United States
             Systems Laboratory
             Environmental Monitoring
             P.O. Box 93478
             Las Vegas, NV 89193-3478
June 1991
Report No. 600/4-91/014
          Research and Development

                      K.H. Lee
        Lockheed Engineering and Sciences Company
             Environmental Programs Office
               Las Vegas, Nevada 89119
                 Contract 68-CO-0050
                  Technical Monitor

                    R.S. Lunetta
             Environmental Protection Agency
          Advanced Monitoring Systems Division
       Environmental Monitoring Systems Laboratory
               Las Vegas, Nevada 89193
           LAS VEGAS, NEVADA 89193-3478

The information in this document has been funded wholly or in pan by the United States Environmental
Protection Agency under contract number 68-CO-0050 to Lockheed Engineering and Sciences Company.
It has been subjected to the Agency's peer and administrative review, and  it has been approved for
publication as an EPA document.  Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
The author wishes to acknowledge the help and technical assistance provided by Mr. David R. Williams
of Lockheed Engineering and Sciences Company. Mr. Williams has many years experience in photo-
interpretation of wetland systems. He contributed to the Aerial Photography Application Section and
supplied several of the personal references cited under the Example Project Costs Section.

The purpose of this  investigation was  to  research and document  the  application of remote sensing
technology for wetlands detection.  Various sensors and platforms are evaluated for:  suitability to
monitor specific wetland systems; effectiveness of detailing wetland  extent and capability to monitor
changes; and relative  cost-benefits of implementing and  updating wetlands databases.

The environment to be monitored consists  of physiographic and ecological wetland resources affected
directly or indirectly  by anthropogenic  activity.  Aircraft and satellite remote sensing can be used .to
record and assess the  condition of these  resources. Monitoring of environmental conditions is based on
the observation and interpretation of certain landscape features. Although some forms of monitoring are
continuous, resource monitoring from aircraft and satellite platforms is periodic in nature, with change
being documented through a series of observations over a given span of time.

This report summarizes the findings  of a bibliographic search on the methods used to inventory and/or
detect  changes in wetland environments.  The bibliography contains  numerous  citations and  is not
intended to be all-inclusive.  Books, major journal and symposium proceedings were examined.  The
findings  documented  will provide the  potential  user  with  a basic understanding of remote sensing
technology as it is applied to wetland monitoring and trend analysis.


Notice/Acknowledgement  	  ii
Abstract	 i i i
Tables	•	   vi

Introduction   	 	 	    1
     Remote Sensing of Wetlands 	   1
     Classification Systems	    3

Aerial Photography Applications	   4
     Photo-Interpretive Process 	 	   4
     Small Scale Photography	    5
     Medium-to-Large Scale Photography 	 	    5
     Aerial Photo Previous Work 	   5
     Change Detection with Aerial Photography 	    6

Satellite or Aircraft Based Sensors   	    8
     Digital Image Processing  	   8
     Change Detection with Digital Imagery 	    9
     Satellite Sensors 	   10
          Landsat MSS Previous Work   	   10
          Landsat TM Previous Work  	   11
          SPOT Previous Work 	   12
     Aircraft Multispectral Scanners  	:	 .   13
          Aircraft MSS Previous Work  	   14
     Radar Applications	   16
     Videography 	   16

Geographic Information Systems  	   18
     Vector CIS Systems	   19
     Raster CIS Systems	   19
     Digital Data Sources  	   20

Example Project Costs	   21
     EPA-EMSL  	 22
     USF&WS-NWI	   23
     USAGE	   23
     State Programs   	   24

References	   26

Bibliography  	  31
Glossary	•	  36

     A - Sources of Digital Spatial Data   	  46
     B - State Sources of  Digital Data   	  59
     C - Field Verification Methods 	  64


Number                                       •                Page

1   Spectral Regions for  Wetland Vegetation Discrimination       4

2   Remote Sensing Sensor Specifications & Acquisition Costs    21

3   Project and Program Summary Table                          25

There are two primary types of remote sensing projects relative to wetlands. The first, resource mapping,
involves acquisition of baseline  data on type, extent, and health of wetland communities. The second
involves detection of change, either natural or anthropogenic, in those communities.

Most of the potential users of this data are responsible for the management and protection of the wetland
resources within their respective states.   To accomplish this,  various  types of data are required for
decision making. In considering use of remote sensing sensors and systems to provide such data, several
factors  must be considered.  What levels of precision and accuracy are  required in identifying and
measuring wetland resources?   How much "leeway" will be tolerated  in the correct identification of
wetland communities?  How "close" can the estimation of area! extent of specific wetland types be when
compared to actual acreage? This factor of precision must be viewed from an economic standpoint, i.e.,
what is the relative cost of each  system versus its degree of precision?

Wetlands are dynamic ecosystems which are defined by federal regulating agencies as possessing three
essential characteristics: (1) hydrophytic vegetation, (2) hydric soils, and (3) wetland hydrology, which
is the driving force creating all wetlands (Federal Interagency Committee for Wetland Delineation, 1989).
These ecosystems are difficult to quantify because of complex interactions between these parameters.
Water level fluctuations and diverse geographic settings make  the  many types of wetlands difficult to
characterize as a group. Changes occur in wetlands and along boundaries in response to variations in the
hydrologic cycle in which seasonal, annual, and long term fluctuations are the driving factors.  Wetlands
are also changed,  altered,  and  encroached upon by human activity, such as urban and agricultural
developments, stream channel alterations, and draining and damming activities.
When wetlands are to be inventoried several issues need to be addressed about the method in which the
baseline or change data should be acquired, categorized, verified, integrated, stored, and distributed.
Because wetlands generally have poor accessibility due to uneven and unstable terrain and frequently tall
vegetation,  any  field work undertaken to inventory them is usually expensive,  time consuming,  and
sometimes inaccurate  in  location.  Reducing the amount of field work via remote sensing  (aerial
photography, aircraft or satellite multispectral scanners) is a viable solution. Baseline inventorying should
consider climatological influences, and generally should have a multi-seasonal approach to capture all the
inherent variances of plant phenology.  If wetland maps are based on persistence and extent of surface
water alone, their boundaries will vary seasonally.   Change detection and trend  analysis requires
ecological information be collected under optimum conditions so valid comparisons can be made between
several  points in time.  The availability of current and historical climatological and image data are
relevant in the wetland inventory planning process.  Roller (1977) lists the advantages to monitoring
wetlands with remotely sensed data over conventional field surveys as: (1) economy, (2) timeliness, (3)
favorable viewing perspective, (4) synoptic observation, and (5) permanent graphic records. It should
be added that with the advent of geographic information systems (CIS), land cover/land use data layers
derived from aircraft or satellite-mounted multispectral sensors or digitally encoded aerial photographs,
can be  incorporated fairly easily  since their formats are digital.   Furthermore, remotely sensed data

collected for wetland change or monitoring activities  may  be utilized  in  future projects  not  under
consideration at the time of acquisition.  With  more sophisticated digital image processing techniques
available, imagery can be enhanced to answer other environmental questions.  Limitations of monitoring
wetlands with remotely sensed data are evident when detailed environmental information is required, such
as a complete floristic make-up of a plant community, or if controversial regulatory decisions are to be
made regarding boundary delineation. Generally there are requirements for field verification (reference
data) in order to attach reliability or confidence estimates to the inventory. Logistic difficulties can arise
because  ideally ground reference data collection should  coincide with the acquisition of the remotely
sensed data.

When planning wetland inventories or change detection studies which will utilize remotely sensed data,
a classification system must be selected or derived which will satisfy the users requirements. Many times
the resulting wetland information derived may not be directly compatible or comparable to  previously
used classification systems or available reference data. Therefore, the end-user must decide the inventory
level of detail (vegetal components,  minimum  mapping unit, classification scheme,  etc.)  and whether
multistage and/or multitemporal sampling  is required.   Relationships that  exist between the desired
mapping parameters and pre-existing classifications may form the basis for the final classification scheme.
If the classification system is hierarchial, it can be amended when information is attained at a higher level
of detail.  For instance, satellite data could be used to  map  an entire state  and/or region, small scale
photographs  for regions and counties, large scale photographs for counties and townships, and hand-held
oblique photos from light aircraft or helicopter, with ground transects or plot sampling for individual
wetland systems (Roller,  1977).

Rowland (1980) found that  differentiation  between vegetation signatures was  best  in the  longer
wavelengths  (red through infrared) for the broad variety of wetlands occurring in northern New England.
This phenomenon is attributed to the green biomass of vegetation which  uniquely  absorbs visible light
and strongly reflects near infrared (IR).  Chlorophyll absorption in the visible  spectrum  and multiple
scattering in the near-IR due  to internal plant structure allows for the characterization of vegetative
communities based on spectral patterns. In general, vegetation canopies have bidirectional reflectance,
but patterns can be further influenced by the understory component, leaf structures, density of vegetation,
etc.  Furthermore,  water absorbs a  majority of the infrared energy which  results in a sharp contrast
between vegetation and water in the IR wavelengths.

Visible wavelengths are more sensitive  to  atmospheric haze and  particle scattering,  although  the
blue-green spectrum can penetrate water  for short distances.  When submergent vegetation is being
targeted, collection of remotely sensed data in the visible wavelengths can provide superior results under
the right atmospheric and hydraulic conditions.

Schloesser et al.,  (1985) found that  low-altitude color aerial photography (1:6,000 scale) with limited
ground survey information, could economically identify beds of submerged macrophytes in the St. Clair-
Detroit River system. Conditions of minimal cloud cover with acquisition time at approximately solar
noon allowed maximum  solar illumination and  water penetration with  minimal  sun  glint.   Five
macrophyte genera and a sand substrate category were delineated on the photos with  a overall accuracy
of 68% by six individual interpreters of varying experience. The authors suggest that improved accuracy
could  result from   a more detailed  dichotomous key,  and greater use of biological  experience in
development of the  key.


 There are a variety  of classification systems used which are generally tailored to satisfy the needs of
 resource managers who  must monitor and regulate wetlands under specific mandates.  The result of
 different systems is maps and inventory products which are not mutually compatible, i.e. different scales.
 media formats, minimum mapping units, and classification schemes. Moreover, all classification systems
 are not designed to  incorporate the results of digital remote sensing, thereby requiring modifications
 and/or ancillary  data to be incorporated into the inventory to make them complete or compatible.

 In 1979, the  U.S. Fish & Wildlife Service (USFWS) published a  nationwide classification system
 (Cdwardin, et al.,  1979) for use in the National Wetlands Inventory (NWI). The Cowardin classification
 strategy utilizes a hierarchial approach where "Systems" form the highest level.  Five systems are defined
 as marine,  estuarine,  riverine,  lacustrine,  and  palustrine.    Marine and  estuarine each  have  two
 subsystems, subtidal and intertidal; the riverine has four subsystems, tidal, lower perennial, upper
 perennial, and intermittent;  the  lacustrine has two, littoral  and limnetic; and  the  palustrine has no
 subsystems (Cowardin, et al., 1979).  Subsystems contain classes which are based on substrate material.
 water regime, or vegetative type, and can be further defined within subclass. Dominance type is a level
 subordinate to subclass and describes the dominant plant or animal species.  Special modifiers can also
 be identified, such as water regime or water chemistry modifiers, and others which describe animal or
human modifications to the wetland.  This FWS classification is designed to allow: (1) ecological units
which contain homogenous attributes to be described and  grouped;  (2) organization of these units to
facilitate resource management decisions; (3) consistent inventory and mapping of wetlands on a national
 level; and (4) uniformity in  concepts and terminology (Cowardin, et al., 1979).  The Cowardin system
 replaced the previous system employed by FWS in their 1955 nationwide wetland inventory which was
designed to emphasize the value  of wetlands for general wildlife and  waterfowl habit, i.e.  Circular 39
(Martin, et al., 1953, Shaw and Fredine, 1956).

The accuracy of the NWI has never been tested for the nation as a whole, but a few assessments have
been done which compare the NWI classification to localized wetland mapping results. Problems arise
when comparisons are made between digitally processed land cover/land  use data and a classification
 system like Cowardin which is designed for use with medium-to-large scale aerial photography.  Only
generalized classes can be compared when utilizing digital classifications of satellite or aircraft based
scanners for comparisons with existing NWI data. An overall identification accuracy of 95 percent was
determined by Swartwout, et al. (1981) for the NWI in Massachusetts.  Hardin (1985), found similar
 accuracies for the  NWI in Delaware.  Luman (1990),  found good correspondence in Illinois when the
 NWI information was regrouped and then compared to satellite-based classifications.  Pickus (1990), also
found high correspondence (90%)  when comparing  NWI digital data, that was re-grouped as wetland
versus non-wetland, to a satellite-based land cover classification in southeast Louisiana.

There is also a classification system  developed by the U.S.  Geological Survey (USGS)  which is a
multi-purpose land use/land  cover system designed for remotely sensed data (Anderson, et al., 1976).
This system  is also  hierarchial  with a framework of nine general Level  1 categories that are further
subdivided into 37 Level 2 categories. Higher detailed categories can be designed into the system to suit
the user at the third  and  fourth levels. Wetland is one of nine Level 1 categories, with a subdivision at
the Level 2 for  forested versus non-forested wetlands.  The flexibility to either generalize or specify
species at higher levels allows inventory compatibility when the remotely  sensed or ancillary inventory
data obtained is not described to the same level of detail.  The higher levels (i.e., 3 and 4) are not usually
 attainable strictly from remotely  sensed data, and normally require ancillary information,

The interpretation of aerial photographs to accurately map specific wetland vegetation and boundary
information is the most common method utilized by Federal, State, and local agencies.  Photo analysis
involves the identification and delineation of specific features recognizable by their distinct "signatures"
(combinations of image characteristics).  These characteristics include: tone, texture, shape, size, shadow
height, and spatial relationship.  Additionally, examination of the aerial photos stereoscopically enables
the interpreter to observe the vertical as well as the horizontal spatial relationships of the subject features.
Due to the complexity of the interpretative process and the wealth of data within aerial photos, accurate
photo interpretation requires considerable expertise.  The accuracy of the interpretations depends  on the
quality of photography, the experience of the photo-analyst with specific wetland settings, and the amount
of ground truth verification to be conducted.

Photographic systems acquire spectral information with films of various .spectral sensitivities.  In order
to maximize  the photo-interpretation result, it  is important  to select  a film type which will provide
maximum contrast between different plant communities. Choices available for camera systems are color,
color infrared, and panchromatic (i.e. black/white) films.  Table 1 lists the spectral region and  season
recommended by Roller (1977), in which to collect aerial photography that would take advantage of
phenological differences among common wetland vegetation. .
   Vegetation type             Season of maximum contrast          Spectral region
   Aquatic submergent
      shallow depths                     Summer                      Near infrared
      deeper depths                      Fall                          Visible

   Floating                             Summer                      Near infrared

   Marsh emergent  and meadcw          Fa!I                          Visible

   Shrubs                           Summer or Fall                    Visible or Near IR

   Trees                                 Fall                          Visible

The most notable resource mapping effort is that of the National Wetland Inventory (NWI) conducted by
the USFWS. Begun in 1977, it creates in map form a database of the Nation's wetlands and deepwater
habitats.  Updates of specific areas are also conducted to  document wetland gains  or losses.   NVv'l
products have included detailed maps, wetland soils and plant lists, and reports on specific regions or
states. Color infrared photography (1:58,000 scaJe) was acquired nationwide between  1981-1984 by the
U.S.G.S. under the National High Altitude Aerial Photography Program (NHAP).  Photo interpretations
derived from NHAP photography for the NWI were categorized by dominant vegetation and hydrologic
characteristics according to the USFWS classification system. The NWI identifies and maps wetlands
to the smallest acreage visible on the photographs being used (Carter, 1982).  The resulting NWI maps
were produced using USGS 1:24,000 scale topographic maps as a base.  These topographic quadrangles
(when available) are used as base maps because they provide a geometrically accurate starting point on
which to compile more recent photo-interpreted information. Standard USGS topographic maps depict
vegetative categories by unbounded symbols which makes them useful only for gross generalizations.
The smallest area that can be depicted on a 1:24;000 scale map is about 0.4 hectare (0.9 acre).  Starting
in 1988, the USGS has initiated another nationwide acquisition of high flight CIR photography that is
scheduled to be completed in 1992.  The National Aerial Photography Program (NAPP) will  collect
1:40,000 scale CIR, leaf-on, quarter quad-centered photography on a state-by-state basis.
Small scale CIR photography other than NHAP and NAPP have been utilized to map wetlands.   Carter
et al. (1979) found  that a 1:130,000 scale  photograph had a recognizable or interpretable limit of
approximately 0.5 hectare (1.2 acre).  This fact  underscores the need to define the level of detail the
wetland  inventory will require  based  on  its ultimate use.  Wetland mapping projects are  usually
individually designed programs geared to satisfy either legal requirements for regulation, the needs of the
mapping agency, or organizations conducting environmental assessments. For most regulatory permitting
procedures and site specific planning activities, accurate boundary delineation is critical. The use of small
scale  photography (i.e. 130,000  scale or smaller) is primarily limited to the identification of plant
communities with distinct spectral characteristics.
Specific locations and vegetation types are easily identifiable on large scale photography, even by photo
users with minimal experience. Very large scale photographs are also useful in investigations of wetlands
that are small, isolated, or narrow and linear.  These characteristics, especially when combined, make
these type wetlands virtually impossible to detect with small scale photography.
In a study reported by Howland (1980), color infrared (CIR) film was found to be superior to color and
multi-band black and white photography for vegetation discrimination in a diverse inland wetland in
Vermont. Besides film choice, attention must also be given to the time of year and even the time of day
when acquiring imagery.  Grace (1985) found that for inland non-persistent emergents in the southeastern
U.S., mid-August to mid-September was the best time to collect photography for annual comparisons

 since most species would be represented and in a mature state. A study by Carter, et al. (1979), found
 that the time period between August to mid-October was optimum for identifying open water and marsh
 categories in Tennessee.  In another study by Gammon and Carter (1979) of the Great Dismal Swamp.
 it was found that photographs obtained during vegetation dormancy  allowed for better identification of
 wetland boundaries, areas covered by  water, drainage  patterns,  separation of coniferous forest from
 deciduous, and classification of some understory components. In tidal environments, Brown (1978) found
 August to early  October best for saline wetlands, late June to early September best for fresh/brackish
 wetlands,  and that fresh water wetlands needed coverage twice, mid-to-late June and late August due to
 specific plant species being predominant at different times.

 Wetland resource mapping projects have been performed for advanced wetlands identification projects
 at  EPA's Environmental  Monitoring  Systems Laboratory  (EMSL) in Las  Vegas,  and  the Photo
 Interpretation Center (EPIC) located in Vim Hill Farms, Virginia (Norton, 1986a, 1986b; Duggan, 1983;
 D.R.  Williams, 1983; Mack, 1980). Hydrogeologic and ecologic evaluation methods are applied in field
 work to assist in advanced wetlands mapping efforts.  Utilizing this type of information allows managers
 to  take a "proactive" stance relative to jurisdictional determinations and suitability of land parcels for
 development, rather than obtaining data on a case by case basis. Other detailed  resource mapping is
 conducted for EPA's Regional Offices  for wetland areas of special  or critical concern (Norton, et al.,

 The following list of identifiable features is provided to acquaint the reader with data available from such
 projects.  A nominal scale of 1:12,000 -1:24,000 and use of color/CIR photos is assumed.

 o Wetland boundary delineation
. o Area
 o Edge, drainage length/densities
 o Shape of upland/wetland edge
 o Fetch/exposure
 o Vegetation growth form
 o Cover density  and distribution
 o Species composition and health
 o Tidal flooding regime
 o Tidal conduits, inlets, outlets
 o Erosion
 Change detection projects which utilize 'aeriai  photography involve the use of one to a number of
 historical sets of photographs to document changes either natural or anthropogenic (D.R. Williams, 1989,
 1985; Grace, 1985; Niedzwiedz and Batie,  1984; Hardisky and Klemas, 1983; and others). Listed below
 are features that are useful  in change detection  analyses  which are  interpretable from color/CIR
 photography with scales ranging from 1:10,000 to 1:24,000.

 o Pre-developmem vegetation patterns
 o Natural vegetation removal

 o Surface drainage network
 o Sediment
 o Dredging, turbid water
 o Fill or spoil deposition
 o Erosion
 o Ground staining, scarring
 o Vegetation stress or damage
 o Contaminant sources
Although there is no "average" change detection project, a description follows of some requirements and
products which are typical.  Federal  agencies charged with enforcement of Section 404 of the Clean
Water Act, which authorizes permits for the discharge of dredge or fill materials into the waters of the
United States,  extensively utilize aerial  photointerpretive data  to document alleged illegal development
of wetland properties.  Many times chronological documentation must be performed so that the illegal
filling of wetlands can be  substantiated.  Regulatory enforcement related projects typically  involve
numerous historical photographs with detailed comparative analysis, interpreted overlays, field verification
of interpreted  results, and very large scale graphic displays.  Therefore,  costs  for in-depth wetland
delineation are substantially higher than for conventional wetland resource  mapping efforts.  Costs to
produce this type of detail, which may be required in court cases, have averaged approximately 51,000
per square  mile ( D.R. Williams,  LESC, Personal Communication,  1990).   In other instances,
investigations are aimed at quantifying the loss of wetland resources over time by climatological  affects.
D.C.  Williams and Lyon (undated) correlated changing Great Lakes water levels to wetland extent, and
Grace (1985) investigated the effect of thermal discharges on seasonal wetland production.

A state agency in Georgia acquires aerial photography at predetermined intervals to monitor development
along the Atlantic coastline (S. Stevins, Georgia Dept. of Natural Resources, Personal Communication,
1990). The U.S. Army Corps of Engineers performs similar monitoring in most of their districts, as they
are responsible for making jurisdictional determinations of wetlands regulated under  Section 404 of the
Clean Water Act. The progressive destruction of wetlands in  association with hazardous waste sites is
also documented, and this information is used to plan clean up activities at CERCLA (Comprehensive
Environmental  Response Compensation, and Liability Act of 1980) sites  (Norton and Prince,  1985).

At an agricultural conversion site in Florida seven sets of available historical photos, black-and-white and
color photographs with scales ranging from  1:10,000 to  1:24,000, were  acquired  and interpreted.  The
purpose  of  the investigation was to assess  and document the  type and amount of wetland habitat
destruction (D.R. Williams,  1981).  Natural vegetation present before conversion, vegetation removal,
and surface drainage network construction were documented over time.   Ground information was
collected prior to the photo analysis,  and  comparisons  conducted to  confirm  map  accuracy after
completion.  Frequently, photography  is flown at periodic intervals to ascertain if a developer is in
compliance with Federal and/or State issued permits. If such investigations become court cases, large
courtroom displays of interpreted data are prepared with expert witness testimony usually given by the
principal investigator or experienced photo-interpreter.

 Digital image  processing of satellite or aircraft-acquired data into land cover categories  involves the
 examination of the reflectance or spectral patterns of pixels contained within the image. Image data is
 organized into a matrix, i.e. raster format, with each pixel (or cell) covering a certain dimension on the
 ground. Each pixel contains one data value representing the spectral intensity of that location on the
 earth's surface for a particular wavelength, i.e., 7 wavelength bands, 7  intensity values.  Fundamental
 energy-matter  interactions control and influence the spectral characteristics of land cover types.  The
 proportion of energy reflected, absorbed, and transmitted for each cover type varies depending on the
 material and ground conditions (Lillesand and Kiefer,  1987).  Most of the earth's features have unique
 spectral characteristics, and thereby can be identified and mapped on the basis of their spectral signatures.

 Pixels commonly correspond to more than one type of land cover and therefore the pixel value represents
 a weighted average. The result is that each raster pixel is a combined product of all the resources present
 at that ground location. Presence of "mixed pixels" causes problems when boundary or edge information
 is required.  Boundary effects can take the form of mixed pixels occurring as transition areas  between
 adjacent cover types.  The spectral characteristics of those mixed pixels are unlike the land-cover types
 on either side, and depending on the classification technique utilized to process the digital data, may be
 incorrectly grouped as a separate class.

 Other possible sources of error or inconsistencies when dealing with satellite or aircraft data originate
 from atmospheric scattering, sun  angle,  topographic  influences,  and the  process of geographical
 rectification of the imagery.  The scattering, sun angle, and topography can be modelled or accounted
 for by assumptions and/or ancillary data, while the rectification process has a reportable positional error
 once the geometric transformation is completed.

 In digital image processing, data can be analyzed in several ways, most commonly  with  either image
 enhancement or image classification algorithms. In order to summarize the remotely sensed data  into land
 cover/land use, those processes which are addressed as classification techniques are employed.  The
 general approach  in  the  classification of digital  imagery fall  into two categories;  supervised  or
 unsupervised  approaches.  In  a supervised classification, known land cover types such  as forest,
 agriculture, wetland, and urban have been identified either through field work, aerial photography, maps,
 or personal experience (Heaslip, 1975).  These sites  are then located in the  image  and homogeneous
 examples are  delineated.   These delineated  areas  serve as training sites because their  spectral
 characteristics  are related to known resources, and can be used to "train"  the classification algorithm
 (Jensen, 1986).   Statistisa!  information  :s  generated for each identified  training cover type  and
subsequently that information is used to classify all unknown pixels remaining in the image.  Conversely
 in unsupervised classifications, the location of particular land cover types  is not known, and  the statistical
parameters required to define the training  classes are then determined with clustering algorithms. It is
then the responsibility of the analyst to label these clusters with the appropriate land cover category.

Evaluation of the land-cover classification accuracy requires comparison between the deification results
and reference data for the area.  This reference data can be taken from various sources such as existing


specialized maps, photo-interpreted  aerial photographs, and  actual "field checks" (see Appendix  Ci.
However, the reference data which is utilized during classification should not be used in the accuracy
assessment as -it heavily biases the results.  Congalton (1988) compared several sampling schemes for
assessing the accuracy of classified remotely sensed data, and  found that generally a one percent sample
should be  obtained.   To  assess the agreement between  the classification and the  rererence data.
site-specific comparisons are made by calculating the frequency of coincident classes, point by point, on
the reference data and the classification results.  These values are reported in an error matrix (sometimes
called a confusion matrix or contingency table).  The total overall percent correct is the ratio of the sum
of diagonal values to total number of cell counts in the matrix.  Proportions of diagonal  values to  row
sums are the category accuracy relative  to errors of commission, and proportions of diagonal values to
column sums as category accuracy relative to errors of omission (Story and Congalton, 1986).  Detailed
statements of accuracy are derived from the error matrix in the form of individual land-cover category
accuracies. For each class, percent commission and percent omission are calculated from the error matrix
as well as the overall accuracy.
Change detection projects which utilize digital imagery are generally more limited in the amount and type
of historical data available. Digital change detection tasks are somewhat complex and require consistent
multiple date imagery acquisition to insure reliable mapping of baseline conditions and change trends.
Several different image processing methods can be employed for change detection studies.  These methods
(image differencing, image ratioing, classification comparison, comparison of pre-processed imagery, and
vector change analysis) are specific to the form in which the imagery is obtained, and to what precision
the two images can be co-registered (Jensen, 1986). It should be noted that depending  on whether the
imagery used  is "raw" or unclassified, or the result of classification or transformation  algorithms, the
different methods of digital change analysis will produce slightly different information.  For example.
image differencing results in an  image (or map)  that reveals  only the locations that have undergone
change, and is  directly dependent upon the spatial registration between the  images.   Conversely.
comparisons between two classified images depict both the changed area locations and the nature of the
change  (Howarth and Wickware,  1981). The accuracy of this type of derived change detection map, and
the statistics generated from it, are contingent on both the spatial  registration between the maps and the
individual classification accuracies of the maps (Hodgson et al., 1988).  Furthermore, the composite map
produced from digital change analysis will not have an accuracy  greater than the least accurate map in
the analysis (Newcomer and Szajgin, 1984).

Digital  change detection offers an alternative to manual photo-interpretation, but can be more difficult
to perform accurately due to the spatial, spectral, and temporal constraints placed on multi-date imagery
analysis. The results from photo interpretation of large-scale photographs may have higher accuracies,
but are time consuming, difficult to replicate, prone to errors of omission, and costly in terms of data
acquisition. Digital change detection studies must utilize analysts familiar with the environment to be
inventoried, the quality/accuracy of the data set(s), and the characteristics of change detection algorithms
in order to be successful (Jensen,  1986).

There are two primary sources for obtaining satellite acquired digital imagery, Earth Observation Satellite
Company (EOSAT) and Systeme Pour 1'Observation de la Terre (SPOT)  Corporation.  EOSAT is a
participant in a cooperative agreement between the National Oceanic and Atmospheric Administration
(NOAA) and the USGS to facilitate the commercialization of satellite based technology to the private
sector.  At present there are  two satellites being operated by EOSAT, Landsat 4 and 5.  The Landsat
systems carry two types of sensors, the multispectral scanner (MSS), and  the thematic mapper (TM).
Both the Landsat 4 and 5 satellites have exceeded their design lives by three years with Landsat 5 being
the primary system utilized in North America.  The MSS provides data in four spectral bands:  visible
green, visible red, and two reflected infrared bands at about 80-meter ground resolution.  The TM has
six spectral bands with 30-meter resolution and a thermal infrared band with 120-meter resolution.  The
Landsat TM sensor possesses higher spectral resolution because of the seven individual bands which
capture reflectance data in the visible through the thermal portions of the electromagnetic spectrum.  The
TM system is the newer of the two systems aboard Landsat 4 and  5, and still has resource mapping
capabilities that have not been  fully explored.   These satellites are in polar orbits and are capable of
collecting spectral information every 16 days over the same surface area.

France launched the first SPOT satellite in February 1986. This satellite has sensors on board which are
capable of producing 10-meter resolution black-and-white panchromatic images, and 20-meter resolution
three  band multispectral images.  SPOT sensors can be directed off-nadir to produce  stereoscopic
coverage and have a repeat collection  cycle of 26 days. The spatial resolution of the SPOT satellite is
finer than Landsat TM, but the lessor spectral  resolution and smaller area per scene of SPOT imagery
make  comparisons  difficult  between the two sensors.   The two systems  could be considered
complimentary in that the SPOT imagery has been merged with Landsat imagery, with resulting hybrid
imagery that reflects the spatial resolution of SPOT with the spectral resolution of Landsat (Salvaggio and
Szemkow, 1989; Welch and Ehlers, 1987).
Landsat MSS Previous Work

Landsat MSS has been investigated for its utility in wetland mapping and inventory updating of broad
wetland communities (Jensen, et al., 1980; Hardin,  1985; May,  1986; and others).  Some of the resource
mapping problems encountered with the use of Landsat MSS data for wetlands originate from the low
spatial resolution (80-meter) of the MSS  sensor which hampers detection of wetlands smaller than  1.6
hectares  (4 acres), the inability to place  boundaries with high  reliability, and uncertain availability of
multi-date imagery (16-day repeat cycle and cloud free conditions required) for use in discrimination of
different vegetation types based on phenology.

A study by Jensen, et al. (1980) compared Landsat MSS. color aerial photography, and radar images for
the identification of giant kelp beds in southern California.  It was found that kelp was separable from
ocean in Landsat MSS, but that the acreage estimates generated  from Landsat MSS were consistently
underestimated in comparison to those derived from the color photography.  Such underestimations were
attributed to the photo interpreter's ability to identify less-dense areas of kelp on the photographs.

A study  undertaken by Ernst-Dottavio, et al. (1981) was aimed  at determining whether Landsat MSS


 could identify inland wetlands  in northern Indiana.   In  that study, spectral characteristics of inland
 wetlands were quantified utilizing a helicopterTinounted radiometer sensitive to the four Landsat MSS
 channels.   The study found hardwood swamps, shallow  marshes, and shrub swamps to be spectrally
 similar, therefore the classification resulted in low individual class accuracies. However, the deep marsh
 environment  showed higher accuracies, and an overall  accuracy of 71  percent was reported for the

 In a study by Wood  (1983)  black-and-white photography, aerial 35mm color slides,  CIR highflight
photography, Landsat MSS, and Landsat TM  simulator  were used  to document historical changes  in
 wetlands in California.  The Landsat MSS classification results showed considerable confusion when
compared to land use maps complied from the CIR aerial photography.  As a  result, the wetland areas
 were greatly overestimated. Wood suggests that better classification accuracies could have been obtained
 with perhaps  a different image date, or a multi-temporal  approach.

 Hardin (1985) reported an overall accuracy of 72 percent for distinguishing wetland from non-wetland
 in Delaware utilizing multi-date Landsat MSS images acquired in 1974.  Classification accuracies ranged
between 74 percent to 28 percent for individual marine species. Freshwater wetlands were less accurately
 classified, and most misclassifications occurred  between the various wetland types.  Hardin believes state
statutory requirements make it unlikely for Delaware to ever discard the use of low altitude photography
 in favor of satellite data, even if accuracy were greatly improved.
In another study using Landsat MSS data, May (1986) evaluated  the feasibility of using automated
classification techniques to update habitat maps  based  on the Co ward in system.  In  comparing the
resulting Landsat  MSS based habitat classifications with existing habitat maps prepared  from aerial
photographs, the study  found a high degree of omission and commission errors, and low mapping
accuracies for all habitat categories.  It was noted that the  fundamental problem encountered in the study
was the inability to accurately co-register the Landsat MSS to the photo-derived habitat maps.
Landsat TM Previous Work
Landsat TM, has improved wetland monitoring capabilities over the Landsat MSS system. In comparison
to the Landsat MSS, the TM sensor provides seven narrower spectral bands, better spatial resolution,
improved radiometric sensitivity, and a higher number of quantization levels (i.e. digital numbers for TM
= 256, for MSS = 128).  TM is still subject to a repeat collection cycle of 16 days, and use of the data
is contingent upon near cloud-free conditions.

In a study by Wood (1983) Landsat TM Simulator data was acquired over California's Central Valley
to supplement existing aerial photography and Landsat MSS.  The objective was to develop a series of
historical maps to document the extent of wetland change  between.1937 and 1982. Very little confusion
was found between wetland and other land use classes based on unsupervised classification results. The
comparison of four resulting classes (permanent  wetland, seasonal wetland, native vegetation, and water)
against the manual photo interpretation of CIR transparencies, resulted in an overall agreement exceeding

In a study by Jacobson, et al. (1987) a waterfowl habitat inventory map was  generated for an area
northeast of Anchorage using Landsat TM acquired in August 1987. The result was a  thematic map
depicting six waterfowl habitat classes (deep freshwater,  shallow freshwater, turbid water, aquatic bed,


 deep marsh, and  shallow marsh).   Direct comparisons between the  TM classified waterfowl habitat
 information and the existing NWI for the area was not possible due to  different classification strategies.
 The 51 unique wetland types identified by the NWI for this area were reduced to 24, and contained only
 those categories which were related to waterfowl habitat.  Frequency  analysis was performed between
 the two data sets.  The TM results when compared to the  reduced NWI data set as coincident, omitted.
 or committed categories, achieved an accuracy of over 90% for lacustrine systems and about 25%  for
 seasonal and temporarily flooded wetlands.  When point and linear wetlands were removed from the NWI
 data, the TM classification missed 78% of the wetlands under two acres.  The NWI mapped the area as
 having 83%  of the wetland basins under 2 acres in  size.  The authors conclude that Landsat TM  can
 provide reliable wetland  information for basins greater than two acres, and because of the computerized
 nature of the resulting data bases, updates can be incorporated when more detailed information becomes
 available.   Generally,   when  current TM imagery and NWI  information  are  combined,  wetland
 preservation and planning issues can be addressed more readily.

 A study to identify suitable wetland habitats for wood  stork with Landsat TM data was undertaken by
 Hodgson, et al. (1988) in east-central Georgia.  In the analysis, a computer classified foraging habitat
 map was created using TM imagery collected for two years, with  each year having imagery for both
 "wet" and  "dry" seasons.  Seven categories were generated: deep water, shallow water, macrophytes,
 cypress/mixed  wetland,  bottomland/hardwoods,  pine/mixed  uplands,  and  agriculture/clearings.
 Classification accuracy was evaluated by comparing the TM results to previously verified foraging sites.
 Overall accuracy results  of 74% and 88% for the two years investigated were reported.

 Pickus (1990) found Landsat TM to be an effective tool for delineating wetlands under the Environmental
 Protection  Agency's  (EPA)  Section  404 Wetlands  Advanced  Identification program  in southeastern
 Louisiana.   Based on comparisons between field collected reference data and NWI data, correlation
 between the classified TM imagery and field verification resulted in an 85% accuracy in distinguishing
 wetlands. Accuracies of 86% and 71 % were obtained in the identification of bottomland hardwoods  and
 uplands respectively.  Results were poor for cypress/tupelo (48%) and water (56%) categories.  When
 compared to digital NWI data, the  classified TM wetlands identification had a very high correlation
 (90%).  Ancillary information was utilized in this project to create a geographic information system (CIS)
 which  contained in addition to the satellite derived land cover, hydrography,  soils, transportation  and
 elevation layers.   Pickus concluded that for this  project,  digital  analysis  of Landsat TM,  when
 incorporated into a GIS containing ancillary data, was sufficient for advanced wetlands identification.

 In the  State of Illinois,  preliminary .evaluations have begun to  assess  the  potential of satellite data for
updating of the NWI  (Luman,  1990).  TM data from May 1987 were classified into generic Anderson
Level  1 land-cover categories (agriculture, grass, shrub, forest, wetland, and water) for a study  site
located in southern Illinois.  Sample point comparisons were made against a variety of ancillary data,  and
the overall  map accuracy was found to be 85%.  When individual categories were evaluated, accuracies
 ranged from 52% for  shrub, 76%  for wetland, and 97% for non-turbid water. Lurnan concludes that  TM
data  would be useful  for detection of previously unmapped wetland habitats.
SPOT Previous Work
Very  few  references  were  located  in  which SPOT imagery was  evaluated  for  wetland mapping
applications. SPOT Corporation has published brochures describing applications in which SPOT imagery

                                               12                     .

 was used for wetland monitoring and vegetation inventorying projects. One project involved the detailed
 mapping of a national  wildlife refuge into 18 classes based on density, species composition, and other
 factors.  In this instance, SPOT panchromatic and multispectral imagery were merged for analysis with
 a resulting minimum mapping unit of 0.01 hectare. No accuracy assessment was cited for this example.
 SPOT also lists projects where submerged aquatic vegetation was located and mapped successfully, and
 another where a wetland area dominated by a particular species was mapped and correlated to vegetation
 index factors for assessing biomass distribution.  Again, no accuracy statements were made about either
 of these applications.

 In a study by Mackey (1990), eleven dates of SPOT multispectral imagery collected over three years were
 evaluated to  determine seasonal  and annual  changes in a 400-hectare, southeastern freshwater  marsh.
 Unsupervised classification techniques were used to generate land cover maps for each date.  Satellite TM
 and aircraft  multispectral imagery  have been analyzed for this site in previous years with reported
 accuracies ranging from 70% to 85% for distinguishing open water, freshwater marsh, shrub-scrub, and
 cypress/tupelo cover types (Jensen, et al.,  1987,  1986,  1984).  No accuracy assessment  was performed,
 but the author contends that accuracies from this project would be  consistent with previous work
 performed by Jensen.  The resulting data were analyzed for trend analysis  with a prediction for a drier,
 more persistent wetland community developing.

 Luman (1990) analyzed  two SPOT scenes (each a hybridization containing both panchromatic and
 multispectral information) for utility in updating the NWI in Illinois. The existing NWI  information was
 regrouped to match a  state-based classification system for direct comparison against the  results from
 digital classification. Good correspondence was found between the satellite-based classifications  and the
 NWI maps,  especially where the NWI had been simplified to the state-based classification, and the NWI
 source information had been of good-to-high quality.   Luman concludes that the use of the two SPOT
 sensors merged (MSS  and panchromatic  data),  is acceptable for  detailed analysis of wetland  habitat
 structure,  and for validating water regime modifiers  within  the Co ward in classification.   The higher
 spatial resolution  of SPOT appears to provide a tool in which to monitor wetland change effectively.
Aircraft multispectral scanners (aircraft MSS) are multi-band sensors which can have from four to 230
separate  channels or  bands  in  which the  sensor  collects information from  various parts of the
electromagnetic spectrum. These systems are designed to measure and record the radiant energy reflected
and emitted from the ground,  and have a wider range of spectral sensitivity than photographic systems.
Their exact configurations vary by the spectral ranges which are sensed and how narrow the bandwidth
the sensors are capable of recording. Very specific data acquisition parameters are required for planning
aircraft MSS  missions.  The area to be covered must  be specified  in terms of location,  size, and
boundaries, with the position of the flight  lines clearly  defined  (usually lat/long coordinates).   If
mosaicking of the data is-required, all flight  lines should be flown in the same direction to minimize
illumination differences.    Also specified are the  appropriate spectral  channels,  acquisition time
(time/day/month/year), and flying height  which controls the resulting resolution and amount of digital
data obtained. Resolutions as  fine as  1 meter are possible from aircraft based sensors. Supporting field
data collection requirements should be clearly defined prior to the mission (see Appendix C). Following
data acquisition, post-processing is necessary to calibrate the data for systematic scanner distortions. The
resulting digital data is in  raster format. There are a  limited number of these operational systems


 available for public use (EPA-EMSL. NASA-AMES, GEOSCAN).  Landsat TM data is more practical
 than airborne MSS because it is routinely collected and less expensive. Conversely, aircraft MSS systems
 can offer flexibility in terms  of resolution,  timing,  and finer  spectral  sensitivity.  Most of  the
 aircraft-mounted multispectral systems are also equipped with mapping cameras  to allow simultaneous
 collection of aerial photography.

 The EPA Environmental Monitoring Systems Laboratory - Las Vegas (EMSL-LV) operates an aircraft-
 based  MSS in which a highly precise navigation system has been incorporated to provide very accurate
 spatial registration.  In the past few years,  the Air Force's Global  Positioning System  (GPS)  has
 revolutionized  positioning technology.  The use of GPS adds a dimension of geometric control and
 correction that has not been available in the past.  The major problem with aircraft scanners is poor
 geometry with data being adversely affected by variations in aircraft attitude or deviations from the flight
 line.  Relatively inexpensive GPS receivers and computer software are available which allow positional
 accuracies ranging from 5 to 25 meters to  be obtained.  This combined with recent developments in
 digital elevation models (DEMs),  offers multispectral data that is geometrically much superior to older
 previous data.

 Applications for aircraft MSS systems include:

    o  Vegetation classification
    o  Land cover / Land use mapping
    o  Water Quality monitoring
    o  Wetland mapping
    o  Monitoring of heated water discharges
    o  Underground fire monitoring
    o  Map updating
    o  Geothermal monitoring
EMSL-LV's system  and  services are available  to  interested  Federal  and State agencies through
interagency agreements (see Appendix A-EMSL).
Aircraft MSS Previous Work
Savastano et al. (1984) reported the results of a study utilizing aircraft MSS to classify and map diverse
nearshore marine and esruarine habitats in St.  Joseph Bay, Florida.  The  objective was to develop
methods for separating seagrass and macroalgal habitats from the ocean bottom using a four band (blue
through near IR) multispectral scanner.  The near IR  provided the capability to discriminate between
emergent vegetation and water, but contained no useful information on submerged habitats as  near-IP.
energy is almost  completely absorbed by water.  The information content of the blue band, which can
penetrate shallow water environments, was lacking due to a low instrument signal-to-noise ratio. When
digitally classified data were compared to ground truth information, it was found that the classification
subdivided the desired categories too finely, breaking out classes that could not be identified on the basis
of existing information.  The study concluded that although it was not possible to collect ground truth
information as detailed as the aircraft  MSS, the resulting habitat  maps were  found to be  generally

 in an tPA Environmental Monitoring Systems Laboratory (EMSL) study by Page (1982), an 11-channel
 aircraft MSS was used to map nearshore kelp beds off the southern California coast.  The study site \vas
 selectively chosen to be an ideal example, as the kelp beds were large and well developed. Unsupervised
 classification was performed on the aircraft MSS imagery using four of the channels (green, red, reflected
 IR, and  thermal  IR),  and results  were compared to  estimates derived from manual interpretation of
 concurrently collected CIR aerial photography. It was  found that the aircraft MSS-derived kelp estimates
 were  considerably  lower than  the photo-derived  estimates.   This  was attributed to  the  inherent
 generalization of mapping boundaries in the  photo interpretation process, where the kelp beds were
 identified as large, visually distinct units. It was concluded that the photo interpretation process resulted
 in the inclusion of non-kelp features (i.e. ocean) within the kelp mapping units.  Page summarized that
 aircraft MSS  data could accurately survey',  map,  and compile acreage estimates for nearshore kelp
 resources. It appeared that the digital approach yielded results which had excellent positional agreement
 with maps produced from photo interpretation, while providing better estimates of kelp bed area! extent.

 Extensive work has been done utilizing aircraft MSS data  for inland wetlands mapping in South Carolina
 (Jensen, et al., 1987, 1986, 1984; Grace, 1985).  In one investigation, 11-channel aircraft MSS data were
 collected  in March  1981, with 3 meter resolution over the  Savannah River floodplain (Jensen,  et al.,
 1986).  The data were analyzed to determine the optimum combination of channels for separability of
 wetland  types,  with green,  red,   and two  near  IR  bands being  ultimately  selected.   Supervised
 classification  techniques were employed  to map  the data into the  Cowardin classification system.
 Categories identified were: persistent emergent marsh, nonpersistent emergent marsh, shrub/scrub, algal
 mat, mixed deciduous upland forest, and mixed  deciduous swamp forest.  An overall accuracy of 83%
 was reported when comparisons were made on a pixel-to-pixel basis to ground transects. Another aspect
 conducted under this  research  was an effort to  relate the types of vegetation present with water
 temperature.   Daytime thermal-IR data were collected during the same mission, and compared against
the classified maps  constructed from the four channel subset.  A statistical analysis was performed  in
which  it  was determined  that  associations did exist between  vegetation types and  their  apparent
temperature class intervals.   However, Jensen concedes that other factors  besides temperature, such as
sedimentation and oxygen stress, may affect the establishment of different plant communities.

 In a recent study undertaken at EPA-EMSL by Mynar (1990), an 11-channel aircraft-MSS was utilized
 to map nearshore habitats in Puget Sound, Washington.  The project objective was to develop aircraft
 MSS data processing protocols with preliminary work focusing on seven test sites. A regional Cowardin
 classification was developed for the Puget Sound marine estuarine system by the Washington State Natural
 Heritage Program.  MSS data collection corresponded with low tide conditions in July 1988, with CIR
 aerial photography collected simultaneously at a scale  of approximately 1:13,000. The resolution of the
 MSS data was approximately five meters.  The final MSS-derived classifications for two sites were
analytically compared  with  classifications derived from manual  interpretation of the simultaneous CIR
 aerial photography.  The overall comparable at the two sites were quite low, 55 and 53 percent.  Mynar
attributes the low accuracies to poor field verification data, and lack of accurate co-registration between
the two databases, i.e. MSS  classification and photo-interpreted CIR photography.  It should be noted that
lack  of agreement does hot necessarily indicate the  MSS classifications did  not correctly depict the
habitats. A larger "minimum mapping unit" results from photo interpretation methods in which resources
 are interpreted and outlined on photographs.  Low accuracy statistics have previously resulted when
manually interpreted results are  compared to high resolution MSS classified data where each  individual
pixel is evaluated and categorized (Page,  1982). The rectification process consisted of three separate,
 yet related  components:  MSS data  registration to an  earth coordinate  system (UTM);  aerial
photo-interpreted results registered to UTM coordinates;  and the co-registration of the photo-interpreted


 results and the MSS data.  Due to the nature of the estuarine environment, reliable ground control points
 were difficult to locate (i.e. shifting sand,.mud. and water)..  Mynar concludes that MSS acquisition must
 be timed accordingly to capture  the resource(s) under investigation, that ground verification data must
 be compatible with remotely sensed data, and image rectification needs to be more precise in order to
 access whether MSS imagery  is appropriate for nearshore habitat inventories.   A  future  mission is
 planned  for 1991 to acquire aircraft MSS over Puget Sound utilizing the enhanced capabilities of GPS
 to increase spatial accuracy.
Radar can be useful  for identifying broad wetland classes over large areas, particularly if the area is
perpetually cloudy.  The  longer wavelength radar systems have the ability to penetrate clouds and
herbaceous vegetation canopies.  Radar is known to be sensitive to differences in the dielectric constant
of surface materials,  and produces intense backscarter or return signals from some wet forested areas.
Disadvantages of using radar are the high costs, limited data availability, and the complexity of the

In a study conducted by Place (1985), Seasat radar images were evaluated for their ability to improve
wetland mapping when combined with conventional sources, i.e. aerial photography. The Seasat satellite
was launched in June 1978 and failed later that year in October.  It was an L-band (23.5 cm) synthetic
aperture radar (SAR), having H-H polarization,  and 25-meter resolution.   If vegetation or moisture
conditions are of interest, Seasat is useful only for areas which are.generally flat.  Place compared Seasat
interpreted imagery and NWI data at four test sites on the coastal plain between Maryland and Florida.
Place concluded, based on statistical comparisons, that photo-interpreters who used Seasat radar images
to compliment their conventional sources were able to map forested wetland more accurately (greater than
85%), than those who did  not.
Airborne videography is not  a replacement  for aerial  photography,  but  rather  a substitute or
complimentary data source when the quality and/or cost of an aerial  survey  can not be supported.
Scientists at the USDA Agricultural Research Service Unit at Welasco, Texas, have been doing basic
research on applications of both normal color and infrared video for r angel and and other vegetation, and
the US Fish  and Wildlife Service in North Dakota have been using video to assist in assessments of
wetland and riparian ecosystems (Driscoll, 1990).

Advantages of video systems are: immediate availability cf imagery; in-flight error-proofing capabilities:
ability to use narrow band filters for finer spectral resolution;  ability to function  in a wider range of
atmospheric conditions than conventional photography; and ability to utilize satellite-type data processing
software for  analysis.  Disadvantages include a lower resolving power than aerial photography and
difficulties with calibration.
Several video camera systems  have been developed  ranging from biack-and-white with
capabilities, u> muitispectrai raise color systems.  A  comprehensive description of the various video

systems available can be found in Everitt and Escobar (1989)   Imagery fmm ,K
used successfully to detect and assess ecological conditions such as nlfn             ^^    *
pests, soil moisture, soil drainage and salinity, vJ^y^^l^^1^ ^ SpeCieS' insec<
Escobar, 1989).   High resolution airborne vide? data  have also be  n% ifz^r    T* (E™'m and
               .                                 aa   ave aso be n   izr
 databases (Ehler et  al.,  ,989).   Video image data lends  itse  to  ^e S Son n?    Can°?raphic
 processing techniques, which can then be linked to geosraohic infnrm fUt"IZatlon of automated image
 update information layers.                        geograpn.c mformation systems (CIS) and used to

 Wu (1989) utilized a commercially available VHS-format  cammMpr ,
 color photography for a multisensor mapping effon S^^^^1^1^.1
 were used as ground-truth in conjunction wfth SPOT  aZrne MSS  '2 %t*™' ^ VideO i
 video images captured the dynamics, volume, and spatial distdbu^fte ££JSL ^
 Wu concludes however, that further study of the  spectral  resnon.pVh!  !       g    °n Very wel1'
surface features is needed to determine which vid^ sys  ms £He  ^T^T °f Vegetati°n  ^
the target/scene.                                                  spectral bands most responsive to

                          GEOGRAPHIC INFORMATION SYSTEMS
 Spatial relationships are apparent when viewed as cartographic products, but to analyze and model spatial
 processes quantitatively, the information needs to be in digital form,  Geographic information systems
 (GIS)  are distinguished from other database management systems by their ability  to perform spatial
 analyses with multiple levels of data in a selected geographic area. Computerized GIS are widely utilized
 to store, query, retrieve, display, and manage large amounts of digital data assembled from many sources.
 This data is typically geographic, environmental, cultural,  statistical or political.  The technological
 migration from costly minicomputers to inexpensive workstations has greatly expanded the user base of
 GIS.   Use of remotely sensed data has become more common in GIS applications as many commercial
 GIS packages  have  image display capabilities  and some  rudimentary  image  processing functions.
 Geographic data can be represented using either vector/polygon or raster/grid data formats.
 GIS does not generate primary data, it captures and processes information in a spatial context and serves
 as a platform on which decisions can be made.   The ability to geometrically  transform and integrate
 multiple data types is very important when accounting for differences in scale, projections, resolutions,
 and coordinate systems.  The analysis of the data can be as simple as measuring line lengths and areas,
.or as complex as using modeling techniques to create "what if scenarios.  Analyses which are commonly
 used include measurements (distance, area, volume), interpretation or processing of a basic layer of data
 to create additional layers, boolean analytical processing, overlay analysis, distance searches, statistical
 calculations, and report generation. It should be noted there are functions which are difficult to perform
 in an image processing system  (raster) that  are relatively  easy  in a GIS (vector) and vice versa.  For
 example, geometric or overlay  operations are easier to perform in the raster domain whereas network
 analysis  or topologic operations are more suited to the vector domain (Ehlers et al.,  1989).

 Both vector and raster methods of representing the spatial extent  of geographical information can be
 translated or interchanged  though the use of data exchange formats.  The conversion of raster or other
 forms  of spatial data into vector form can be approached in several ways.  The most  common is manual
 digitization from existing paper sources which is labor intensive, thereby expensive, and requires  in-place
 quality control procedures  to reduce error propagation.  Data exchange formats  are a common approach
 to solving the solution of  integrating various digital source  data.  Some GIS software packages offer
 conversion packages which transform raster or vector data into a single variable file (SVF) format. The
 (SVF) format is one type of bridge between  vector and raster file structures.

The simultaneous display of raster images with vector cartographic data offers the capability to perform
 change detection analyses.  In  many applications where GIS databases exist, this tool can be used to
 update map resource information on  a regular basis, or for  rudimentary  location and query functions.
 Often just a visual interpretation of geocoded data allows an index to be placed on the surface of the
 image, which then becomes available for interactive queries  in the GIS.  For example, a transportation
 network vector layer might be overlaid on a raster satellite image.  Tne vector fiie then could be edited
 and new attributes connected to features  for updating statistical information if construction altered the
 transportation network.  Over time newly acquired images could be utilized the GIS with minimal cost
 and labor.

 Burrough (1986) lists the following recommendations for the use of vector and raster data structures in

 Vector:  vector ;jr archiving phenomenologically structured data (e.g., soils, land-use units, etc.), for
 network analysis (e.g., transportation, etc.), for high quality line drawing, and digital terrain modeling.

 Raster: raster for map overlays, map combinations, spatial analysis, altitude matrices, and for simulations
 and  modeling when surfaces are encountered.

 Combination Vector/Raster:  for plotting high quality lines with vectors, in combination with efficient
 area filling in color with raster structures such as run length codes or quadtrees.
The most dominant type of GIS are vector-based and have a data structure based on topology to store the
relationships among various spatial objects.  Cartographic information is characterized by point, line, and
area features which are further defined by information about what  is on either side,  and how it is
connected to the other lines.  This coordinate information is then cross-referenced to attribute files which
contain spatial location in relation to other descriptive attributes. Vector/polygon data structures describe
the unique lines or forms  of specific geographic features (streets, lakes,  rivers,  etc).   The coordinate
space is presumed to be continuous,  not quantized as with raster data, which allows all positions, lengths,
and dimensions to be defined with vector data structures (Burrough, 1986). This vector representation
of data is a way to replicate the feature as exactly as possible in a digital form.

Vector GIS have the capability to organize diverse types of data into a single database which contains the
coordinate location of the feature, and it's geographical, cultural, or scientific attributes. These attributes
are the key identifiers in organization and description of the data layers contained within the GIS. The
relational database capabilities allow for access to non-GIS attribute data. Raster data are processed into
a single layer or dimension with one  pixel value before becoming a layer  in a  vector-based  GIS.
Maintenance of these spatial descriptors as a part of the data base, allows GIS to perform normal data
base management functions as well  as spatial manipulations.
In a raster environment, each pixel (or cell) represents one data type spatially, so if a house, tree, and
road intersect at a pixel those elements are put into separate layers. Raster-based systems are compatible
with the data produced  by satellite or aircraft based sensors, and raster-based scanned  maps and
photographs.  These systems produce imagery in raster form comprising a rectangular array of pixels,
which is then analyzed and stored to create a grid map.  The processed data may be in a digital form such
as a thematic  map  created by computer classification, or in graphic form such as a paper map derived
from visual image  interpretation. Raster systems, because of data storage configuration can take longer
to process and display, and usually require more computer disk space. Moreover, raster data may not
explicitly represent feature boundaries, and instead have a stair-stepped appearance.


 Many digital spatial data sources have come into existence,  and the use of such sources compared with
 generating new data is almost always less expensive and time consuming (see Appendix A).  Remote
 sensing can provide a means for acquiring very recent land cover/land use data, and when merged with
 other data (i.e. soils, hydrology, roads, etc.) allows the greatest use to be made not only of the remotely
 sensed data, but all data available to the investigator from other sources.  For some applications maps.
 field  data,  and/or aerial  photography would  need to be  scanned or  manually  digitized into  a
 machine-readable form for integration with existing digital products.  All data layers, whether procured,
 generated through  digital  image processing,  or manual  digitizing, should have reported  spatial  and
 thematic accuracies associated with the real world.  When acquiring digital data from reliable sources,
 minimum standards for accuracy are usually ensured. Data of questionable integrity should be eliminated
 from consideration unless  the intent is  to use the information  as a data layer on which  updates.
 refinements, or corrections are to made later when more accurate  information becomes available. The
 overall  accuracy of any CIS database is dependent upon the data layer with  the lowest accuracy or
 resolution.  The integration of various types of spatial and geographic information enables remote sensing
 to provide  useful input for  a broad range of  applications  in a  cost-effective  manner.  For many
 environmental  applications,  remotely sensed  data represents only one  source of  input for studying
 complex environmental problems. Some other sources of spatial data include maps, aerial photographs,
 census information, field measurements, and meteorological records.

 The challenge then, is not only where to get information but how to integrate information flow between
 different geoprocessing technologies.  Once a CIS is functioning, information can be drawn fr TI a single,
 consistent, uniform database which avoids the possibility of having separate collections of competing data.
A comprehensive CIS will support various computer mapping and graphic products.

CIS  rovides  a convenient and organized method for analysis of wetland ecosystems.  CIS  provides
flexibility in which to build a cohesive database on which additional information can be incorporated, and
refinements made if project objectives change. Relationships can be modeled and different scenarios put
forth that may need to be addressed for impact assessments or  trend analyses on the wetland ecosystem.
The results  of such capabilities provide the resource/project manager the best information available for
better protection and management of wetland resources.

                         EXAMPLE PROJECT COSTS
Each wetland resource mapping project is highly individualized depending upon legal requirements and/or
the needs of the agency conducting the investigation.  The following estimated project costs should be
regarded with caution as differing study objectives result in varying costs.

Communication must begin in the planning phase of any inventory project, and focus on data acquisition
requirements and subsequent processing/interpretation to ensure that the final product will meet the user's
needs.  Decisions made regarding the scope of such endeavors must include the RS/GIS specialists so any
preconceived expectations  can be realistically evaluated and modified if necessary.

Table 2 contains generic information regarding remote sensing system specifications and acquisition costs.
Table 3 contains summary information pertaining to described projects/programs found described later
in this  section.
                            & ACQUISITION COSTS
                   Sensor Costs   '   Spatial     Temporal     Coverage
                   (sq. mile)       Resolution                 (sq. mile)
CIR Photo
Aircraft MSS
& CIR Photo
1.1 x 1.1 km
30 x 30m
10 x 10 m
15 x 15m
1-3 m
16 days
26 days


 The following resource mapping projects were undertaken at EPA's Environmental Monitoring Systems
 Laboratory in Las Vegas.

 Aquatic  Macrophvte Mapping - Clark Fork & Tributaries. MT

 Color infrared  photography,  1:18,000 scale,  were collected  July  1988 concurrently with  aircraft
 multispectral scanner data.  The purpose  of the project was shoreline interpretation of wetlands and
 associated vegetation (algal blooms,  rooted macrophytes, and riparian tracts),  and non-point pollution
 source features  (cropland, pasture, confined feeding, landfills, waste water treatment facilities, outfall
 locations, timber harvests, golf courses/urban recreation, nursery/orchard, industrial, commercial, and
 residential).  The areal coverage and interpretation efforts centered on several  river courses located in
 Montana, Idaho, and Washington.  Only the area contained within the photographs were analyzed and
 mapped  for the  above class groupings.

 Project  costs reflect  EMSL project management, photo  acquisition and  film  processing,  manual
 photo-interpretation of 740 flight line miles, graphic personnel support, and a deliverable of bound reports
 (3 volumes/set)  containing the photo-interpreted overlays attached to the photographs. Not included are
 costs associated with aircraft usage,  flight  crew,  and aircraft maintenance.   No map  transfer  or
 digitization of the photo-interpreted data was performed. The cost for this project is estimated at S37.00
 per square mile ($0.06 per acre).
Lake Pend Oreille Watershed Characterization Using Landsat TM Data

A Landsat Thematic Mapper subscene (100km X 100km) dated July 1989 was used in this analysis.  An
Anderson Level 1 land cover characterization (forest, agriculture, rangeland, barren, water, wetland, and
urban) was  derived using unsupervised image processing techniques.  The resulting overall accuracy
reported for only  the forest,  agriculture,  rangeland, and barren  categories was  78% (Lee,  1990).
Statistical data were generated regarding areal extent of land cover resources for approximately 1,000
square miles (640,000 acres).  This information was used by the State of Idaho for non-point pollution
modeling efforts for the Lake Pend Oreille watershed.

Project costs reflect EMSL project management, TM imagery acquisition, supporting NHAP photographs.
image processing,  accuracy assessment, final report,  and graphic products (slides and map acetate
overlays at 1:100,000 scale).  The cost is estimated at $20.62 per square mile ($0.03 per acre).
          .   __«....*. -.  _  *«_*__«-._.   ___   ___ »-         -
         iJ liaTuwOOua lucntiiiwottn wfim  i^onaaal i ivi i-/«no
Subsets of two Landsat Thematic Mapper scenes dated April 1988 and August 1988 were used in a pilot
project  to  assist in advanced  identification of wetlands.  The analysis  focused on four 7.5  minute
quadrangles (approximately 240 sq. miles) located in southern Illinois in which several different image
processing methodologies  were  tested for optimum bottomland  hardwood wetland  identification.
Completion of this portion if the pilot is scheduled for September 1990.  Once completed, the next phase
will apply the' derived results to a watershed level analysis of wetland  resources  (not included  in cost


estimation).  An Anderson Level 1 land  cover was derived, except for the wetland  class which was
further  subdivided  into a Level 2  classification (forested vs.  non-forested wetlands).   The overall
agreement between  the satellite-based classification and photo-interpreted CIR photography collected in
August  1988, was approximately 81 percent.  A field site verification effort is planned for spring 1991,
with an accuracy assessment to follow.

Project costs reflect EMSL project management, TM imagery acquisition, supporting photo-interpretation
of  existing CIR  imagery  (1:24,000),  image processing,  accuracy  comparisons between  the  TM
classification and photo-interpretation, final report, and graphic products (slides and plots).  The cost is
estimated at $122.00 per square mile ($0.19 per acre).
Pearl River Wetlands Advanced Identification

A CIS database was constructed for the Pearl River Watershed (approximately 200 sq. miles) located in
southern Louisiana in which layers were generated for hydrography, soils, land cover, transportation,
elevation, and field reference data. The data layers were integrated or compiled from digital and map
sources obtained from Landsat TM imagery (land cover), NWI, SCS, and USGS. Two dates of Landsat
TM imagery were obtained from pre-copyrighted archived data. Wetlands were identified as wetland vs.
non-wetland, and further subdivided into local species such as bottomland hardwoods and cypress/tupelo
groupings.  Overall accuracy between the CIS based wetland identification and  the field verification
information was 85% (Pickus, 1990).

Project costs reflect EMSL project management, TM image processing,  CIS data layer digitization or
integration, development of a user friendly CIS interface (ARC/INFO), and a formal project report: The
cost  is estimated at $133.00 per square mile ($0.21 per acre).


The production of NWI information is performed by blocks of USGS quadrangles.  USGS color infrared
NHAP and NAPP photography are photo-interpreted according to the Co ward in classification (Cowardin,
et al., 1979) with interpreted results transferred to the appropriate quadrangles. These maps are available
in paper form by quadrangle, and subsequently become digitized in selected areas  where cost-sharing
between the USF&WS and the customer/user is provided. To date, 60% of the conterminous U.S., and
20% of Alaska have been completed.  The average cost per acre ranges from $0.03 to $0.06,  depending
on the  number and complexity of wetlands found within the quadrangle (D. Woodward, USFWS-NWI,
personal  communication,  1990.)
The following project  is being performed by the U.S.  Army Corps of Engineers, Detroit District,
Engineering Division, Great Lakes Hydraulics & Hydrology Branch.

Resource Analysis and Land Cover/Current Use Inventory Data Base for the U.S. Side of the Great
Lakes. 1989 - Present.


A CIS incorporating land cover/current use inventory information was created with source information
derived from photo-interpreted CIR aerial photography (1:24,000 dated Aug/Sept 1988) for approximately
1500 linear miles of the  U.S.  Great Lakes  shoreline excluding the State of Michigan.  The State of
Michigan has  been constructing a digital  land  cover data base over the  past  ten  years  utilizing
photo-interpreted 1:24,000 scale CIR photography. The land cover/current use classification used in this
Corps project is based upon the classification system utilized by the State of Michigan, Department of
Natural Resources called MIRIS (Michigan Resource Inventory System). The MIRIS classification is a
refined  Anderson Level 3 which  subdivides the wetland category into wooded wetlands (wooded and
shrub/scrub classes), non-wooded (aquatic beds, emergents, flats), and Great Lakes coastal submergent
classes.  The final photo-interpreted map  overlays are digitally scan-encoded into vector format.

Project costs  are based on contracts which required photo acquisition and film processing, manual
interpretation, map transfer, and digital scan encoding/quality control to vector form (Intergraph).  The
costs do not include Corps district project management or additional coordination provided by the  State
of Michigan. The cost is estimated at $42.00 per square mile ($0.07 per acre) (R.L. Gauthier, USACE,
personal communication,  1990).

STATE MAPPING PROGRAMS   (Also see Appendix B)

The Florida Department of Transportation (DOT), State Topographic Office mapped the state's wetland
resources under contract to the Florida Freshwater Gaming Commission. Landsat TM data from 1986-88
for the  entire state was classified to a user-specified Anderson Level 3.  The TM data analysis was
supplemented with soils  data  and  existing photography  (CIR  1:40,000, B&W 1:24,000) to allow
refinement of classes during production. The classification categories included: coastal strand, saltmarsh,
wet and dry prairies, swamps (mangrove, cypress, hardwood, bay), pine forest, mixed hardwoods, upland
forest, exotic species, disturbed communities,  bare  ground, and open water. The classified data is
available only in digital form with  inherent 30 meter pixel resolution.  Project costs reflect TM data
acquisition, DOT image analysis, and limited field reviews. The costs do not include Freshwater Gaming
Commission project coordination, or their independent field reviews of the final classification. The cost
is estimated at $9.50 per square mile or $0.02 per acre (G. Maudin/A. Shopmyer, Florida DOT, personal
communication, 1990).

In 1972, the state passed the Marshland Enforcement Act which mandated an aerial surveillance program
be established to monitor  and regulate the state's coastal estuaries.  Black-and-white aerial photography
is periodically acquired at a scale at 1:40,000.  In addition, routine light aircraft  and helicopter flights
collect 35roffi photos to monitor shoreline development, assure compliance to state laws, and to locate
unauthorized developments. Reports are generated on a case-by-case basis for unauthorized activities.
No standard minimum mapping unit was given, but is estimated to be between 0.5 and 1 acre (S. Stevins,
GDNR, personal communication, 1990).  No cost data were available.

In Michigan, the entire state has been encoded into a digital GIS (intergrapn) by the Michigan Department
of Natural Resources (MDNR), Michigan Resource Inventory System (MIRIS).  Color infrared aerial
photography (dated 1979  and 1985,  1:24,000) were photo-interpreted, for land cover/land  use with a
minimum mapping unit of one  acre.  Wetlands were mapped to an revised Anderson Level  3 flowland


hardwood, shrub wetland, emergent wetland, and aquatic beds).  Data is available by county in a variety
of formats (digital, map, and statistical). The cost is estimated at S39.00 per square mile or $0.06 per
acre (M. Scieszka, MDNR, personal communication, 1990).

In Wisconsin, the entire state's wetlands were mapped in approximately five years and was completed
in 1984 by the Wisconsin Department of Natural Resources  (WDNR).   Black  and white  infrared
photography at a scale of 1:20,000 were photo-interpreted with acquisition costs averaging $30,000 per
county. Interpretative data was transferred to 1:24,000 scale township-centered rectified base maps with
a minimum mapping unit of two acres.  Data is compiled on a county-by-county basis with three to four
counties revised annually.  All current work is performed by contractors. The state is in the process of
creating a GIS which incorporates both past and ongoing work. The costs reflect Wisconsin DNR project
management, photo acquisition and film processing, photo-interpretation, map  transfer, and manual
digitization of the map  overlays.  Maps are available in paper and digital formats where digitization is
completed.  The cost is estimated at $43.00 per square mile or $0.07 per  acre (L. Stoerzer, WDNR,
personal communication, 1990).
                         PROJECT AMD PROGRAM SUMMARY TABLE
  Data   Analytic
           Map       Encoded
                       Map Unit
Clark Fork
Pearl R.

0.5  acre
1-2  acre
1-2  acre
1-2  acre
1-2  acre
1-2  acre
1 acre
0.5-1 acre
1 acre
2 acre
Analysis Symbols
AP  -  Aerial photography
MSS - Aircraft multispectral  scanner
TM  -  Landsat Thematic Mapper
LC  -  Land  Cover
LU  -  Land  Use
WL  -  Wetland
NPS - Non-point source

 Anderson, J.R., E.E. Hardy, J.T. Roach, and R.E. Witmer,. 1976.  "A Land Use Cover Classification
 System For Use With Remote Sensor Data", U.S.  Geological Survey Professional Paper 964, 27p.

 Brown, W.W., 1978. "Wetland Mapping In New Jersy and New York", Photogrammetric Engineering
 and Remote Sensing, Vol. 44 No. 3, pp. 303-314.

 Burrough, P.A., 1986.  Principals of Geographic Information Systems for Land Resources Assessment.
 Oxford University Press.  New York.

 Carter, V., 1982.  "Application of Remote Sensing to Wetlands".  Jn_Remote Sensing for Resource
 Management  (C.J. Johannsen and J.L. Sanders, eds.), Soil Conservation Society of America, Ankeny,

 Carter, V., D.L.  Malone, and J.H. Burbank, 1979.  "Wetland Classification and Mapping In Western
 Tennessee", Photogrammetric Engineering and Remote Sensing, Vol. 45, No.3, pp. 273-284.

 Cowardin, L.M., V. Carter,  F.C. Golet, and E.T. LaRoe, 1979.   "Classification of Wetlands and
 Deepwater Habitats of the United States".  U.S. Fish and Wildlife Service, Report No. FWS/OBS-70/31,

'Duggan, J.S., 1983. "Puget Sound Wetlands Inventory: Photography Volumes 1-6", EPA-Environmental
 Monitoring Systems Laboratory, Las Vegas, NV.  Report No. TS-AMD-82072.

 Driscoll, D.,  1990.  "Remote Sensing:  USFS Pest Management Group" in CIS World  Magazine,

 Ehlers, M., G. Edwards, and Y. Bedard, 1989.   "Integration of Remote  Sensing with Geographic
 Information Systems: A Necessary Evolution", Photogrammetric Engineering Ar.u Remote Sensing, Vol.
 55, No. 11, pp. 1619-1627.                           .  '

 Ehlers, M., R.J. Hintz, and R.H. Greene, 1989. "High Resolution Airborne Video System for Mapping
 and  GIS  Applications",  Proceedings  12th  Biennial Workshop on  Color  Aerial Photography and
 Videography, American Society of Photogrammetry and Remote Sensing, pp. 171-177.

 Ernst-Dottavio, C.L.,  R.M.  Hoffer, R.P.  Mroczynski,  1981.   "Spectral Characteristics of Wetland
 Habitats", Photogrammetric Engineering and Remote Sensing, Vol.47, No.2, pp.  223-227.

 Everirt, J.H. and D.E. Escobar, 1989. "The Staus  of Video Systems for Remote Sensing Applications",
 Proceedings 12th Biennial Workshop on Color Aerial Photography and Videography, American Society
 of Photogrammetry and Remote Sensing, pp. 6-29.

 Federal Interagency Committee for Wetland Delineation, 1989. Federal Manual for Identifying and
 Delineating Jurisdictional Wetlands.  US Army  Corps  of Engineers, US  Er.vircnincntal Protection
 Agency, US Fish and Wildlife Service, and USDA Soil Conservation Service.  Cooperative technical
 publication, 76 pages plus appendices.


Gammon, P.T. and V. Carter, 1979. "Vegetation Mapping With Seasonal Color Infrared Photographs",
Photogrammetric Engineering and Remote Sensing, Vol. 45, No. 1, pp. 87-97.

Grace, J.B., 1985.  "Historic Macrophyte Development in Par Pond", Submitted to the U.S. Dept. of
Energy, Environmental Sciences  Division,  Savannah River Laboratory, Report No. DPST-85-841,
Aieken, South Carolina.

Hardin, D.L., 1985.  "Remote Sensing of Wetlands for Fish and Wildlife Habitat Management in
Delaware - A Comparison of Data Sources",  Integration of Remotely Sensed Data In GIS for Processing
of Global Resource Information, CERMA Proceedings, Washington, DC.

Hardisky, M. A., and V. Klemas,  1983. "Tidal Wetlands Natural and Human-Made Changes from 1973
to 1979 in Delaware: Mapping Techniques and Results", Environmental Management, Vol. 7, No. 4, pp.

Heaslip, G.G., 1975. "Environmental Data  Handling". John Wiley and Sons, New York.

Hodgson, ME., J.R. Jensen, H.E. Mackey, and M.C. Coulter, 1988.  "Monitoring Wood Stork Foraging
Habitat Using Remote Sensing and GIS", Photogrammetric Engineering and Remote Sensing, Vol. 54,
No. 11, pp.  1601-1607.

Howarth, P.J. and G.M. Wickware,  1981.  "Procedures for Change Detection Using Landsat Digital
Data", Journal of Remote Sensing, Vol. 2, pp. 277-291.

Rowland,  W.G.,   1980.   "Multispectral Aerial  Photography  for  Wetland Vegetation Mapping",
Photogrammetric Engineering and Remote Sensing, Vol. 46, No. 1, pp. 87-99.

Jacobson, J.E., R.A. Ritter, and G.T. Koeln, 1987.  "Accuracy of Thematic Mapper Derived Wetlands
As Based On National Wetland Inventory Data", ASPRS/ACSM/WFPLS Fall Convention, American
Society of Photogrammetry and Remote Sensing, Falls Church, VA.

Jensen, J.R., E.W. Ramsey, H.E. Mackey, EJ. Christensen, and R.R. Sharitz, 1987. "Inland Wetland
Change Detection  Using Aircraft MSS Data", Photogrammetric Engineering and Remote Sensing, Vol.
53 No. 5, pp. 521-529.

Jensen, J.R., M.E. Hodgson, E.J. Christensen, H.E. Mackey, L.R. Tinney, and R.R. Sharitz,  1986.
"Remote Sensing Inland Wetlands: A Multispectral Approach", Photogrammetric Engineering and Remote
Sensing, Vol. 52,  No. 1, pp.  87-100.

Jensen, J.R., 1986.  "Introductory Digital Image Processing".  Prentice Hall, New Jersey.

Jensen, J.R., M.  Hodgson, E.J. Christensen, H.E. Mackey,  and S.S. Sharitz,  1984.  "Multispectral
Remote Sensing of Inland Wetlands in South Carolina: Selecting the Appropriate Sensor", Submitted to
the U.S. Dept. of Energy, Savannah River Laboratory, Aieken, S.C.

Jensen, J.R., EJ.  Christensen, and R. Sharitz, 1984.  "Non-Tidal Wetland Mapping in South Carolina
Using Airborne Multispectral Scanner Data", Remote Sensing of Environment, 16:1-12.

Jensen, J.R., I.E. Estes, and L. Tinney, -1980.  "Remote Sensing Techniques for Kelp Surveys",
Photogrammetric Engineering and Remote Sensing, Vol. 46, No. 6, pp. 743-755.

Lee, K.H., 1990.  "Internal Report-Watershed Characterization Using Landsat Thematic Mapper (TM)
Satellite Imagery: Lake Pend Oreille, Idaho".  EPA-Environmental Monitoring Systems Laboratory, Las
Vegas, Report No. TS-AMD-90C10.

Lillesand, T.M. and R.W. Kiefer, 1987. "Remote Sensing and Image Interpretation". John Wiley and
Son$, New York.

Luman, D.E., 1990.  "The Potential for Satellite-Based Remote Sensing Update of the Illinois Portion
of the National Wetlands Inventory",  Quarterly Report, submitted  to  the Illinois  Department of

Mack, W.M., 1980.  "Aerial Survey of Utah  Wetlands", EPA-Environmental Monitoring Systems
Laboratory, Las Vegas, NV.  Report No. TS-AMD-08047.

Mackey Jr., H.E., 1990.  "Monitoring Seasonal and Annual Wetland Changes in a Freshwater March
with SPOT  HRV Data",  Technical Papers  1990 ASCM-ASPRS Annual Convention, pp. 283-292.
American Society of Photogrammetry and Remote Sensing, Falls Church,  VA

Martin, A.C., Neil Hotchkiss, P.M. Uhler, and W.S. Brown, 1953.  "Classification of Wetlands of the
United States", Special Science Report  Number 20. U.S. Fish and Wildlife Service, Washington, D.C.,
14 pp.

May Jr., L.N., 1986.  "An Evaluation of landsat MSS Digital Data for Updating Habitat Maps of the
Louisiana Coastal Zone", Photogrammetric Engineering and Remote Sensing, Vol. 52, No. 8, pp. 1147-

Mynar H, F.,  1990.  "Classification of Puget Sound Nearshore Habitats  Using Aircraft Multispectral
Scanner Imagery", EPA, Environmental Monitoring Systems Laboratory, Las Vegas,  Report TS-AMD-

Newcomer, J.A. and J.  Szajgin, 1984.  "Accumulation  of Thematic Map Error in Digital Overlay
Analysis", The American Cartographer, Vol.11, No.  1, pp. 58-62.

Niedzwiedz, W.R. and S.S. Batie, 1984. "An Assessment of Urban Development into Coastal Wetlands
Using Historical Aerial Photography: A Case Study", Environmental Management, Vol. 8, No. 3, pp.

Norton, D.J.,  1986a.   "Suitability Study of Chincoteague Wetlands, Virginia",  EPA-Environmental
Photographic Interpretation Center, Vint Hill, VA. Report No. TS-AMD-85037.

Norton, D.J., 1986b. "Initial Feasibility Study: Wetlands Identification Sussex County, Delaware", EPA-
Environmental Photographic Interpretation Center, Vint Hill, VA. Report No. TS-AMD-86071.

Norton, D.J., S.W. Engle, and J.D. Simmons, 1985. "Water Resources Management Applications of
Remote Sensing at EPA's Environmental Photographic Interpretation Center", Proceedings of the 51st
Meeting American Society of Photogrammetry and Remote Sensing, Washington, D.C.

Norton, D.J.  and J. Prince, 1985.  "Using Remote Sensing for Wetlands Assessment in Superfund
Hazardous Waste Sites", Proceeding of the 19th International Symposium on  Remote  Sensing of
Environment, ERIM, Ann Arbor, MI.

Page, S.H., 1982. "Remote Sensing of Kelp Beds Demonstration Report: Channel Islands National Park,
California", EPA-Environmental Monitoring Systems Laboratory, Las  Vegas, Report No. TS-AMD-

Place J.L., 1985.  "Mapping of Forested  Wetland:  Use of Seasat RADAR Images to Complement
Conventional Sources", Professional Geographer, 37(4), pp. 463-469.

Pickus, J., 1990. "Pearl River Wetlands Advanced Identification: A Geographical Information Systems
Demonstration Project", EPA-Environmental Monitoring Systems Laboratory, Las Vegas, Report No.
215-90C05.                                                                  ,

Roller, N.E.G., 1977.  "Remote Sensing of Wetlands",  Environmental Research Institute of Michigan,
(NASA-CR-153282), Ann Arbor, Michigan, 165 pp.

Salvaggio, C. and P. Szemkow,  1989. "Generation of High-Resolution Hybrid Multiband Imagery from
Existing High-Resolution Panchromatic and Low-Resolution Multispectral Images", Technical Papers,
1989 ACSM-ASPRS Annual Convention, American  Society of Photogrammetry and  Remote Sensing,
Falls Church, VA.

Savastano, K.J., K.H. Faller, and R.L. Iverson, 1984.  "Estimating Vegetation Coverage in St. Joseph
Bay, Florida  with  an Airborne Multispectral Scanner", Photogrammetric Engineering and  Remote
Sensing, Vol. 50, No. 8, pp. 1159-1170.
Shaw, S.P. and C.G. Fredine, 1956.  "Wetlands of the United States", U.S. Fish and  Wildlife Service,
Circular 39, 67 pp.                                                                        (

SPOT Image Corporation, 1897 Preston White Drive, Reston, VA 22091, (703)620-2200.

Story, M. and R.G. Congalton,  1986. "Accuracy Assessment: A User's Perspective",  Photogrammetric
Engineering and Remote Sensing, Vol. 52, No. 3, pp. 397-399.

Swartwout, D.J., W.P. MacConnell, and J.T. Finn,  1981.  "An Evaluation of the National Wetlands
Inventory in Massachusetts",In-Place Resource Inventories Workshop, University of Maine, Orono, ME,
pp. 685-691.

Welch, R. and  M. Ehlers, 1987.   "Merging Multifesolution SPOT HRV and Landsat  TM Data",
Photogrammetric Engineering and Remote Sensing  Vol. 53, No. 3, pp. 301-303.

Williams, D.C. and  J.G.  Lyon, undated.  "Use of A Geographic Information System.Data Base to
Measure and Evaluate Wetland Changes in the St. Marys River, Michigan", Submitted to the Detroit
District, U.S. Army Corps of Engineers.

Williams, D.R., 1989. "Aerial Photographic Analysis of Utah Sand and Gravel Company Settling Ponds,
Salt Lake County, Utah", EPA-Environmental Monitoring Systems Laboratory, Las Vegas,  NV. Report
No.TS-PIC 88218.

Williams, D.R.,  1985.  "Current and Historical Photographic Inventory of Agricultural Development,
Great Cedar Swamp, Middlesborough County, Massachusetts", EPA-Environmental Monitoring Systems
Laboratory, Las Vegas, NV. Report No. TS-AMD-85532.

Williams, D.R.,  1983.  "Wetland Delineation and Photography: Piceance Creek, Colorado and James
River, South Dakota", EPA-Environmental Monitoring Systems Laboratory, Las Vegas, NV. Report No.

Williams, D.R.,  1981.  "Historical Inventory of Development Activity M & K Ranches, Gulf County,
Florida", EPA-Environmental Monitoring Systems Laboratory, Las  Vegas, NV.  Report No. TS-AMD-

Wood, B.L., 1983.  "Wetland Mapping  in Colusa  County, California",  NASA/AMES  International
Renewable  Resource Inventions for Monitoring Conference, Corvallis, OR.

Wu, S.T., 1989.  "Utility of a Digital Video and Image Analysis System for Forest and Coastal Wetland
Mapping",   Proceedings  12th  Biennial Workshop  on Color Aerial Photography and Videography,
American Society of Photogrammetry and Remote Sensing, pp.  164-170.

 Alam, M.S., S.D. Shamsuddin, and S. Sikder, 1990.   "Application of Remote Sensing For Monitoring
 Shrimp Culture Development In A Coastal Mangrove Ecosystem In Bangladesh", Technical Papers, 1990
 ACSM/ASPRS Annual Convention, American Society of Photogrammetry and Remote Sensing, Falls
 Church, VA, Vol. 4, pp. 23-32.

 Balogh, M.E. and D.L. Becker, 1986.  "Riparian Vegetation Inventory: Parker II Division of the Lower
 Colorado River, Blythe, California", EPA-Environmental Monitoring Systems Laboratory, Las Vegas,
 Report No. TS-AMD-85561.

 Bartlett, D.S., V. Klemas, O.W. Crichton, and G.R. Davis, 1976.   "Low-Cost Aerial Photographic
 Inventory of Tidal  Wetlands", University of Delaware, College of Marine Studies, Newark, DE.
 Submitted to Dept. of Natural Resources and Environmental Control, State of Delaware.

 Benson,  A.S. and  S.D. DeGloria,  198S.   "Interpretation Of Landsat-4  Thematic Mapper and
 Multispectral Scanner Data For Forest Surveys", Photogrammetric Engineering and Remote Sensing, Vol.
 51, No. 9, pp. 1281-1289.

 Best, R.G., M.E. Wehde, R.L. Linder, 1981.  "Spectral Reflectance of Hydrophytes". Remote Sensing
.of Environment, Vol. 11, pp. 27-35.

 Bogucki, D.J., 1978.   "Remote Sensing To Identify, Assess, and Predict Ecological Impact On Lake
 Champlain Wetlands", State University of New York, College at Pittsburgh.

 Boule, M.E.,  G.B. Shea,  1978.   "Snohomish Estuary  Wetlands  Study",  Northwest Environmental
 Consultants Inc., Seattle, Washington. Submitted to Seattle District, U.S. Army Corps of Engineers.

 Byrne, G.F., P.P. Crapper, and K.K. Mayo,  1980.  "Monitoring Land-Cover Change by Principal
 Components Analysis of Multi-temporal Landsat Data", Remote Sensing of Environment, Vol. 10, pp.
 175-184.                         .

 Carter, V. and D.G. Smith, 1973.  "Utilization Of Remotely Sensed Data In The Management Of Inland
 Wetlands", U.S. Geological Survey, Contracts IN-385 and 1-414, 15p.

 Chavez, P.S., C. Guptill, and J.A.  Bo well, 1984.  "Image Processing Techniques For Thematic Mapper
 Data", Technical Papers, 50th Annual  Meeting of The American Society of Photogrammetry,  Vol. 50,
 No.  2, pp. 728-742.
 Civco, D.L.,  1989.  "Topographic Normalization of Landsat Thematic Mapper Digital Imagery",
 Photogrammetric Engineering and Remote Sensing, Vol. 55, No. 9, pp.  1303-1309.

 Civco, D.L., W.C. Kennard, and  M.W. LeFor, 1978.  "A Technique For Evaluating Inland Wetland
 Photointerpretation: The Cell Analytical Method (CAM)",  Photogrammetric Engineering  and Remote
 Sensing, Vol. 44, No. 8, pp. 1045-1052.

Congalton, R.G., 1988. "A Comparison of Sampling Schemes Used In Generating Error Matrices for
Assessing the Accuracy of Maps Generated from Remotely Sensed Data", Photogrammetric Engineering
and Remote Sensing ,  Vol. 54, No.5, pp. 593-600.

Cowardin, L.M. and   V.I.  Myers,  1974.   "Remote Sensing for Identification and Classification of
Wetland Vegetation", Journal of Wildlife Management, 38(2), pp. 308-314.

Dottavio,  C.L., and F.D. Dottavio,  1984.   "Potential Benefits of New Satellite Sensors To Wetland
Mapping", Photogrammetric Engineering and Remote Sensing, Vol. 50. No. 5, pp.599-606.

Ernst, C.L., and R.M. Ho.ffer, 1981.  "Using Landsat data with Soils Information to Identify Wetland
Habitats", In Satellite Hydrology.  Proceedings, Pecora V Symposium.  American  Water Resources
Association, Minneapolis, Minnesota.

Farmer, A.M., and M.S. Adams, 1989. "A Consideration of the Problems of Scale in the Study of the
Ecology of Aquatic Macrophytes", Aquatic Botany, Vol.  33, pp. 177-189.

Federal  Interagency Committee for Wetland Delineation, 1989. Federal Manual for Identifying and
Delineating Jurisdictional Wetlands.  U.S.  Army Corps  of Engineers, U.S. Environmental Protection
Agency, U.S.  Fish and Wildlife Service, and U.S.D.A. Soil Conservation Service, Washington, D.C.
Cooperative technical publication, 76 pp. plus appendices.

Prayer,  W.E., T.J. Monahann,  D.C.  Bowden, and F.A. Graybill, 1983.    "Status and Trends of
Wetlands and Deepwater Habitats in  the Conterminous U.S. 1950's to 1970's",  U.S. Fish & Wildlife
Service, National Wetlands Inventory, St. Petersburg, FL.

Gilmer, D.S.,  E.A. Work, J.E. Colwell, and D.L. Rebel,  1980.  "Enumeration of Prairie Wetlands  With
Landsat and Aircraft Data", Photogrammetric Engineering and Remote Sensing, Vol. 46, No. 5, pp. 631-

Hall,  L.B., R.C. Clar, J.D. Von Loh,  J.N. Halls, M.J. Pucherelli, and R. McCabe,  1988. "The Use
of Remote Sensing and GIS Techniques for Wetland Identification and Classification in the Garrison
Diversion Unit-North  Dakota", Technical Papers  1988 ACSM-ASPRS  Annual Convention, American
Society of Photogrammetry and Remote Sensing, Falls Church, VA.

Jaynes, R.A., L.D. Jr., Clark, and K.F. Landgraf, "Inventory of Wetlands and Agricultural Land Cover
in the Upper Sevier River Basin, Utah", University of Utah, Center for Remote Sensing and Cartography,
Salt Lake  City, UT.

Jensen, J.R. and B.A. Davis, 1987.  "Remote Sensing of Aquatic Macrophyte Distribution in Selected
South Carolina Reservoirs", Technical Papers 1987 ASPRS-ACSM Annual Convention, Vol. 1,  pp. 57-
65. American Society of Photogrammetry and Remote Sensing, Falls Church, VA.

Jem;in,  J.R.,  J.E. Estes, and L. Tinney, 1980.   "Remote Sensing Techniques for Kelp Surveys",
Photogrammetric Engineering and Remote Sensing, Vol.  46, No. 6, pp. 743-755.

Johnson, M.O. and W.D. Goran, 1987.  "Sources of Digital Spatial Data for Geographic Information
Systems", U.S. Army Corps of Engineers, Construction Engineering Research  Laboratory, Technical
Report No. N-88-01,  33 pp.

Karaska, M.A., S.J. Walsh, and D.R. Butler, 1987.  "Impact of Environmental Variables on Spectral
Signatures Acquired by the Landsat Thematic Mapper", Technical Papers, 53rd  Annual Meeting of the
American Society of Photogrammetry and Remote Sensing, Vol. 1, pp. 371-384.

Kempka, R.G. and R.P.  Kollasch,  1990.   "California Waterfowl Habitat  Evaluation Using Remote
Sensing Techniques",  Ducks Unlimited Contract Final Report: California Department of Fish and Game
Wildlife Management Division, 92 pp.

Klemas, V., 1983.  "Evaluation of Spatial Radiometric and Spectral Thematic Mapper Performance for
Coastal Studies", Quarterly Status Report, Delaware University, 4p.

Klemas, V. and M.A. Hardisky, undated. "Remote Sensing of Estuaries: An Overview", University of
Delaware, College of Marine Studies, Newark, Delaware.

Laboratory for Application of Remote Sensing (LARS) Purdue University, 1979. "Application of Remote
Sensing Technology to the  Solution of Problems  in the Management of Resources in Indiana", W.
Lafayette, .Indiana.

Long,  K.S.,  1979.   "Remote Sensing of Aquatic Plants",  USAGE Waterways Experiment  Station
Technical Report No. A-79-2, Vicksburg, MS.

Lyon,  J.G.,  1979.  "Remote Sensing Analysis of Coastal Wetland Characteristics:  St. Clair Flats,
Michigan", Proceedings of the 13th International Symposium on Remote Sensing of Environment, ERIM,
Ann Arbor, MI, pp. 1117-1129.

Mackey, H.E. Jr. and J.R.  Jensen,  1988.  "Macrophyte Mapping with Video Technology in a Fresh
Water Lake", Proceedings of the First Workshop on Videography, American Society of Photogrammetry
and Remote Sensing, pp. 86-71, 265.

Mackey, H.E. Jr.  and J.R.  Jensen, 1988.  "Remote  Sensing of Wetlands applications Overview",
Proceedings of the First Workshop on Videography, American Society of Photogrammetry and Remote
Sensing, pp.  32-33.

Mackey, H.E. Jr., J.R. Jensen,  M.E. Hodgson, and K.W. O'Cuillin, 1987.  "Color Infrared Video
Mapping of Upland and Wetland Communities", Proceedings of the llth Biennial Workshop Color Aerial
Photography and Videography in the Plant Sciences. American Society of Photogrammetry and Remote
Sensing, pp. 252-260. •

Masry, S.E. and S. MacRitchie, 1980.  "Different Considerations in Coastal Mapping", Photogrammetric
Engineering and Remote Sensing, Vol. 46, pp. 521-528.

McEwen, R.B.,  W.J. Kosco, and V. Carter, 1976.  "Coastal Wetland Mapping", Photogrammetric
Engineering and Remote Sensing, Vol. 42, No.2, pp. 221-232.

Mead,  R.A. and P.T. Gammon, 1981. "Mapping Wetlands Using Orthophotoquads and 35mm Aerial
Photographs", Photogrammetric Engineering and Remote Sensing, Vol. 47, No. 5, pp. 649-652.

Meisner, D.M., 1986.  "Fundamentals of Airborne Video Remote Sensing", Remote Sensing of
Environment, 19:63-79.

Meisner, D.M.  and D.E. Lindstrom, 1985. "Design and Operation of Color Infrared Aerial Video",
Photogrammetric Engineering and Remote Sensing, Vol. 51, No. 5, pp. 555-560.

Mouchot, M.C., G.  Sharp, and E. Lambert, undated.  "Thematic Cartography of Submerged Marine
Plants Using the Fluorescence Line Imager", Canada Centre for Remote Sensing, Ottawa, Ontario.

Nelson, R.W., W.J. Logan, and E.G. Weller, 1983.  "Playa Wetlands and Wildlife on the Southern
Great Plains: A Characterization of Habitat", U.S. Fish and Wildlife Service, Fort Collins, Colorado,
Report No. FWS/OBS-83/28, 163 pp.

Newbury, G.E., 1981.  "Changes in the Wetlands of Hunting Creek, Fairfax County, Virginia", U.S.
Army Engineer Topographic Labs, Fort Belvoir, VA.

Nixon, P.R. D,E. Escobar, and R.M. Menges, 1985.  "Use of Multi-band Video System for Quick
Assessment of Vegetal Condition and Discrimination of Plant Species", Remote Sensing of Environment,

Pelletier, R.E., R.T. James, and J.C. Smoot, 1990.  "Evaluation Hydrologic Modelling Components of
the Florida Everglades With AVHRR and Ancillary Data", Technical Papers, 1990 ASCM-ASPRS
Annual Convention,  Vol. 4, pp. 321-330, American Society of Photogrammetry and  Remote Sensing,
Falls Church, VA.

Sabins  Jr. F.F., 1987.   Remote Sensing Principals and Interpretation.  W.H. Freeman and Company,
New York.

Scarpace, F.L., B.K. Quirk, R.W. Kiefer, and S.L. Wynn, 1981.  "Wetland Mapping from Digitized
Aerial Photography", Photogrammetric Engineering and Remote Sensing,  Vol. 17, No. 6, pp. 829-838.

Scarpace, F.L., R.W.  Kiefer, S. L. Wynn, B.K. Quirk, and G.A.  Frederichs, 1975.  "Quantitative
Photo-Interpretation for Wetland Mapping", Proceedings of the 41st Meeting of the American Society
of Photogrammetry, American  Society of Photogrammetry and Remote  Sensing, Falls Church, VA,

Schowengerdt, R.A., 1983.  Techniques  for Image Processing and Classification.   Academic Press,
Orlando, FL, 249p.

Sharp,  G.,  J.  Carter,  D.L. Roddick, and G.  Carmichael, 1981.   "The Utilization of Color Aerial
Photography and Ground Truthing To Assess Subtidal Kelp (Laminaria) Resources in Nova Scotia,
Canada",  Technical  Papers   1981   ASPRS-ACSM Annual  Convention,  American  Society  of
Fhotogrammetry and Remote Sensing,  Faiis Church, VA.

Shima L.J., R.R.  Anderson, and V.P. Cater,  1976.  "The Use of Aerial Photography in Mapping the
Vegetation of a Freshwater Marsh", Chesapeake Science, Vol. 17, No. 2, pp.74-85.

SPOT Image Corporation, 1897 Preston White Drive, Reston, VA 22091, (703)620-2200.

Steffenson, D.A., and E.E. McGregor,  1976.  "The Application of Aerial Photography to Estuarine
Ecology", Aquatic Botany, Vol. 2, pp. 3-11.

Steward,  W.R.,   V.  Carter,  and P.D.  Brooks, 1980.   "Inland  (Non-tidal)  Wetland  Mapping",
Photogrammetric Engineering and Remote Sensing, Vol. 46, No.  5, pp. 617-628.

Stohr, C.J.  and   T.R.  West,  1985.   "Terrain and Look Angle Effects Upon Multispectral Scanner
Response", Photogrammetric Engineering and Remote Sensing, Vol. 51, No, 2, pp. 229-235.

U.S. Army Corps of Engineers, 1988.  "Environmental Impact Statement for Operations, Maintenance
and Minor Improvements of the Federal Facilities at Sault Ste.  Marie, Michigan, Appendix F: Sediment
Aquatic Plant, and Bathymetry  Mapping  from Airborne Scanner Data", Detroit District Corps of
Engineers, Detroit, MI.

Wallsten,  M., and  P.O. Forsgren,  1989.   "The  Effects of  Increased Water  Level  on Aquatic
Macrophytes", Journal of Aquatic Plant Management, Vol. 27, pp.32-37.

Weismiller, R.A., 1979. "The Application of Remote Sensing Technology to the Solution of Problems
in the Management of Resources in Indiana", Semiannual Status Report, Purdue University.

Weismiller, R.A., S.J. Kristof, D.K. Scholz, and P.E. Anuta, 1977.  "Change Detection in Coastal Zone
Environments", Photogrammetric Engineering and Remote Sensing, Vol. 43, No. 12, pp. 1533-1539.

Wicker, K.M. and K.J. Meyer-Arendt, 1981. "Utilization of Remote Sensing in Wetland Management",
Proceedings, Pecora VII Symposium, Falls Church, VA. American Society of Photogrammetry, pp. 217-

Williamson, F.S.L., 1974.  "Investigations on Classification Categories for Wetlands of Chesapeake Bay
Using RS  Data",  Smithsonian Institution Annual Report, Oct.  1972-Oct. 1973, 98 p.

Wobber, F.J., 1974.   "Remote Sensing Trends in State Resources Management", Photogrammetric
Engineering and Remote Sensing, Vol. 40, No. 9.

Work, E.A. and D.S. Gilmer, 1976. "Utilization of Satellite Data For Inventorying Prairie Ponds and
Lakes", Photogrammetric Engineering and Remote Sensing, Vol. 42, No. 5, pp.685-694.


This glossary is included to assist the reader of this .report in understanding those relatively uncommon
terms that are widely used in remote sensing and space operations.  Many of these terms are defined
specifically as they are used in remote sensing, space operations, and related activities; these definitions
may different from standard definitions for the same word.


ABSORPTION.  The process by which electromagnetic radiation (EMR) is assimilated and converted
into other form(s) of energy, primarily heat.  Absorption takes place only on the EMR that enters a
medium, and not on EMR incident on the medium but reflected at its surface.  A substance that absorbs
EMR may also be a medium of refraction, diffraction,  or scattering; however,  these processes involve
no energy retention or transformation and are distinct from absorption.

ABSORPTION BAND.  A  range of  wavelengths (or frequencies) of EMR that is assimilated by a

AIRPHOTO IMAGES. A 9" by 9" photograph acquired vertically downward or obliquely from the air
on 10" roll film.  Airphoto images may be in the form of paper prints or transparent film.

AIRVIDEOGRAPHY.  The growing new field  of acquiring vertical airvideo images and making
measurements from digitized frames of such  imagery for  the purpose of  monitoring or managing
agricultural and natural resources, tax assessment,environmental degradation,  etc.

ALGORITHM (computing terminology). A statement of predefined steps to be followed in the solution
of a problem.

ARC/INFO.   A  vector  (arc)  based geographic Information System (GIS)  developed use  on
minicomputers and personal computers  (PC ARC/INFO).

AVHRR. Advanced Very High Resolution Radiometer imagery produced by NOAA Satellites.

BACKGROUND. Any effect in a sensor or other  apparatus or system above which the phenomenon of
interest must manifest itself before it can be observed.  See background noise.

BACKGROUND NOISE.  (1) In recording and reproducing, the total unwanted disturbance within a
useful frequency band, independent of whether or not a signal is present.  The signal is not to be included
as part of he disturbance.  (2) In receivers, the random oscillation in the absence of signal modulation
on ths carrier. Ambient oscillations detected, measured, or recorded with the signal become part of the
background noise.  Included in this definition is the interference resulting from primary power supplies,
that is commonly described as hum.

BAND, SPECTRAL.   An  interval in the electromagnetic spectrum defined by two wavelengths,
frequencies, or wave numbers.

 BRIGHTNESS..  The attribute of visual perception in accordance with which an area appears to emit
 more or less light.

 BRIGHTNESS VALUE (BV or DN digital number).  A number in a range of 0-63, 0-127, or  0-255
 that is related to the amount of radiance in Watts/cm striking a detector in the multi-spectral scanner.

 BYTE.  A unit of measure based upon a base 2 number system where each byte represents 256 data

 CELL.  In remote sensing, an area on the ground from which EMR is emitted or reflected.

 CLASSIFICATION.  (1)  An administrative system wherein information, equipment, or processes are
 categorized according to their importance.  (2) A systematic arrangement of objects (which have been
 imaged) into a logical structure or hierarchy.  (3) A computer process of determining the grouping of
 cells in  coregistered, multispectral,  multitemporal,  and/or multisensor rasters  to map the location of
 materials and identify their type (e.g., type of crop, surface cover map, map line type by color, etc,)

 CIR IMAGE.  A £olor InfraRed image or photograph renders healthy, green vegetation in bright red.
 Damaged,  diseased, or  dying vegetation will appear in shades of pink, tan, and yellow. The original
 sensing devise used for  collecting color infrared images may be a electronic scanner or special film type
 which maps the photographic infrared radiation just beyond the range of human vision into the red
 intensity in the display or film. The red radiation from the scene is mapped into green in the display or
 film. The green radiation  from the scene is mapped into the blue display or film.  The blue radiation
. from the scene is filtered out and not recorded.

 A physical or biological impact on growing plants which  begins to  cause a deterioration in their  vigor,
 that is, their water and/or  chlorophyll  content, will cause more rapid decreases in their reflectance of
 photo infrared radiation and corresponding increases in their red reflectance.

 CLUSTER.  A homogeneous group of units which are very "like" one another.  "Likeness" between
 units is usually determined by the association, similarity, or distance between the measurement patterns
 associated  with the units.

 COREGISTRATION.  This indicates that the associated raster and vector objects overlay each other
 accurately  in a geographic  sense.

 DETECTION.   The act  of discovering or perceiving  an  object on an  image.   It does  not  imply
 recognition or identification of that object.

 D.E.M. Digital Elevation Model data is distributed by the USGS in raster from on open reel magnetic
 tapes. It consists of two basic types..  The DMA type was originally created by the Defense Mapping
 Agency  in a fixed cell  size and  also a 3 arcsecond by 3 arcsecond cell size and distributed  in  1 by 1
 degree files. A newer format of elevation data is available for those 7.5' USGS quadrangles which have
 been processed into 1 by 1 arcsecond elevation cells.

 DIGITAL IMAGE (or digitized image, or digital picture function).  An image in digital format that is
 obtained by partitioning the  area of the image into a finite  two-dimensional array of small uniformly
 shaped mutually exclusive  regions, called resolution cells, and assigning a "representative" grey shade


 to each such spatial region.  A digital image may be abstractly thought of as a function whose domain
 is the finite two-dimensional set of resolution cells and whose range is the set of grey shades.

 DIGITIZATION,  MANUAL. The process of conversion of analogue or. graphic data into digital form
 by an operator with or without mechanical or computer aids.

 DIGITIZE. To use numeric values to represent data.

 DIGITIZER, GRAPHIC. Machine that changes graphic cartographic information into a digital format
 for computer input.

 ELECTROMAGNETIC SPECTRUM. (1) A system that classifies, according to wavelength, all energy
 that moves, harmonically, at the constant velocity of light. (2) A continuum that is conventionally broken
 into arbitrary segments (as ultraviolet, visible,  radio).

 FALSE COLOR.  Reproduction that shows objects in colors other than their true color.  Usually, the
 color refers to color infrared, but not necessarily.

 GAIN (electronic).  Ratio of output signal to input of a device.
 GEOMETRIC ACCURACY.  Four types:  Geographic (latitude-longitude) based on the standard Earth-
 fixed coordinates reference system, which  employs latitude and longitude.  Positional - the ability to
 locate a point in an  image with respect to a map. Scene Registration - the ability to superimpose the same
 point in two images of a scene taken at the same time (different spectral bands). Temporal Registration -
 the ability to superimpose  a point in two images of the same scene taken at different times (same or
 different spectral bands.

 GIGABYTE, GBYTE, OR GB.  A computer unit measurement used to indicate 1,000,000,000 bytes,
 1,000,000 kilobytes, 1,000 megabytes, or .001 terabytes.

 GIS  A geographic Information System is a computer system designed to allow users to collect, manage,
 and analyze large volumes  of spatially referenced and associated attribute data. The major components
 of a CIS are:  a user interface, system/data base management capabilities; data base creation/data  entry
 capacity; spatial data manipulation and analysis packages; and display/product generation functions. From
 USGS Open File Report 88-105: A Process for Evaluating Geographic Information Systems.

 GPS Global Positioning System technology was developed by the US Department of Defense for military
 applications. The  technology makes positioning on Earth possible to within millimeters of accuracy.
 GPS is based on a constellation of NAVSTAR satellites orbiting earth (by 1992 a total of 24 satellites will
 have been launched in 12-hour orbits, so at least four satellites - the minimum required to obtain 3D
 positional data - wi!! be visible to a GFS at almost any point  on Earth).  The GPS receiver, a  small
 portable device that weighs only a few pounds, acquires the signals from these four satellites and  using
 a method called "satellite ranging", calculates  the position on Earth by measuring the length of time it
 takes for the satellite signal  to reach the receiver.  Within one second,  positional coordinate data  -
 longitude, latitude,  and elevation - is displayed on the GPS receiver screen.

 GRAY TONE.  A number or value assigned tn  a position (\,  y) cr.  ST.  image.  The  number is
 proportional to  the integrated  output, reflectance, or transmittance of a small area, usually called a


resolution cell or pixel, centered on the position (x, y).  The gray shade can be measured as or expressed
in any one of the following ways:
        (1) transmittance   (2) reflectance
        (3) a coordinate of the ICI color coordinate system
        (4) a coordinate of the tristtmulus value color  coordinate system
        (5) brightness     (6)  radiance
        (7) luminance      (8) density
       (9) voltage      (10) current

GREEN BIOMASS.  Green biomass is synonymous with phytomass.  It is the amount of wet or dry
weight,growing, chlorophyll containing plant material  per unit ground area.  It is usually expressed in
grams per square meter, tons per' acre, or metric tons per hectare.

GREENNESS.  The biophysical property  of the surface of the earth  indicating its greenness in a
biological sense as related to vigor, water wellness, and chlorophyll content. This greenness property
is a qualitative estimate of green biomass and is the property computed and stored in a raster object by
Kauth's greenness, brightness, wetness transformation  for LANDS AT MSS or TM images.

GROUND RESOLUTION.  The minimum distance between two or more adjacent features or the
minimum size of the feature which can be detected; usually measured in conventional distance units, e.g.
feet or inches.

GROUND TRUTH. Information concerning the actual state of the environment at the time of a remote
sensing overflight.

GROUND CONTROL. Accurate data on the horizontal and (or) vertical positions of identifiable ground

HIS  Refers to the Hue, Intensity, Saturation color domain where these three characteristic are used to
specify a color.

HISTOGRAM.  A graphical representation of a frequency distribution by means of lines or rectangles
that represent class intervals along the x-axis, and represent corresponding class frequencies along the y-
axis.  A graphical representation of the number of times a value occurs in a raster image plotted against
each of the data values which could be present in the original raster image.

IDENTIFY.  The process of recognition or identification whereby an object or unit on an image may be
classified or categorized as to a specific function or type.

IMAGE.   The  recorded  representation of an  object produced by optical,  electro optical, optical
mechanical, or electronic means.  It is the term generally used when the EMR emitted or reflected from
a scene is not directly recorded on film.

IMAGE ENHANCEMENT.   Any of several processes to improve the interpretability of an image.
These include contrast enhancement, ratioing, spatial filtering, and so on.

IMAGE PROCESSING. All the various operations which can be applied to photographic or image data.
These include, but are  not  limited  to, image compression, image restoration, image enhancement,
preprocessing, quantization, spatial filtering, and other image pattern recognition techniques.

IMAGERY. The visual representation of energy recorded by remote sensing instruments.

INFRARED (IR).  Pertaining to or designating the portion of EM spectrum with wavelengths from the
red end of the visible spectrum to the microwave portion of the spectrum, or from 0.7 micrometer to 1

INTENSITY. One of the three coordinates which specifies color in the HIS color domain.  Intensity is
that coordinate or value which represents the brightness or average radiance level of a color.

INTERACTIVE IMAGE PROCESSING. The use of an operator or analyst at .a console with a means
of assessing, preprocessing, feature extracting, classifying, identifying and displaying the original imagery
or the processed imagery for his subjective evaluation and further interactions.

INTERPRETER.  An individual trained in the process of detecting, identifying, analyzing, locating and
quantifying information portrayed on imagery and determining its significance.

KILOBYTE.  In computer terminology, refers to 1024 bytes of core memory storage.

LAND COVER.  Cultural objects and natural and cultivated vegetation occupying the landscape that can
be grouped or classified  and subsequently mapped using remotely sensed  imagery.

LAI Leaf Area Index is a unitless, biological, measure which is the one sided surface area of the leaves
of a crop, forest, grassland, etc. per unit ground area.

LANDSAT. A satellite vehicle used to house the sensor systems which collect multispectral images using
at various times a RBV or Return Beam Vidicon device, the MSS or MultiSpectral Scanning device, and
the TM or Thematic Mapper scanning device.  LANDSAT also relays data from ground observation
stations. Originally the LANDSAT satellite was referred to as the ERTS or Earth Resource Technology

LEGEND.  A description,  explanation, table of symbols, scale bar and other information  printed  on
margins of maps or mosaics.

MAP, PLANIMETRIC. Map showing only the horizontal location of detail.

MAP, THEMATIC. Map designed to demonstrate particular features of concepts. In conventional us
 MULTIBAND SYSTEM. A system for simultaneously recording EMR from the same scene in several
 bands from  essentially the same spectral region, such  as the visible or visible and near-IR.   May  be
-applied  to cameras with different film/filter combinations or scanning radiometers that use dispersam
 optics to split wavelength bands apart for viewing by several filtered detectors.

 MULTISENSOR IMAGES.  Images collected by different sensors or other devices which have been
 manipulated  so as to be coregistered in both cell size and location are called multisensor data.

 MULTITEMPORAL IMAGES. Images collected at different times by the same device and brought into
 coregistration are multitemporal  images. For example, airvideo images collected at approximately the
 same data on two successive years can be digitized and one of them warped to overlay the other and
 stored in the same project file.  These multitemporal images can then be analyzed in various ways to map
 the changes between the dates.

 MULTISPECTRAL. Generally used for remote sensing in two or more spectral bands, such as visible
 and IR.

 NANOMETER.  A unit of measure equal to one millimicrometre  (millimicron) or one-millionth of a

 NASA  National Aeronautics and Space Administration.
 NEAR INFRARED. The preferred term for the shorter wavelengths in the infrared region extending
 from about 0.7 micrometers (visible red) to about three micrometers. The longer wavelength end grades
 into the middle infrared. The term really emphasizes the radiation reflected from plant materials, which
 peaks around 0.85 micrometers.  It is also called solar infrared, as it is only available for use during the
 daylight hours.

 NOAA  National Qceanographic and Atmospheric Administration.

 NWI National Wetlands Inventory being conducted by the  USF&WS.

 PICTURE ELEMENT  (PIXEL).  A unit whose first member is a resolution cell and whose second
 member is the gray shade assigned by a digital count to that resolution cell.

 PIXEL. Abbreviation of picture element.

 PLANIMETRY, AUTOMATIC.  The process of calculating the area of a patch from data defining the
 boundary of the patch.

 PLATFORM.  The object, structure, vehicle, or base upon which a remote sensor is mounted.

 POLYGON.  Plan figure consisting of three or more vertices  (points) connected  by line segments or
 sides. The plane region bounded by the sides of the polygon is the interior of the polygon.

 PREPROCESSING.  An operation applied before pattern identification is performed. Preprocessing
 produces, for the categories of interest, pattern features which tend to be  invariant under  changes such
 as translation, rotation,  scale, illumination levels, and noise.   In essence, preprocessing converts  the


measurements patterns ro a form which allows a simplification in the decision rule.  Preprocessing can
bring into registration, bring into congruence, remove noise, enhance images, segment target patterns,
detect, center, and normalize targets of interest.

RANGE, BRIGHTNESS.  Variation in light intensity from maximum to minimum.  This generally
refers to a subject to be photographed.

RANGE, DYNAMIC. The difference between maximum measurable signal and minimum detectable
signal.  The upper limit usually is set by saturation and the lower limit by noise.

RASTER CELL. A raster is made up of a sequence of numbers representing a measured variable or
the results of computing upon such measurements.  These numbers have an order and a position in space
and describe the variation if some phenomena of  interest such as the elevation above sea level of the
ground or the  intensity of the red radiation in a video image.  The individual value in a raster can thus
be  thought of as the average  value representing  a specific area or  cell of  the ground or other area
represented.  For convenience, this cell is usually thought of as a square or  rectangle  although many
image collection  devices actually create  the value from  the measurement of a circular or elliptical
observation cell.

RATIO.  A reflectance ratio of any target is the percent of reflectance in one  spectral region divided by
the percent of reflectance in another region.  The  disolay is a representation  of the reflectance ratio of
the two spectral regions.
RECTIFYING.  A process by which the geometry of an image area is made planimetric.  For example,
if the image is taken of an equally spaced rectangular grid pattern, then the rectified image will be an
image of an equally spaced rectangular grid pattern.  Rectification does not remove relief distortion.

REFLECTANCE.  A measure of the ability of a body to reflect light or sound. The reflectance of a
surface  depends on the type of surface,  the wavelength  of the lamination, and the illumination and
viewing angles.

REGISTERING.  The translation-rotation alignment process by which two  images of like geometries
and of the same set of objects are positioned coincident with respect to one another so that corresponding
elements of the same ground area appear in the same place on the registered images.

REFLECTION (EMR theory)  EMR neither absorbed nor transmitted is reflected.  Reflection may be
diffuse,  when the incident radiation is scattered upon being reflected from the surface, or specular, when
all or most of the EMR is reflected at an angle equal to the angle of incidence.

REMOTE SENSING.  In the broadest sense, the measurement or acquisition of information of seme
property cf an object or phenomenon, by a recording device that is  not in physical or intimate contact
with the object or phenomenon under study; for instance,  the utilization at a distance (as from aircraft,
spacecraft, or ship) of any  instrument and  its attendant  recording  and display devices  for gathering
information pertinent to the environment, such as measurements offeree fields, electromagnetic radiation,
or acoustic energy.  The technique employs such devices as the camera, lasers, and radio frequency
receivers, radar systems, sonar, seismographs, gravimeters, magnetometers, multispectral scanners, and
scintillation counters.

RESAMPLE.    The process of interpolating the values of the cells in a raster object to yield a new
raster object which has larger or smaller cells and perhaps a raster which has been warped, reoriented
and/or rescaled relative to the input raster.

REPRESENTATIVE FRACTION (RF).   The  relation between niap or photo distance and ground
distance, expressed as a fraction (1/25,000) or ratio 1:25,000, or  1 mm on map = 25,000 mm on the

RESOLUTION.  A generic term which describes how well a system, process, component, material, or
image  can reproduce an isolated object or  separate closely  spaced objects or lines.   The limiting
resolution, resolution limit or spatial resolution is described in terms of the smallest dimension of the
target or object that  can just be discriminated or observed.  Resolution may be  a function of object
contrast, spatial position as well as shape.  The ability of a remote sensing system to distinguish signals
that are close to each other spatially, temporally, or spectrally.

RESOLUTION CELL.  The smallest most elementary areal constituent of grey shades considered  by
an investigator in an image. A resolution cell is referenced by its spatial coordinates. The resolution cell
or formations of resolution cells can sometimes constitute the basic unit for pattern recognition of image
format data.

RGB An acronym for Red, fireen, and Blue.

SAMPLES. Samples of cells in a raster or image are selected from known areas from ground visitation,
detailed airphoto interpretation, or other personal experience and are used to represent a feature or land
cover of interest and  value in a process such as feature mapping.

SATURATION.  One of three coordinates which specifies color in the HIS domain. Saturation is that
coordinate or value which designates  how far away a color  is from gray or neutral color of equal

SCANNERS.   The sweep of a mirror, prism, antenna,  or other element across the track (direction of
flight); may be straight, circular, or other shape.

SENSOR. An device which gathers energy and presents it in a form suitable for obtaining information
about the environment. Passive sensors, such as thermal infrared and microwave, utilize EMR produced
by the surface or  object being sensed.  Active sensors, such as radar, supply their own energy source.
Aerial cameras use natural or artificially produced EMR external to the object or surface being sensed.

SIGNATURE. Any characteristic or series of characteristics by which a material may be recognized.
Used in the sense of spectral signature, as in photographic color reflectance. A category is said to have
a signature only if the characteristic pattern is highly representative of all units of that category.

SPECTRAL BAND.   An  interval in the  electromagnetic spectrum  defined by two wavelengths,
frequencies, or wave numbers.

SPECTRAL INTERVAL.  The width, expressed  either in wavelength or frequency, of a particular
ponion of the electromagnetic spectrum. A given sensor, such as radiometer detector or camera film may
be designed to measure or be sensitive to energy from a particular spectral interval.


SUN ANGLE.  The angle of the Sun above the horizon.  Both the quantity (lumes) and spectral quality
of light being reflected to the aerial camera or sensor are influenced by Sun angle.  Also called Sun
elevation, Sun elevation angle.

THERMAL BAND.  A general term for intermediate and long wavelength infrared-emitted radiation.
as contrasted to short wavelength reflected (solar) infrared radiation.  In practice, generally refers to
infrared radiation emitted in the 3- to 5- and 8- to 14-micrometer atmosphere windows.

THERMAL INFRARED.  The preferred term for  the middle wavelength ranges of the  IR region
extending roughly form 3 micrometers  at the end of the near infrared, to about 15 or 20 micrometers
where the far infrared commences.  In practice the limits represent the envelope of energy emitted by the
Earth behaving  as a greybody with a surface temperature around 290 K (27 C).  Seen  from any
appreciable distance, the radiance envelope has several brighter bands corresponding to windows in the
atmospheric absorption bands.  The thermal band most used in  remote sensing extends from 8 to 14

UNIT. The simplest and most practical picture element(s) observed, compared, or measured in a pattern
recognition sequence.  Most units are not simple picture elements but are often complex spatial formations
of picture elements such as houses, roads,  forest, etc.

VECTOR.  A linear line segment, normally short, used to construct any line form on a plotter, drafting
unit or display.

WAVELENGTH.  Wavelength = I/frequency. In general, the mean distance between maximums (or
minimums) of roughly periodic pattern. Specifically,  the shortest distance between particles  moving in
the  same  phase  of oscillation in a wave disturbance. Optical and IR wavelengths  are measured in
nanometers, micrometers, and Angstroms.



   (formerly the National Cartographic Information Center)
Reston - ESIC
U.S. Geological  Survey
507 National Center
Reston, VA  22092
(703) 860-6336/FTS 959-6045
Rolla - ESIC
1400 Independence Road
Rolla, MO   65401
(314) 341-0851/FTS 759-0851

Lakewood -  ESIC
Federal Center
Box 25046,  MS 504
Denver, CO  80225-0046
(303) 236-5829/FTS 776-5829

Los Angeles - ESIC
Federal Bldg., Rm.7638
300 N. Los  Angeles St.
Los Angeles, CA  90012
(213) 894-2850/FTS 798-2850

Stenhis Space Center -  ESIC
Building 3101
Stennis Space Center, MS  39529
(601) 688-3544/FTS 494-3544
 Salt Lake  City  - ESIC
 8105 Federal  Bldg.
 1245 S.  State St.
 Salt Lake  City, UT   84138
(801) 524-5652/FTS 588-5652

 Menlo Park -  ESIC
 Building 3, MS  532
 345  Middlefield Road
 Menlo Park, CA  94025
(415) 329-4309/FTS 459-4309

 Anchorage  - ESIC
 4230 University Drive
 Anchorage,  AK  99508-4664
(907) 271-4159/FTS 868-7011
 San Francisco - ESIC
 504 Custom House
 555 Battery St.
 San Francisco,  CA  94111
(415) 705-1010/FTS 465-1010

 Washington,  D.C.  - ESIC
Dept. of the Interior Bldg.
 1849 C St.,  NW Rm.2650
 Washington,  D.'C.   20240
 (202)  208-4047/FTS 268-4047
For more detailed information contact the ESIC office nearest you.
Fact  sheets,  user  guides,  price  lists,  and  order  forms are
available upon request.   ESIC personnel will  assist  with ordering
available   digital   products  and  can  provide  more  technical
information.   ESIC has  established affiliated offices with many
State governments.

Available USGS Circulars include:
895-A, Overview  and USGS Activities
895-B, Digital Elevation Models
895-C, Digital Line Graphs from 1:24,000 Maps
895-D, Digital Line Graphs from 1:2,000,000 Maps             '
895-E, Land Use  and Land Cover  Digital Data
895-F, Geographic Names  Information System
895-G, Digital Line Graph Attribute Coding Standards



Data Format;
Vector files in DLG format
1:24,000, 1:100,000, 1:2,000,000
Data Description;
Separate files are available for:
  Land net
Data Coverage:
1:2,000,000  data are  available  for the  entire United  States,
1:100,000 data are complete for transportation and hydrology.
1:24,000 are not as well developed.
For current coverage call ESIC.
9-track magnetic tape and/or paper maps

Data Format:       '
Vector files in GIRAS format
(GIRAS is an arc-node format.  Programs are available from ESIC for
converting this format to the DLG standard).
Raster available in binary or character  (ASCII or EBCIDIC) form
Vector  1:100,000 or 1:250,000
Raster  200 meters
Data Description:
Separate files are available for:
  Land use/land cover
  Political units
  Census county subdivisions
  Hydrologic units (watersheds)
  Federal land ownership  (park, forest,  etc.)
Composite Theme .Grid (CTG) which contains all themes  available for
a given area.
Data Coverage;
Contact ESIC  for latest  edition  of "Index to Land  Use and Land
Cover Information".
9-track ASCII tapes and paper maps


                     DIGITAL ELEVATION MODELS
Data Format:
Data Description:
A regular  array of elevation values  referenced  to  the Universal
Transverse Mercator  (UTM)  coordinate  system with a  spacing of 30
meters.  Data are collected either by  digitizing  7.5 foot contours
overlays or by  scanning photographs.
Data Coverage;
Scattered  7.5 minute  quadrangles throughout the United States.
Contact  ESIC  for  the  "Index to  Digital Line Graph  and Digital
Elevation Model Data".
9-track magnetic tape  and/or paper maps

Data Format:
1 degree by 1 degree blocks
Data Description:
Data   consists  of   a  regular   array  of   elevation  values,
latitude/longitude,  referenced with a  spacing  of  3-arc seconds.
Data, were produced from  digitizing 1:250,000 topographic maps.
Data Coverage;
Entire United States and many  other parts of the world.
9-track magnetic tape



Product (Data Type)

A. Digital Elevation Model
B. Digital Line Graph  (DLG)
   Land Net
                    (DLG) (30' x 30' blocks)
Digital Line Graph
   Land Net
Land Use/Land Cover  (Polygon)
Census Tracks
Political Boundaries
Hydrologic Units
Federal Land Ownership
Geographic Names by State

Digital Elevation Model
(1 deg x 1 deg blocks)

Digital Line Graph (DLG)
 (sold in 21 sections)
Boundary Layer (per section)
Transportation Layer  (per section)
Hydrography Layer (per section)
Prices for orders of five or less are as follows:
Number of files per order
                    Total Price
Prices for orders of six or more files are as  follows:
Base Charge of $90 plus $7 per file.

   (formerly the National Cartographic Information Center)
U.S. Department of the Interior
Fish and Wildlife Service
National Wetlands Inventory
9720 Executive Center Drive
Suite 101, Monroe Building
St. Petersburg, FL  33702

Data Formats
Generally available in USGS 7.5 minute topographic maps.
Limited digital data available.

Data Description:
Photo-interpreted   wetland   inventory   on  USGS   base   maps.
Classification  of  categories  is  based on  US Fish  and  Wildlife
Service  publication  "Classification  of  Wetlands and  Deepwater
Habitats of the  of  the United States",  USFWS Report No.  FWS/OBS-

Data Availabilitv/Cost
Data may be ordered through any USGS ESIC office.
Paper Composite of NWI 7.5 minute maps are $1.75 each.


 US Department of Commerce
 Bureau of the Census
 Customer Services Branch
 Data User Services Division
 (301) 763-4100

 Da*ta Format;
 ANCII or EBCDID, labeled or unlabeled

 Data Description;
 TIGER   -  Topologically   Integrated  Geographic   Encoding  and
 Referencing System.  Census Bureau's new digital map data base that
 provides coordinate based  cartographic information by County.
 TIGER data base includes:
  Features  (roads, railroads, rivers), feature names and
  classification codes, alternate feature names and codes,
. feature shape coordinates.
  Address ranges with ZIP  codes for metropolitan areas.
  Census statistical area  boundaries  (census blocks/tracks)
  Political boundaries  (state, county, and incorporated places)

 The  TIGER/Line  files  contain  basic  data  for each   individual
 segment.   Each  segment record  contains the  appropriate  census
 geographic area codes, latitude/longitude coordinates, the name and
 type  of the  feature,  the relevant  census  feature  class code
 identifying the feature segment by category, and for metropolitan
 areas, the address ranges and associated ZIP codes for each side of
 a street segment.

 Data Coverage;
 Entire  United  States,  Puerto  Rico,   the  Virgin Islands,  and US
 Pacific territories.

Precensus TIGER/LINE Files     April 1989
Postcensus TIGER/LINE Files     Late Spring/Summer 1991
TIGER/Data Base                 Fall 1991
TIGER/Boundary                  Fall 1991
TIGER/Comparability             Fall 1991
TIGER/Area                      As Required

 9-track magnetic tape

Acquisition Costs;
 $190 for the first county  ordered
 $ 15 for each additional county ordered within the sane  state


 EOSAT  Company
 4300 Forbes  Blvd.
 Lanham, MD   20706
Data  Format:
U Data  -  Corrected for  radiometric only
C Data  -  Corrected for  radiometric and geometric distortions

Radiometric:  Data are  scaled to pixel dynamic ranges of  0-256 for
TM and  0-127 for MSS, and compensated for detector  gain and offset

Geometric:   Data  are compensated  for earth rotation, spacecraft
altitude,  attitude  and sensor  variations.  (Note:  data  is  not
rectified to a coordinate system, geocoded products  are available) .

LANDSATS  4 and  5

  Thematic Mapper  (TM)   C or U available  4/84 to present
  Band  sequential  (BSQ)  - Fast Format only

  Multispectral Scanner (MSS)
  BSQ or Band interleaved (BIL)  - C or U  available  6/81 to present
  BSQ or  BIL -  C only for 1/79 to 5/81
  BIP prior to  1/79  (U only)

9-track magnetic tape,  6250 or 1600 bpi

TM approximately 30 meters with thermal band 6 at  120 meters
MSS approximately  80 meters

Data Description:
As a  result  of the Land Remote  Sensing  Commercialization Act of
1984,   Landsat  data   are  currently  acquired,   processed,  and
distributed by  EOSAT Company under  a cooperative  agreement with
NOAA and  the  USGS.   Users  must sign a  form  when  acquiring data
stating  they  will  not  copy  or  distribute the  data  without
authorization from EOSAT.

Spectral Sensitivity of the TM and MSS  Sensors
TM data - 7 bands
  Band 1 - blue
  Band 2 - green
  Band 3
  Band 4
         - red
         - near IR
  Band 5 - near IR
  Band 7 - mid ±R
  Band 6 - thermal IR   10.4-12.5
MSS data - 4 bands
 Landsat 4 & 5
    Band 1
    Band 2
    Band 3
    Band 4
                    Landsat 1, 2,& 3
                        Band 4
                        Band 5
                        Band 6
                        Band 7
                    green    0.5-0.6 micrometers
                     red      0.6-0.7
                     Near IR  0.7-0.8
                     Near IR  0.8-1.1
Data Coverage:
Historical data can be obtained through USGS-EROS Data Center.
Acquisition Costs;
                  (size: 185 x 170 km)
                             x 85 km)
TM Full Scene
TM Quarter Scene  (size: 92.5
TM Moveable
TM moveable
TM Geocoded
TM Geocoded
TM Geocoded
(size:  100 x 100 km)      $
(size:   50 x 100 km)      $
(size:  185 x 170 km)      $
(size:  92.5 x 85 km)      $
(size:  1/2 by 1 degree map
MSS Full Scene    (size: 185 x 185 km)



 U.S.  Environmental Protection Agency
 Environmental Monitoring Systems Laboratory
 Advanced  Monitoring Systems  Division
 P.O.  Box  93478
 Las Vegas,  NV  89193-3478
 Ross  S. Lunetta
 (702)  798-2175
 FAX  (702)  798-2637

 Data  Format:
 Aircraft-mounted multispectral scanner.
 EMSL-LV's scanner  system is  a Daedalus  Enterprises  Model  1260
 instrument.   This system uses a rotating mirror to direct  radiated
 energy from a spot on the Earth's surface onto sensing  detectors.
 Energy is focused  on sensors with a telescope assembly, and  is
 split into spectral components by a  prism and a  dichroic mirror.
 The  Daedalus 1260  is  an 11-band system with a sensitivity  range
 from  0.3  to 14 micrometers (ultraviolet though thermal  infrared).
 Geometric control  is  greatly improved over past missions  (FY91)
 with  the  use of Global Positioning System (GPS) technology.   With
 a  single  receiver  (autonomous  mode)  RMS accuracies of fixes are
 approximately 25 meters unless the Air  Force  invokes intentional
.system degradations called "selective availability" (SA),  in which
 case  accuracies degrade to approximately 100 meters.  However,  if
 a  second  GPS receiver  simultaneously acquires  data at  a  known
 location,  then  "differential  corrections"  can  be  applied  that
 improve accuracies  to approximately  5  meters, whether SA  is  in
 effect or not.

 GPS fixes are acquired by an onboard  GPS receiver and logged on a
 small computer.  The  fixes  include accurate latitude,  longitude,
 and  altitude information  which is made more accurate  later when
 differential corrections  are  applied.   The  fixes  also  include
 precise information about the time that the fixes were taken.  This
 is critical because it takes time for the receiver to  report its
 fix  to the  computer,  and the  aircraft  has moved an appreciable
 distance  in that time.  The  recorded times  are used later  during
 data  reduction.  The sequence of precise times and positions from
 the GPS fixes, and the scan line time  are used to interpolate scan-
 center position and heading.   This allows accurate compensation for
 changes in heading,  mispositioning left  or right of the  scan line.
 and changes in altitude.  In  addition  to  knowing aircraft position,
 heading and altitude, scan line by scan line,  it is important  to
 understand the terrain below.   Height above terrain is critical  to
 scanner geometry, as is variation in  terrain elevation  across the
 scan  line.  Digital elevation models (DEM) from the USGS are placed
 into  mosaics  covering  the  mission  area,  and  are used by the
 ereometr i c correction software.


 US Environmental Protection Agency
 Environmental Monitoring Systems Laboratory
 Advanced Monitoring Systems Division
9-track magnetic tape.

Data Description;
The  MSS  is  aircraft-mounted  and  available  for  contract data
collection.   The customer has control over the area of coverage,
resolution  of  data,  and  wavelengths  to be  collected.    Post-
processing  operations applied include those designed to calibrate
the data  and to correct for systematic  scanner distortions.    In
addition, simultaneous aerial photography can be  collected  with a
Wild RC-8 metric mapping camera.

Spectral Sensitivity of the Daedalus DS-1260
Channel 1
Channel 2
Channel 3
Channel 4
Channel 5
Channel 6
Channel 7
Channel 8
Channel 9
Channel 10
Near Ultraviolet
Near Infrared
Near Infrared
Mid Infrared
        0.38-0.42  micrometers
Either one of two thermal detectors can be employed
Channel 11
Thermal Infrared
(InSb)   3.00-5.00
(MCT)    8.00-14.0
Acquisition Costs;
MSS collection only   $15 - $20 per square mile
Simultaneous aerial photography    $10 -  $15 per square mile


National Environmental  Satellite,  Data,  and Information Service
National Geophysical  Data Center
NOAA, Code E/GC1
325 Broadway
Boulder, CO   80303-3328
(303) 497-6900

Data Format:

Digital data  is distributed on standard magnetic tape: 9-track,
ASCII, 1600bpi.   Other formats such as 6250 bpi tapes  and floppy
diskettes  may be requested.

Data Description;

Aeromagnetic  and  geomagnetic data; oil and gas lease data; ocean
core sample  locations;  ocean  bottom  characteristics;  dangers to
navigation; coastal,  deepwater,  or gridded bathymetry; land and
marine seismic  data;  geothermal  data;  land  and marine geology;
satellite  data  (GEOSAT, MAGSAT, GEOS-III,  SEASAT); and  topography

Acquisition Costs;

Varies, call  or write for quotes,  literature.

Soil Conservation Service  (SCS)
National Cartographic Center
South National Technical Center
501 Felix, P.O. Box 6657
Fort Worth, TX  76115

Data Format;
Raster encoded as ASCII

Vector: mostly 1:24,000
Raster: mostly 4-hectare cells

Data Coverage:
Scattered.  Call/write SCS at above address  for current edition of
"Status of Detailed Soil  Survey Digitizing" or  "Status  of Soil
Survey Digitizing".

Data Description;
Data are available  on a county basis.   The program is funded by
state SCS  offices;  thus,  extensive data are available  for some
states,  and  none  for  others.    For  areas that  have not been
digitized,  cost-sharing  information  may  be   available  form
individual state soil scientists.

9-track magnetic tape

Acquisition Costs:
$150 per tape, $25 additional charge if SCS provides tape.

 SPOT Image Corporation
 1897 Preston White Drive
 Reston,  VA  22091-4326
 (703)  648-1813

 Data Format:
 10 jneter panchromatic or 20 meter multispectral
 Data Coverage:
 Worldwide,  though  historical data are not yet extensive.
 Data Description;
 The  standard product is a SPOT scene measuring, 60  km  x  60 km when
 acquired from a vertical (nadir)  viewing angle.  SPOT'S sensors
 also have  an  off-nadir viewing  capability.   Panchromatic  and
 multispectral scenes  are available and  are  processed  to one  of
 three  levels:
 Level  1A - raw data which  have  undergone radiometric corrections
 only to  normalize  detector  response within each spectral band.   No
 geometric corrections have been applied to the data.
 Level  IB -  raw data  which have  undergone  both radiometric  and
 geometric corrections to reduce distortions from systematic orbital
 effects.   Level IB images have a locational  accuracy  of about  860
 meters and  are not considered planimetric.
 Level  2  -  precision  processed data  which  include  radiometric>
 geometric,  and bi-directional corrections based  on ground control
 points  (GCP).   Imagery is oriented  true north  and  available  in
 several  projections.   Level 2 data  have a location accuracy  of
 approximately 30  meters, provided  the image was recorded  at a
 vertical viewing angle and relief is not over 10 meters.

 Spectral Sensitivity of the SPOT Sensor

 Multispectral  20  meter resolution
 Band 1    green           0.50-0.59 micrometers
 Band 2    red             0.61-0.68
 Band 3    Near Infrared   0.79-0.89
 Panchromatic  10 meter resolution
 Band 1    visible         0.51-0.73 micrometers

 9-track  magnetic tape,  1600 or 6250 bpi,  EIL format

Accrui'sitions Costs:
	Panchromatic	Multispectral
Level  1A, IB                      $1900                $1700
Level 2                             3500                 3200
 SPOT QuadMaps                                           $500
   1:24,OOO  corresponding to USGS 7.5 siinute t.opo.
   Meet  QuadHaps are less than 2 years old,


This listing was extracted from the USGS Earth Science Information
Center .(ESIC),  listing entitled "Sources for Digital Spatial Data".

National Cartographic Information Center,  Geological Survey
4230 University Drive, Anchorage, AL  99508   (907)271-4159
Several federal lands.inventoried.

Satellite imagery over Alabama
Geological Survey of Alabama
Box 0, 420 Hackberry, Tuscloosa,  AL  35486  (205)349-2852
Complete  coverage by  Landsat collected  between 1972-74,  color
composites available, non-automated system.

Arizona Land Resources Information System
State Land Office, 1624 W. Adams, Room 302
Phoenix, AZ  85007   (602)225-4061
General vegetation cover, 50 Meter cells,  1986
Vegetation/land cover, soils 1981

California Department of Water Resources,  Planning Division
1416 9th Street, Room 252-3, Sacramento, CA  9.4236-0001
Land use, agriculture, hydrology
Bay Area Spatial Information Data Base - GEOGROUP
2560 9th Street, Suite 319, Berkeley, CA  94710  (415)549-7030
Hydrography, wetlands, land use,  eco-zones

Florida Agriculture Lands Mapping
Florida Department of Transportation
State Topographic Office, Tallahassee, FL  32301  (904)488-2168
Land .cover (15 classes) from Landsat TM imagery, concentration on
agriculture, multi-temporal analyses  also available.   Complete
library of Landsat and  related image data (1976,1984-85).  Users
group located at DOT.

Coastal Resource Data Base
Hawaii State Dept. of Planning and Economic Development.
250 S. King Street, Honolulu, HI  96813   (808)548-4025

 Land Use - State of Idaho
 Soil Conservation Service,  Soils  Division/Carto-GIS Laboratory
 304  N.  8th Street,  Room 345,  Boise,  ID   83702   (208)334-1525
 Land use,  agriculture,  forestry,  water;  1:100,000 vector

 Illinois Geographic Information System
 Illinois State  Geological Survey,  Computer Research and Services
 615  E.  Peabody,  Champaign,  IL 61820  (217)344-1481
 Land resources,  ecoregions,  surface  hydrology

 Satellite imagery over  Iowa
 Iowa Department of Natural  Resources, Geological Survey Bureau
 123  N.  Capitol  Street,  Iowa City,  IA 52242
 Landsat coverage of  the state for  MSS 1973-81 and  TM 1982-84.
 Digitized data  sets in  EROS or Goddard  format.

 Satellite Imagery over  Kansas
 Kansas  Geological Survey, The University of Kansas
 1930 Constant Avenue, Lawrence, KS  66046
 Positive and negative transparencies of  over 2000 images covering
 all  of  Kansas.   Non-automated system.

 Kentucky Natural Resource Information System
 Kentucky Natural Resources  and Environmental Protection Cabinet
 Capitol Plaza Tower,  14th  Floor,  Frankfort,  KY  .40601  (505)564-
 5174.   Landforms,  surface hydrology.

 Louisiana  Areal  Resource Information System
 Louisiana  Office of State Planning
 P.O.  Box 94095,  Baton Rouge,  LA   70804   (504)342-7410
 Land use 1979 Satellite imagery over Louisiana
 Louisiana  Department of Natural Resources
 Louisiana  Geological Survey
 Box  G,  University Station,  Baton  Rouge,  LA  70893  (504)388-5320
 Catalog/Index of available  imagery,  including the type, quality,
 data, and  scale  over the entire state of Louisiana.

Maryland Automated  Geographic Information System
Maryland Department of  State  Planning,  Office of Planning Data
 301  Preston Street,  Baltimore, MD 21201-2365   (301)225-4450
 Vegetation, wetlands

Michigan Resource Inventory Program
Michigan Department of Natural Resources, MIRIS
P.O. Box 30028, Lansing, MI  48909  (517)373-1170
Landcover  information  includes:  agriculture,  forest,  non-forest,
water, wetlands, barren lands.  1:24,000 scale, vector format.

Minnesota Land Management Information System
Minnesota State Planning Agency, Land Management Information Center
Metro Square Building, Room LL-65, St. Paul, MN  55101  (612)295-
1208.  Land use, soils, forestry, agriculture, watersheds, 1 ha.

Mississippi Automated Resource Information System
Mississippi Research and Development Center, CIS Division
3825 Ridgewood Drive,  Drawer  2470,  Jackson,  MS  39211  (601)982-
Land use, soils, flood-prone areas, agriculture, 25 hectare cells

Nebraska Natural Resource Information System
Nebraska Natural Resources Commission, Data Bank Section
301 Centennial Mall South, Box 94876,  Lincoln, NE 68509
(402)471-2081.  Land use; agriculture, soils

North Carolina Land Resource Information System
North Carolina Dept. of Natural Resources and Community Development
P.p. BOX 27687, Raleigh, NC  27611-7687  (919)733-2090
Soils, hydrography, and nursery areas

New Mexico Natural Resource Information System
New. Mexico Department  of Natural Resources
408 Galisteo Street, Villagra Bldg., Santa Fe, NM  87503
(505)827-7830.    Vegetation,  Agriculture,  hydrography,  soils,

New Mexico MOSS Data
Bureau of Land Management, New Mexico State Office
Box 1449, Federal Bldg., Santa Fe, NM  87501   (505)988-6081
Partial  state coverage  of  vegetation,   soils,  range management,

 Soil Landform Analysis
 Bureau of Land Management,  Nevada State Office
 300 Booth St., Box 12000, Federal Bldg./Reno,  NV  89520  (702)784-
 5731.   Soils, surface water.
 Clark County Resources Information System Data Base
 Clark County Department of Comprehensive Planning
 225 Bridger Ave.,  7th Floor, Las Vegas, NV  89115
 Land use, vegetation, hydrography, soils.   Vector format.

 New York Land Use  and Natural Resource  Inventory
 New York Department of Commerce
 99 Washington Ave.,  Albany, NY  12245  (518)474-7721
 Land use, agriculture, forestry, wetlands,  water  resources

 Ohio Capability Analysis Program
 Ohio Department of Natural Resources, Division of Soils  and Water
 Fountain Square, Building E-2, Columbus,  OH 43224  (614)265-6778
 Land use/land cover, soils, watershed boundaries

 Rhode Island Statewide Planning Program
 Rhode Island Department  of "Administration,  Statewide Planning
.Program.   265 Melrose Street,  Providence, RI  02907  (401)277-2656
 Land use, vegetation, agriculture, wetlands;  10 acre  cells

 South Carolina Resource Information System
 University of South Carolina, Computer  Science Division
 1244 Blossom Street, Columbia, SC  29208  (803)777-7366
 Land use, vegetation

 Texas Natural Resources Information System
 Texas  Department of Water Resources Systems Central
 Box 13231,  Capital Station, Austin, TX  78711  (512)475-3321
 Land resources,  surface water, and satellite imagery  over Texas
 Listing  available  from above  referenced source for the  available
 CCT's  located in the archive,  contains Landsat MSS and TM imagery.

 Utah Automated Geographic Reference System
 Utah Automated Geographic Reference
 1271 state Office  Bldg.,  Salt Lake City,  UT  84114  (801)533-6290
 Topography,  public land survey

Virginia Commonwealth Data Base
Virginia Polytechnic Institute, Department of Geography
Blacksburg, VA  24061   (703)961-5841
Landforms, cover types

Washington State Digital Land Use/Land Cover
Washington Dept. of Natural Resources, Info Management Division
1102 S. Quince St., EV-31, Olympia, WA  98504   (216)753-01262
Vegetation, cover types, agriculture, forestry; 1:24,000 scale

Wisconsin Department of Natural Resources
Bureau of Water Regulation and Zoning
P.O. Box 7921, 101 S. Webster St., Madison, Wl  53707
Wisconsin wetland inventory

Wisconsin Department of Natural Resources
Bureau of Information Management
GEO Unit, P.O. Box 7921, Madison, WI 53707
Hydrography boundaries, land use/land cover, non-point watersheds

FEDERAL LANDS                           .

Bureau of Reclamation, Remote Sensing Section D-1524
P.O. Box 25007, Denver Federal Center, Denver,  CO  80225

National Park Service, Geographic Information Systems Division
12795 W. Alameda Parkway, Box 25287 GISD, Denver, CO  80225-0287
(303)969-2590  FTS 327-2590


 The field verification methods under development at EMSL-LV revolve
 around  a  statistical  approach  employing  simple  random  and
 stratified  random sampling concepts and the use of emerging Global
 Positioning System (satellite navigation) technology.  The primary
 field  verification method is a site visit  by a  trained  individual
 who will thoroughly describe the site in an objective manner.  In
 situations  where site access  is very difficult or it  is equally
 acceptable  and  cost effective, a small aircraft will be  flown over
 the site to acquire verification data.  Aerial verification data
 will be  acquired in one of two ways, either by a trained  individual
 and/or by acquiring high resolution photography.


 Field  Verification, sometimes referred to as 'field checking' or
 'ground  truthing'  is an essential part of the science of  extracting
 data from aerial photography, satellite imagery, or other remotely
 sensed data  (Colwell 1975).  Ground verification plays an important
 role in  both signature development prior to the  interpretation, as
 well as  in the assessment of  the  accuracy of data extracted, either
 manually or digitally,  from imagery.   However,  it is  extremely
 important that  the same  ground data not be used for both signature
 development and accuracy assessment.   The  most  common approach in
 expressing   accuracy  is  a statement  of percentage of  features
 correctly  classified  when  compared   to  known,  field   verified,
 reference data  (Story and Congalton 1986).  Since  land  cover and
 land use are very  temporal  in nature, it is critical that field
 verification data be acquired  either simultaneously,  as soon as
 possible after,  or during  the  same   season of the year as the
 satellite or aircraft scanner data and  photography.   It is also
 important that  the  sampling design allow verification of each land
 cover  and land  use category  mapped.   This  will  provide  a complete
 accuracy assessment.

 Sampling Design

 If  verification  takes  place  after  image classification  it  is
 recommended  that  a stratified random  sample be  used for  the field
 verification effort.    If  field verification  occurs   prior  to
 classification, then the simple random sample should be  used*  The
 concept  of stratified random sampling involves  the partitioning of
 sample space into 'strata' based on prior knowledge of the sampling
 units  that  is known independently of  the sample.   Each  stratum is
 treated  as  a separate  subuniverse  to gain efficiency and reduce
 variance (Kelly 1970).  Congalton (1988)  showed  that simple random
 and stratified  random  sample approaches  provided  the best results
when creating error matrices to assess  map accuracy  when it was
 necessary to show -t-hat  sisail,  but important areas were  present in
    sample.  A benefit of the stratified random sample approach is


 that classes can  be  stratified,  reducing the  size  of the field
 verification team  to one or two specialists capable of  objectively
 describing the sample area and identifying  class types.   The number
 of  verification  sites  is  still  under development  as  of  this
 writing,  but will  be dependent on the number of classes the number
 of study areas, the ability to group study areas containing similar
 resources and/or within similar ecosystems, and cost.   It has been
 recommended   that  a  one  percent  sample  is  best  for  assessing
 classification  errors  (Congalton,  1988).

 Training  of  the Field  Crew

.A minimum of three  people should be  trained by the  appropriate
 staff   (i.e.,  remote  sensing   scientist,   quality   assurance
 specialist,  and person(s) familiar with the resources to be mapped)
 on  correct  field  techniques  to acquire verification  data.   The
 training  will include a theoretical discussion, a detailed step  by
 step explanation  of the verification protocol followed  by one  or
 two  field trials, and a thorough discussion on reporting.   The
 field  crew should report directly to  the  person assigned as the
 Quality Assurance  Officer of the project.   During the course of the
 field  verification process,  the  field  crew will be  audited  to
 ensure proper implementation  of the  protocol.

 Field  Verification Protocol

 As stated previously,  field verification will be performed in one
 of two ways.   Either  a trained field team  will actually go to the
 site,  or  the site will be observed and photographed from a light
 aircraft.   An aircraft  will  be used  in cases where the  site  is
 difficult to access  and to provide cost  savings in areas where  it
 is  expected  that  the aircraft  information will be  of the same
 quality as  an actual  site visit.   In either case  it will  be
 important that  the  verification crew be   able  to  locate  the
 predetermined sample  position with a known degree of accuracy.  .

 Spatial Referencing in the Field

 Inherently,  the process of field verification  implies that one must
 know,  to  some  degree of accuracy,  the  geographic location with
 respect to  a known reference system of  both the  object to  be
 verified  and of the  observer.  This  can  be accomplished in two
 ways.   The most practical method is to use a base map as reference
 for  both  the object  being  verified  (thematic  polygon)  and  as
 directional  guidance to  the  field person.   The accuracy of the
 particular map is the primary operational consideration.  A map can
 be  utilized  for  a project when  the  sample  can' be  clearly and
 precisely identified on both the map and  on  the  ground.   This  is
 often  the case in  areas of relatively high anthropogenic activity,
 where  verifiable  'landmarks'  can be  identified on both the ground
 and  on the  map.   However, some projects  may be in remote  areas


where  there are very few  identifiable  landmarks.   Since using a
base map for field guidance in these situations is unacceptable, an
alternative method  for field  navigation  is  required.   Global
Positioning System  (GPS)  technology  may be us-ed  to  guide these
field  verification activities.

GPS is based on  a  constellation of earth  orbiting satellites that
transmit  radio  signals which  can  be  received anywhere  on  the
earth's surface.  By measuring the Doppler shift and/or  the timing
of the satellite signal, by using known data on the  satellite orbit
(ephemeris  data),   and  by modelling  for known  errors  caused by
atmospheric distortions and clock error, very accurate earth based
positions can  be obtained.  Horizontal  accuracies of navigational
quality GPS units have been reported to  range between  5.5 - 7.9
meters in  autonomous (single receiver)  mode and 2-5  meters in
differential  (two  receiver) mode  (Jasumback 1988).

GPS units   can  instantly compute the receiver  location  and  can
therefore  be  used to  navigate  field  crews  to the  centroid of
thematic polygons.   The accuracy of the units in a navigational,
autonomous  mode  is sufficient to insure that variations will fall
well  within the minimum  mapping  unit  (one hectare)  of mapped
polygons.   The future accuracy of position fixes depends mostly on
the actions of the Defense Department (DoD) and how much they will
degrade  the performance capabilities  of the  NAVSTAR  satellite
system for  civilian users.   This degradation  process  is called
selective availability  (S/A).  When S/A  is operating, the position
fixes may wander up to  100 meters from the true position. To get a
good position  fix  in the presence of  S/A,  differential  GPS can be
used.    Differential  GPS  utilizes  a  stationary  reference  GPS
receiver, set up on a known location and set to record a reference
file.  Any deviations  from "truth" in  this  reference  file will
allow the generation of a time-tagged position correction that can
be  applied to  a  remote  GPS data  file  collected  using another
receiver, at an unknown location, within a  few hundred  miles of the
reference location.   Using differential positioning in this way,
the standard deviation  of the error for  an absolute point location
can be reduced to  1.5 meters, from typical stated accuracy on the
order  of 2 to  5  meters  (Lange,  et al.,  1988).    Systems  are
available to do this differential processing either in real-time or
as a post-processing operation.

Verification Procedures

The verification procedures for actual  site visits will include
location  of   the   sample  point,   description  of   the  site,
photographing the site, and selecting land use and land cover class
names.  The following detailed procedures will  be carried out by
the  field   crew  after   they   receive   the   1 at it-uds/'longitude
coordinates for the sample sites determined by the  sample design.

Differential GPS techniques will be used to locate the ground point
corresponding to the latitude/longitude coordinates.  The following
steps will be carried out to properly identify the ground location:
1)planning; 2) reconnaissance; 3) survey; and 4)  data reduction and

1.  The planning phase consists of:
*    Defining Project Area -  establishing the overall project area
and defining  the limits of the  sample sites.  Maps and/or aerial
photos should be utilized  extensively to familiarize the crew with
the area prior to the actual field work.
*    Scheduling Operations -  This  involves determining the precise
window of satellite availability and scheduling accordingly.  The
incomplete  status  of the  satellite constellation dictates that
surveys are currently restricted to a few hours per day.  Optimiza-
tion of the schedule  is dependent upon the size of the crew, the
level of accuracy desired, and  the logistics of setup and travel
between control points.
*    Establishing Control Configuration - locating known benchmarks
for both horizontal and vertical  control to perform differential
GPS.  This  is usually accomplished by researching  the records of
various Federal, State and local agencies such as the U.S. Coast &
Geodetic Survey or the state geodetic survey.  It is1 advisable to
have, where possible,  at  least two  control  points each for both
vertical and horizontal positions so that there is a double check
for all control locations.
*    Selecting  Survey Locations  -  there  are several  specific
considerations:  Points should be accessible during the satellite
window of availability, often during unusual  hours.  Points should
have  continuous  and  direct line-of-site  to  the path  of  the
satellites  in the sky. Points should  not be near power lines,
substations or large metal  objects which can cause multipath inter-
ference and corrupted data.

2. Reconnaissance

The Reconnaissance phase is an important part of a successful GPS
operation and is usually performed by an individual  or crew at some
point prior to the actual  acquisition  of field verification data.
The purpose of this phase  is to:
*    Locate and Verify Control Point Locations - This is critical
to the success of the overall survey.  Often, monuments have been
damaged, stolen, buried or vandalized.  If a  control point,  cannot
be recovered, a replacement must be located.   This can drastically
change the schedule and logistics of the field survey.
*    Preview instrument locations -  obtain permissions and  verify
that there are none of the above restrictions.
*    Physically establish point locations - This is accomplished by
using a standard surveying marker such as  an iron pipe, a hub and
tack,  or a brass  nail.   All  points should  be documented with
detailed descriptions.


 3.  Survey

 The actual  GPS  survey  consists of:     .
 *    Establishing  a   Schedule  of  Operations  -  This  involves
 determining the window of satellite configuration availability and
 scheduling  the GPS  sessions.  This is dependent  on  the  size of the
 crew, the level of accuracy desired, and the logistics of  setup and
 travel between  control points.
 *    Performing GPS Survey - The  crew  must warm-up,  check and
 program the receiver for proper operation.  Depending  on the unit
 being utilized  sufficient  battery power must be available and the
 receiving antenna must be leveled on a tripod and centered exactly
 over the control  point location.   Log sheets containing critical
 information on position, weather, timing, height of  instrument, and
 local  coordinates  must  be  maintained.    Once the  session  is
 completed,  the  receiving  equipment must be disassembled, stored,
 and log  and  tape  files   documented.    If another  session  is
 scheduled,  this process must  be conducted quickly  and  efficiently
 so  that the crew  can be at the next location and be set up in time
 for the scheduled window of satellite  availability.

 4.  Data Reduction and  Processing

 Data reduction  and  transfer consists of:
 *    Data Transfer - reading the raw  data  from the GPS cassette
•tapes into a structured data base for processing and  backup copies
 *    Pre-processing - GPS  data is  not immediately  useable and
 consists of satellite navigation messages, phase  measurements, user
 input field data  and other information that  must be transferred to
 various files  for processing before  network computations  can be
 accomplished.   There  are  five components  to  the  pre-processing
 *    Orbit  Determination   -  The  software  uses  the  satellite
 navigation  messages to  compute one  unambiguous   orbit  for  each
 *    Single-Point  Positioning  -  Computations   of  the  clock
 corrections and parameters  for each receiver.
 *    Baseline  Definition  -  establishing  general  locations  of
 receiving stations the computation of best pairs  of  sites for
 baseline definition.
 *    Single difference file creation -  single differences are the
 differences between simultaneous phase measurements  to  the same
 satellite from two sites.  This is the basic data  from  which network
 and  coordinate data will be derived.
 *    Data screening - This a routine that  allows .for  automatic and
manual screening  of the  single difference files.  Breaks,  cycle
 slips, and poor observations  can be detected.
 *    Computation  -  This component uses tha pre-processed data to
 compute the network or  sites and  give a  full solution showing
 geographical  coordinates   (latitude,  longitude and ellipsoidal


height), distances of the vectors between each pair of sites in the
network,  and  several assessments  of  accuracy  of  the  various
transformations and residuals of critical computations.

After  the  location  of  the  sample  point  has   been  accurately
determined, a stake or pole will be  driven  into the ground at that
point.  A measuring tape with a  length equivalent  to 1.5 times the
pixel  size will  be attached to  the stake.    The tape  will  be
stretched out to inscribe a circle.   While the use of a stake and
tape  are  considered  ideal  to  ensure  the proper  size  area  is
documented, it is  recognized that ground conditions  and time may
preclude the use of these tools.  It  is highly recommended that the
field  crew be very familiar with  the  size  of  the  verification
samples, and practice with a stake and tape during training.  The
area within  the  circle will be thoroughly described  starting at
ground-level (0-.3m), then .3 to 2 meters above ground level, and
finally  the  area  above 2  meters will  be described.    After the
descriptions are  completed a diagram of  any resource boundaries
will be  sketched  along with any major  features.   A photographic
documentation of the  site  will  be prepared by taking 35mm slides
from the center looking out at eight different directions, namely,
N, NE, £, SE, S, SW, W, NW.  A member of the  field crew should be
located  at the left edge  of the  camera frame at the  end of the
tape.  This will provide a relative scale for comparison of feature
size as well as demarcating  the edge of the sample circle.  Extra
care should be taken to ensure that the photography encompasses the
entire site, from  the ground up,  in each frame.  The final field
verification  procedure  will   be   the   selection  of  the  most
appropriate land use and land cover classes for the site.

If the verification data  is being acquired via a light aircraft,
the point location should  be determining using GPS or if GPS  is not
available  then  Loran.   The  area surrounding the point should be
described as best as possible using the field verification form and
35mm slides should be taken  both vertically and obliquely.

Field Verification Reporting

The following pages present the form which will be .filled out in
the  field during  the verification  process.   This form may and
should  be customized  for unique ecosystems and projects which
require different field information. The form will be completed in
the order  presented with  the last component  being a selection of
land cover and land use class names.  The selection of class names
is performed last  to prevent any bias in the description  of field
characteristics.  Copies of the  completed forms will be sent to the
Quality  Assurance Officer and the technical  staff performing the
accuracy assessment.

                      Field Verification  Form
Field Crew.
 Site   ID
     Method  of  determining Lat/Long:   	GPS    	USGS map
Height of GPS	
GPS    Satellites:
                Time of GPS use
General Description
1) Surface  (0-.3  meters):
Surface roughness and moisture:   check one descriptor from  each
column which best describes the entire area.  If there is a portion
of  the site which  is  distinctly different,  please  label  the
appropriate  descriptor  with  a location,  i.e.,   SW,  S,  SE,  W,
Central, E, NW, N, NE.
	smooth noncultivated       	dry             	_flat (0-3%)
	hummocky noncultivated      	moist          	gentle (3-10%)
	cultivated      	saturated        	hilly  (10-20%)
	no-till         	standing wet    	steep  (20-40%)
                                   .	v steep  (40-60%)
                               	extreme  (>60%)

Surface  features:   record the  percentage of  each surface  type
present on the site; also describe any features which are  not  on
this list.  If  there is a  portion of the  site which is distinctly
different, please label the appropriate descriptor with a  location,
i.e., SW, SC,  SE, WC,  C, EC, NW,  NC, NE.
  jnetal waste
  "misc. waste
   'low shrubs
   "other i
_leaf  litter

Please check appropriate category of water from list below.

	lake  	pond 	river 	stream 	ditch 	marsh 	tidal

Does the surface  appear to be altered by  human activity	,  or
does the surface  appear to be in a  natural  state (this includes
areas that may have had historical human activity, but have had
time to return to natural species)	.

Surface condition general comments:
2) Near Surface (.3-2 meters)

     Near surface  features:   record the percentage  of each type
present on the site and if possible list the species;  also describe
any features which are not on this list.  If there is a portion of
the  site  which   is   distinctly  different,   please  label  the
appropriate  descriptor with a  location,   i.e.,  SW,  s,   SE,  w,
Central, E, NW, N, NE.

          	shrubs (.3-1 meters);
          	shrubs  (1-2 meters);
          	trees  (.3-1 meters);
          	trees  (1-2 meters);
          	metal waste;
          	misc. waste;
Near surface condition general comments:

3)  Above Surface  (>2 meters)

     Above surface features: this category is designed to describe
trees and buildings;  record the species type, stand density, and
for buildings list the type  of structure, i.e., single family home,
apartment,  commercial,  industrial,  etc.     Please   indicate  if
portions of the site are distinctly different.

Is this a tree plantation or orchard:  	yes     	no

Buildings: Type of structure

Above surface condition general comments:

4) Wetlands and Near Coastal Areas

     .This resource  type  is unique since some  of  the features of
interest of  submerged.   If possible,  please  check the substrate
type.  Indicate the  water depth, flooding frequency, and water type
if possible  for each portion of the  site  by  using the following
appropriate descriptors:  SW, SC, SE, WC, C, EC, NW, NC, NE.
Water depth      Flooding              Water type      Substrate
	0-.3 m 	never       	_fresh       	mud
	.3-1 m 	saturated           	salt        	sand  (.05-2mm)
	1-2 m  	daily  (tidal)          mix         	gravel  (2-8cm)
	2-5 m  	seasonally         	cobbles (8-25cm)
	5-10 m 	intermittently     	stones  (25-6lcm)
	>10 m  	semipermanently    	boulders  (>61cm)
          	permanently        	organic matter
Please list as best as possible the species present:

Wetlands/Near  Coastal  general  comments:
Please diagram any distinct resource boundaries or features present
within the  verification site below.   Also indicate any features
which would provide a better accuracy assessment of the digital
database.  Take photographs (8)  looking to the N, NE, E,  SE, S, SW,
W, and NW from the verification site center point.

  Roll #      	, Frame  #'s 	"	•	.