001R95007
Multi-Rescluticn
Land Characteristics
Ccnscrtium
Documentation Notebook
Compiled and Edited by
Thaddeus J. Bara
Senior Scientist
ManTech Environmental Technology, Inc.
Research Triangle Park, North Carolina
Prepared for
Denice Shaw
Technical Coordinator
EMAP-Landscape Characterization
Atmospheric Research and Exposure Assessment Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Contract: 68-DO-0106
M « M mm f* mr- -w i mi f i m i tm
MAKE* ?"; ? '•>=• ••==
I wlrH W= i'j —.i::—.i= =
MANTECH ENVIRONMENTAL TECHNOLOGY, INC.
P.O. Box 12313, Research Triangle Park, North Carolina 27709
A ManTech International Company
-------
Uiilli I OM luliui
Land Characteristic§ Ccn§crtium
Documentation Notebook
The Multi-Resolution Land Characteristics (MRLC) Consortium
An Innovative Partnership for National Environmental Assessment
Five federal environmental monitoring programs, EMAP (USEPA),
GAP (USFWS), NAWQA (USGS), C-CAP (NOAA), and NALC
(USEPA/USGS) have formed a partnership with the EROS Data
Center (USGS) to facilitate the development of comprehensive
land characteristics information for the United States. Each of the
respective programs brings to the Consortium unique experience,
expertise, and resources. Common requirements for source
satellite data, preprocessing, spectral clustering, ancillary data
acquisition and integration, and data management and
distribution have been identified. The goals for the MRLC
Consortium include the generation of landcover data for the
conterminous United States and the development of a flexible and
functional land characteristics database that meets the diverse
needs of the participating programs and federal agencies.
EMAP Environmental Monitoring and Assessment Program
GAP Gap Analysis Program
NAWQA National Water Quality Assessment Program
C-CAP CoastWatch Change Analysis Program
NALC North American Landscape Characterization Project
EDC EROS Data Center
-------
Foreword
This notebook was prepared by ManTech Environmental
Technology, Inc. in response to Technical Directive EF6 of
Contract 68-DO-0106 for the Environmental Monitoring and
Assessment Program (EMAP)-Landscape Characterization,
Atmospheric Research and Assessment Laboratory, Office of
Research and Development, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina. Although
production of this notebook was wholly funded by the
U.S. Environmental Protection Agency, it does not necessarily
reflect the views of the Agency. Mention of trade names or
commercial products does not constitute endorsement or
recommendation for use by the Agency.
-------
MRLC Consortium
Documentation Notebook
May 1995
INTRODUCTION
The Multi-Resolution Land Characteristics (MRLC) Consortium Documentation
Notebook is intended to serve as a primary reference for the MRLC Consortium. While it is
intended that the notebook primarily serve as the chief source of information for use by the
participating agencies, the notebook is also of value to other programs and individuals
interested in reviewing the history and continuing activities of the Consortium. Every effort
is being made to ensure that the notebook contains the most up-to-date and accurate
information available describing the ongoing activities of the MRLC Consortium. Users of
the notebook are strongly encouraged to provide additional information, editorial or content
comments, and other suggestions for inclusion in future updates of the notebook.
This notebook is designed as a loose-leaf notebook allowing for the regular and
continual updating of individual sections and the introduction of new sections as appropriate.
Quarterly updates will be provided to all agencies and individuals on the Documentation
Notebook mailing list. These updates will include replacement material, new material, and
new sections. Full instructions for the updating of notebook material will be provided with
each update package. Currently fifteen sections are active.
Users with Internet access may view a hypertext exhibit of the MRLC Documentation
Notebook via the World Wide Web. The URL for the exhibit is:
http://www.epa.gov/grd/mrlc
In some cases, relevant documentation is too long to effectively include in this
notebook format. Pointers are provided in this document to identify the appropriate
information source. Copies of the specific documentation are retained in the MRLC central
files and will be available upon request to interested parties.
-------
MRLC Consortium
Documentation Notebook
May 1995
MULTI-RESOLUTION LAND CHARACTERISTICS CONSORTIUM
DOCUMENTATION NOTEBOOK
TABLE OF CONTENTS
SECTION DESCRIPTION
1. ABOUT THE MRLC
1.1 Contact List
1.2 Executive Review of MRLC - 10/94
2. MEMORANDUM OF UNDERSTANDING
3. TM SCENE PURCHASE AGREEMENT
4. TM SCENE SELECTION
4.1 Scene Selection Criteria
4.2 Scene Selection Process Flow Diagram
4.3 Scene Selection and Order Status
5. TM SCENE ACQUISITION AND ARCHIVING
5.1 EDC Scene Archive Flowsheet
5.2 TM Scene Acquisition and Archive Status
6. TM SCENE PREPROCESSING
6.1 TM Scene Preprocessing Overview
6.2 Online Map of TM Scene Preprocessing
6.3 Preprocessing Protocol Documentation
6.4 Spectral Clustering of Scenes
7. SCENE LABELLING AND CLASSIFICATION
7.1 SPECTRUM Software
7.1.1 NASA Ames notes on SPECTRUM
7.1.2 Reston, VA SPECTRUM Pilot Workshop Summary
7.2 MRLC Labelling Pilot Programs
7.3 Landcover Classification References for MRLC Agencies
8. MRLC METADATA STANDARDS
-------
MRLC Consortium
Documentation Notebook
May 1995
9. MRLC DATABASE MANAGEMENT
9.1 MRLC Information System Software Design Review
9.2 MRLC Product Relationships
9.3 Derivative Metadata Format
9.4 MRLC Data Set Tables
9.5 GLIS XGLIS Design Document Version 2.0
10. MRLC ACCURACY ASSESSMENT
10.0 GAP Accuracy Assessment Workshop
11. INFORMATION ON PARTICIPATING AGENCIES
11.1 EMAP
11.2 GAP
11.3 NAWQA
11.4 C-CAP
11.5 NALC
11.6 EDC
12. MRLC CONSORTIUM MEETINGS
12.1 Portland,OR--4/93
12.2 Las Vegas, NV -- 5/93
12.3 Sioux Falls, SD - 6/93
12.4 Minneapolis,MN -- 8/93
12.5 Mountain View, CA -- 11/93
12.6 Santa Barbara, CA -- 2/94
12.7 Reno, NV -- 4/94
13. MRLC CONFERENCE CALLS
14. MRLC REGIONAL IMPLEMENTATIONS
14.1 Region 2/3 Implementation
15. MRLC DATA RECIPIENTS
15.1 Summary Listing of MRLC Data Recipients
15.2 Details of MRLC data use by organization and project
-------
MRLC Consortium
Documentation Notebook
February 1995
SECTION 1
ABOUT THE MRLC CONSORTIUM
This section contains reference information pertaining to the entire MRLC Consortium
and its efforts. There are two subsections:
1.1 Contact List
1.2 Executive Summary of MRLC
-------
MKLC Consortium
Documentation Notebook
May 1995
1.1 Contact List
This is the most recent contact list of principal individuals involved with the Multi-
Resolution Land Characteristics (MRLC) Consortium. The list begins with the primary
contacts within the participating programs, followed by other federal personnel and supporting
contractors.
PRIMARY PROGRAM CONTACTS
Peter Campbell
U.S. Environmental Protection Agency
AREAL - MD75
EMAP Center
Research Triangle Park, NC 27711
Voice: (919) 541-2957
FAX: (919) 541-3615
Email: campbell.peter@epamail.epa.gov
Jeff Eidenshink
EROS Data Center
U.S. Geological Survey
Sioux Falls, SD 57198
Voice: (605) 594-6028
FAX: (605) 594-6589
Email: jeiden@edcsnw24.cr.usgs.gov
Donald W. Field
National Marine Fisheries Service
Beaufort Lab
101 Fivers Island Road
Beaufort, NC 28516
Voice: (919) 728-8764
FAX: (919) 728-8784
Email: dfield@hatteras.bea.nmfs.gov
Michael D. Jennings
U.S. Department of the Interior
Idaho Cooperative Fish and Wildlife
Research Unit
College of Forestry
University of Idaho
Moscow, ID 83843
Voice: (208) 885-3565
FAX: (208) 885-9080
Email: jennings@crow.csrv.uidaho.edu
Thomas Loveland
EROS Data Center
U.S. Geological Survey
Sioux Falls, SD 57198
Voice: (605)594-6066
FAX: (605) 594-6589
Email: Ioveland@edcsnwl9.cr.usgs.gov
Denice M. Shaw
U.S. Environmental Protection Agency
AREAL-MD75
EMAP Center
Research Triangle Park,NC 27711
Voice: (919)541-2698
FAX: (919) 541-3615
Email: shaw.denice@epamail.epa.gov
-------
MRLC Consortium
Documentation Notebook
May 1995
Gail P. Thelin
U.S. Geological Survey - WRD
Rm. W2233, 2800 Cottage Way
Sacramento, CA 95825
Voice: (916)978-4645
FAX: (916) 979-2669
Email: gpthelin@wl 14dcascr.wr.usgs.gov
L. Dorsey Worthy
NOAA Coastal Center
2000 Bainbridge Avenue
Charleston, SC 29408-2623
Voice: (803) 974-6234
FAX: (803) 974-6224
OTHER PERSONNEL
Susan Benjamin
U.S. Geological Survey - National Mapping
Division
Ames Research Center 242-4
P.O. Box 1000
Moffet Field, CA 94305
Voice: (415) 604-3914
Fax: (415) 604-4680
Email: susan@tenaya.arc.nasa.gov
John Dwyer
Hughes STX Corporation
South Dakota Operations
EROS Data Center
Sioux Falls, SD 57198
Voice: (605) 594-6060
FAX: (605) 594-6589
Email: spectra@dgl.cr.usgs.gov
Ronald Feistner
Hughes STX Corporation
South Dakota Operations
EROS Data Center
Sioux Falls, SD 57198
Voice: (605)594-6878
FAX: (605) 594-6589
Email: feistner@edcserverl.cr.usgs.gov
Kent Hegge
Hughes STX Corporation
South Dakota Operations
EROS Data Center
Sioux Falls, SD 57198
Voice: (605)594-6976
FAX: (605) 594-6589
Email: hegge@edcserverl.cr.usgs.gov
-------
MRLC Consortium
Documentation Notebook
May 1995
Mark Henderson
ManTech Environmental Technology, Inc.
2 Triangle Drive
P.O. Box 12313
Research Triangle Park, NC 27709
Voice: (919) 541-4205
FAX: (919) 541-4958
Email: henderson@igc.org
Thomas M. Holm
EROS Data Center
U.S. Geological Survey
Sioux Falls, SD 57198
Voice: (605) 594-6142
FAX: (605) 594-6589
Email: holm@edcserverl.cr.usgs.gov
Joy J. Hood
Hughes STX Corporation
South Dakota Operations
EROS Data Center
Sioux Falls, SD 57198
Voice: (605) 594-6045
FAX: (605) 594-6589
Email: jhood@dgl.cr.usgs.gov
Ronald Kanengieter
Hughes STX Corporation
South Dakota Operations
EROS Data Center
Sioux Falls, SD 57198
Voice: (605) 594-6875
FAX: (605) 594-6589
Email: ron@edcserverl.cr.usgs.gov
Morgan Sarges
Hughes STX Corporation
South Dakota Operations
EROS Data Center
Sioux Falls, SD 57198
Voice: (605)594-6931
FAX: (605) 594-6589
Email: sarges@dgx.cr.usgs.gov
Paul Severson
Hughes STX Corporation
South Dakota Operations
EROS Data Center
Sioux Falls, SD 57198
Voice: (605)594-6966
FAX: (605) 594-6589
Email: pseve@edcserverl.cr.usgs.gov
James A. Sturdevant
EROS Data Center
U.S. Geological Survey
Sioux Falls, SD 57198
Voice: (605)594-6080
FAX: (605) 594-6589
Email: sturdevant@edcserverl .cr.usgs.gov
-------
MRLC Consortium
Documentation Notebook
February 1995
1.2 Executive Summary of MRLC
This section consists of presentation materials prepared for the MRLC Summit meeting
in October 1994.
-------
o
• CC
CO
C/)
CO
tr
LLJ
DC
d.
DC
111
Q
111
LL
DC
111
X
O
CO
HI
o
DC
O
CO
111
DC
in
LLJ
O
LL
DC
LLJ
O
O
O
Q
o
1
CO
Q
LLJ
LLJ
Q X
O
DC
LLJ
CO
LLJ
DC
-------
O
O
O
CM
o
O
4
CO
O
tf.
a.
o
CO
LU
CO
Q
O
O
CO
O
o:
uio
o
QQO
< O UJ
H 01 CO
Z 0. <
00°
CM
o:
o
U-
Q.
CO
cn
o: LU O
gi-
83
CO
-------
O
LU
a.
O
LU
LLJ
Q
CC Q
O
DC
LU
CO
LJJ
CC
UJ
r$ Q- O
T < Q
O O LU
0>
^
5
i
9
(O
09
cc
O
LL
O
UJ
UJ
3
QC
O.
-------
CO
O
O
O
cc
O
Q
O
cc
0-
<
Q
D < ^
O H ^
cc < o
S: Q in
I I I
u
o
O)
o
tj
^
-------
CO
LU
O
CC
O
LU
CC
Q
O 5:0.
1891
2 UJ O HI
CL
<
CD
>1000L$
0
h-
05
I
CO
CD
CC
O
LL
O
LU
O
HI
QC
Q-
-------
E
OP
E
'-LI
DCQ
o
a:
111
CO
LLJ
DC
•EGRATION
r-
SPATIAL DATABASE fe
h-
z
•— •
ACCURACY ASSESSN
•
CHANGE DETECTION
e
ICATION
LL
CLUSTERING/CLASSI
«
CE/INTERSENSOR
STRATEGIES
Z
SENSOR PERFORMA
t
RELATIONSHIPS
O)
G
O
u
1
-------
tt)
Lli
DC
o
O LJJ
DC DC
CL CL
LU
LU
CO
CO
O
O
LL
CO
CO
O
O)
73
o
I
1
CO
o
LLI
O
en
o.
-------
0>
o
o
u
•s
E
O
cc
0)
or
O
LABORAT
_i
O
O
-CAP PROJECTS
O
•
CO
AP STATE PROJECT
O
•
* **
AWQA STUDY UNITS
Z
•
cc
V^^^B
o
Q.
DC TECHNICAL SUP
LU
•
LL
LL
f
CO
Q.
<
LU
•
O
IRLC LIASON POSITI
^
•
CO
nr
Uba
0
H
ROGRAM COOPERA
CL
•
-------
£
I
5
&
&
S
O
| Cluttarad
S
9
1
&
2
a
i
•
«
£
i
1
ui
!!
It
si
53
i!
I
O
O
ul
TO
to
Q
O
cc
*
i
*
*
i
i
u
r\
1
I
s
Si
II
II
-------
05
C/D
CC
0
o
DCO-
O
FERENC
LLJ
CC
O
LJJ
CD
ISTRATK
0
LJJ
DC
O
O
O
DC
LU
CO
O
CO
h-
o
Q
O
CC
CL
O
D_
CC
o
at
o
,0
o
CM
o
-------
DC UJ
HI
LU
LU
DC
O
LU
CO
O
DC
g
I-
Z
LU
LU
O
O
o
UJ
o
CC
LU
D_
O
O
LU
o>
fc
O
In
o
O
CC
2
LU
DC
Q_
CC
LU
C/) ^
-------
ui
DC
o
<
CL
CO
LU
a:
o
LL
Q
CO
Q
CO
O
111
X
Q
LLJ
O
o
LLJ
LLJ
g
cc
Q.
CD
DC
CO
<
<
Q
CO
111
LL.
o
CO
LU
CO
1
o
(O
o
o
en
-------
-------
Q
O
(T
0°
uo
o
LU
CO
o
or
D_
bS
LU Q
DC
Q
O
CM
CM
o
z
CO
CO
LU
LU
DC
D
o>
o
O
_!
CC
2
o
LU
< CC
O
DC
Q_ O ^
-------
a
%f.
I
LU
O
CC
^^M^H
O
cc
in
Q_
O
o
o
cc
o
LL
CC
o
111
CC
LL
O O
CO
CO
111
O
O
CC
Q_
111
CC
Q_
ACQUIS
oi
&
O
CNi
O
O
z
LU
LU
O
<
<
-------
Q
tt)
g
o
I
o
cc
O)
5
O
O
CM
O
O
u.
LL1 £
LU
111 t" UJ
ccioQ<
H CO
CO
-------
LL
O
LLJ
LLJ
DC
LU DC
O
LU
X
LU
CO
LU
_
£
DC
O
Q.
D_
O
Q
<
-------
O)
O
O
?5
o
O
Q
oc
Oiii
LJJ
O
CD <
i
O
CO
O
01
<
Q.
LL1
0=0. a.
0<
CL
UJ ill CD
C/)
-------
MRLC Consortium
Documentation Notebook
May 1995
Section 2
Memorandum of Understanding
This section contains the Memorandum of Understanding establishing the MRLC
Consortium which was signed on March 10, 1995.
MEMORANDUM OF UNDERSTANDING
AMONG:
U.S. ENVIRONMENTAL PROTECTION AGENCY
NATIONAL BIOLOGICAL SERVICE
U.S GEOLOGICAL SURVEY
NATIONAL OCEANIC AND ATMOSPHERIC ADMINISTRATION
-------
MRLC Consortium
Documentation Notebook
May 1995
PURPOSE
This Memorandum of Understanding (MOU) is entered into for the purpose of forming a
multi-agency partnership, hereafter referred to as the Multi-Resolution Land Characteristics
Consortium (MRLC). The agencies involved are:
^^ » •
U.S. Environmental Protection Agency - Environmental Monitoring and
Assessment Program (EMAP);
National Biological Service (NBS) - Gap Analysis Program (GAP);
U.S. Geological Survey - National Water-Quality Assessment Program (NAWQA); and
EROS Data Center (EDC);
National Oceanic and Atmospheric Administration.- Coastal Ocean Program,
Coastal Change Analysis Program (C-CAP).
This MOU establishes the basis for the seminal goal of the MRLC: the joint acquisition
of Landsat Thematic Mapper (TM) imagery for the conterminous United States. The joint
acquisition of TM imagery and the sharing of data by the MRLC partners will result hi
significant savings of government funds. Also, this MOU lays the groundwork for the long-term
goals of MRLC. These goals include collaborative research on and development of a flexible
and functional land characteristics data base for use by the MRLC and other Federal, State, and
local organizations.
Although commonalities and differences exist among MRLC partners, program
objectives are complementary. Cooperative efforts will build on the strengths of individual
programs, resulting in better service of the National interest It is the intent of the MRLC,
therefore, to: promote cooperation; coordinate land-cover mapping activities of the programs to
the greatest possible extent; prevent duplication of effort; and, ensure that information produced
by each program is used to the maximum extent possible. Coordination will include joint
synthesis and integration of land-cover mapping as appropriate.
MRLC Memorandum of Understanding
-------
MRLC Consortium
Documentation Notebook
May 1995
BACKGROUND AND SCOPE OF THE PARTNER PROGRAMS
U.S. Environmental Protection Agency
Environmental Monitoring and Assessment Program
The Environmental Monitoring and Assessment Program, managed by the U.S.
Environmental Protection Agency's (EPA) Office of Research and Development, is a monitoring
and assessment research effort designed to investigate and report on the condition of the Nation's
ecosystems. Through the development of research tools to support assessment of the Nation's
resources, EMAP is recognized as a major component of the U.S. ecological research program.
EMAP is assessing the condition of ecological resources—estuaries, surface waters, the Great
Lakes, agroecosystems, rangeland ecosystems, forest, and wetlands". When fully implemented,
EMAP will provide comparable, high-quality data at several spatial scales. EMAP is developing
new ecological monitoring and assessment research tools and generating statistically reliable,
scientifically-defensible data, which are being combined with data from other monitoring
programs to provide comprehensive information for land managers as well as periodic overviews
of the overall effectiveness of national environmental protection policies and practices.
National Biological Service
Gap Analysis Program fGAP1
The Gap Analysis Program provides a regional and national overview of the distribution
and conservation status of biological diversity by producing comprehensive and synoptic
biogeographic data. The analysis is accomplished through the comparison of the distributions of
vegetation and all vertebrate species with land ownership and management, using geographic
information system technology. A central question focuses on how adequately (or inadequately)
are native vertebrate species and vegetation cover types represented in areas that are managed for
long-term persistence. That is, where are the "gaps" hi the overall mix of conservation lands
when it comes to maintaining representative examples of biodiversity components?
Additionally, there is broad utility for GAP's digital spatial data for local, state and national
levels of land and resource planning and decision making. To accomplish the goal of providing
such information in the near-term, GAP is mapping the distributions of vegetation cover types
and vertebrate species against land ownership and land management categories for the
conterminous 48 States at a scale of 1:100,000. Vegetation is mapped first using Landsat
Thematic Mapper imagery along with substantial amounts of ancillary data (aerial photography
and videos, agency records, existing maps, and field observations). Vegetation map units are
used as one of several data layers to model and predict present-day distributions of vertebrate
species. Land ownership and land management categories are compared to the distributions of
vegetation cover types and vertebrate species. By focusing on higher levels of biological
MRLC Memorandum of Understanding
-------
MRLC Consortium
Documentation Notebook
May 1995
organization, this method is less expensive and more effective than a single-species approach to
biological conservation.
U.S. Geological Survey
National Water-Quality Assessment fNAWOA") Program^
The National Water Quality Assessment Program of the U.S. Geological Survey is
designed to describe the status of and trends in the quality of the Nation's ground- and
surface-water resources; and, to link information on these status and trends with an under-
standing of the natural and human factors that affect the quality of water. The program integrates
information about water quality at a wide range of spatial scales—from local to national—with a
focus on water quality conditions that affect large areas of the Nation or that occur frequently
within small areas.
The building blocks of the NAWQA Program are investigations, which are conducted in
60 major hydrologic basins (Study Units) of the Nation. Collectively, the NAWQA's Study
Units cover a large part of the U.S., encompass the majority of National water use, and include
diverse hydrologic systems that differ widely in the natural and human factors that affect
water-quality. This approach ensures that the most important national water-quality issues can
be addressed by comparative studies. Each Study Unit investigation consists of intensive data
gathering activities for 4 to 5 years, followed by 5 W6 years of low-level assessment activity.
Approximately one-third of the Study Units will be examined intensively at a given time; and,
the 10-year cycle will be repeated perennially. NAWQA National Synthesis projects combine
and interpret Study Unit results for priority regional and national water quality issues.
U.S. Geological Survey
U.S. Global Change Research Program
The EROS Data Center's (EDC) Multi-resolution Land Characteristics Monitoring
System is a prototype system in which satellite remotely sensed data are used to map current land
cover characteristics, and monitor, target, and assess environmental changes. This includes
objectives to develop: (1) a global 1-km land characteristics data base; (2) regional 30-m land
characteristics data bases; and (3) a multi-resolution framework for monitoring synoptic
environmental processes and targeting significant areas of change. The project is part of the U.S.
Global Change Research Program, specifically the Global Change Data Systems activity that
specifies generation of data and information about the land essential to understanding, modeling,
and predicting global change processes. This activity calls for: (1) creating regional to global
land characteristics data sets; (2) developing techniques for mapping, monitoring, analyzing, and
predicting rates, patterns, and impacts of landscape changes; and (3) facilitating access to land
MRLC Memorandum of Understanding
-------
MRLC Consortium
Documentation Notebook
May 1995
data for global change research and environmental applications. The Global Change Data
Systems activity is coordinated with other agencies through the National Science and
Technology Council and the Committee on Environment and Natural Resources. Participation is
also based on the Land Remote Sensing Policy Act of 1992 (Public Law 102-555) which
established the "National Satellite Land Remote Sensing Archive" at the EDC for the purpose of
maintaining a permanent government archive of global Landsat and other remote sensed data,
and ensuring proper storage, preservation, and timely access to data in the public domain for
long-term monitoring and environmental studies.
National Oceanic and Atmospheric Administration
Coastal Change Analysis Program
To better understand and manage living marine resources, scientists and managers need
up-to-date information on the distribution and abundance of coastal fisheries habitats and how
these habitats change with time. In accordance with these needs the Coastal Ocean Program of
the National Oceanic and Atmospheric Administration (NOAA) has initiated the Coastal Change
Analysis Program (C-CAP). The purpose of C-CAP is to develop a comprehensive, nationally
standardized information system to assess changes in wetlands and adjacent uplands in coastal
regions of the U.S. C-CAP utilizes satellite sensors to detect change in coastal emergent
wetlands and adjacent uplands; and, it uses aerial photography to detect change in submerged
aquatic vegetation. The ultimate goal of the program is to monitor coastal areas every one to five
years, depending on the rate and magnitude of observed change in each region. The protocols for
establishing C-CAP data were developed through a series of workshops that brought together
approximately 250 technical and regional experts, and representatives of key. state and federal
organizations. Projects using satellite-based TM data to detect change in uplands and emergent
wetlands in the Chesapeake Bay drainage area and using aerial photography to detect change in
seagrasses in North Carolina have been completed. Several research projects are underway to
further refine change detection methodologies. NOAA enters into this agreement under authority
of 15 U.S.C. §§313,2904 and 1525-27.
USE OF DATA BY MRLC PARTICIPANTS
Data from the joint acquisition will be archived and distributed by EDC (at cost of
reproduction plus any preprocessing not already paid for by MRLC) to any MRLC participant as
specified in the purchase agreement between EOS AT and EDC (Annex 1).
MRLC Memorandum of Understanding
-------
MRLC Consortium
Documentation Notebook
May 1995
COOPERATIVE ACTIVITIES
MRLC partners agree to cooperate to the greatest extent possible to maximize the use of
existing data and resources and to prevent duplication of efforts. Cooperation may include, but is
not limited to:
1. Participation in joint research efforts to continue development of remote sensing methodol-
ogies, and collaborative regional projects to develop flexible land characteristics data utilizing
techniques and data from all participants. This includes efforts to maximize data
compatibility among the partner programs;
2. Exchange of data sets, inventory and monitoring methodologies, sampling and analytical
procedures, quality assurance protocols, and training programs;
3. Transfer of such base technology products as procedures for remote imagery acquisition and
interpretation, information management systems, and publications;
4. Participation in formal reviews of study plans and proposals, technical manuals, or project
synthesis reports; and
5. Convening of, and attendance at, joint workshops to ensure:
• early and continuing communication about mapping and monitoring plans and priorities;
• interactive planning and reviews of plans; and
• interim review of results, discussions of areas of concern, and recommendations of actions.
THIRD PARTY LIABILITY
Liability by the Federal Government for acts of its employees is governed by the Federal
Tort Claims Act and other Federal statutes. Nothing contained herein will constitute a waiver by
any party of its sovereign immunity and the limitations set forth by Federal law.
AMENDMENTS AND REVIEW
This Agreement is subject to revision and can be amended, extended, or modified by the
mutual written consent of the participating agencies.
MRLC Memorandum of Understanding
-------
MRLC Consortium
Documentation Notebook
May 1995
PROJECT ANNEX PROVISION
Any activity carried out under this MOU will be agreed upon by the MRLC partners in
writing and in accordance with the MRLC partners legal authority. Whenever more than the
exchange of technical information or visits is planned, such activity will be described in an
Annex to this MOU. which will set forth, in terms appropriate to the activity, an implementation
plan, technical requirements, financial arrangements, and other responsibilities, obligations, or
conditions not addressed in this MOU.
OTHER PROVISIONS
Nothing herein intentionally conflicts with current directives or the applicable laws of any
of the parties entering this agreement. If the terms of the agreement are inconsistent with
existing directives or with the applicable laws of any of the parties entering the agreement, then
those parts of this agreement that are determined to be inconsistent shall be invalid. The
remaining terms and conditions of this agreement not affected by any inconsistency shall remain
in full force and effect
Should disagreement arise about the interpretation of the provisions of this agreement, or
amendments or revisions thereto, that cannot be resolved at the operating level, the area(s) of
disagreement shall be reduced to writing by each party and presented to the other parties for
consideration at least ten (10) working days prior to forwarding the areas of disagreement to
respective higher officials for appropriate resolution.
TERMS OF THE AGREEMENT
The terms of this a'greement shall become effective upon the signature of all approving
officials of the respective parties entering into this agreement. This agreement shall remain in
effect until terminated by (1) mutual written agreement, (2) at least thirty (30) days advance
written notice by any party, or (3) completion of this agreement
MRLC Memorandum of Understanding
-------
MRLC Consortium
Documentation Notebook
May 1995
SIGNEES:
_
Robert J. Huggett, PK±>. / //
Assistant Administrator for Research and Development
U.S. Environmental Protection Agency
Date
H. Ronald Pulliam, Ph.D.
Director
National Biological Service
Date
Gordon P. Eaton, Ph.D.
Director
U.S. Geological Survey
Date
D. James Bake^PhJD.
Under Secretary and Administrator
National Oceanic and Atmospheric Administration
Date7
MRLC Memorandum of Understanding
-------
AnnGX 1
United States Department of the IntcriOHRLc consortium
Documentation Notebook
May 1995
GEOLOGICAL SURVEY
EROS
Sioux FaI&-^l30flJ&£$l>J7198
September 24, 1993
Dr. Arturo Silvestrinl
President, EOSAT
4300 Forbes Boulevard
Lanhan, HO 20706
Dear Dr. Silvestrini:
The intent of this letter 1s to obtain final concurrence on the terms for the
EOSAT United States Product Package, as required by the Federal nultl-agency
consortium identified in ny letter to you dated August 19, 1993, attached
hereto and incorporated herein.
The specific terms defined herein are in addition to those contained 1n the
USG5/EOSAT Basic Ordering Agreement (BOA). Modification 16 of the BOA allows
for the purchase of 430 Landsat TH scenes of the conterminous United States at
$3,256 per scene with an option for 100 additional scenes at no cost1 over
the conterminous United States. In the case of inconsistency between the
terns of the BOA and this letter, the terns of the BOA will be controlling.
except those unique provisions defined herein on data use.
The additional provisions of the USGS/EOSAT United States Product Package that
require our mutual concurrence are:
EOC acknowledges that proprietary ownership of the data delivered to EDC
under this agreement shall remain with EOSAT, until terminated by
contract, agreement, or law, and that these data constitute a special,
valuable, and unique asset of EOSAT. Data use shall be limited to EDC
and its Affiliates1.
All data shall be used for non-cossnercial purposes only.
EOC agrees that the initial order will consist of 430 scenes to be
ordered by November 15, 1993. EDC has prepared, for EOSAT, a list of
the 430 Landsat TH scenes, incorporated herein by reference, that
represents one-tine coverage of the conterminous U.S. A subsequent
order for the 100 scenes will be placed prior to September 30, 1994.
1 Scene* at no cost znaani no per scene
16 of Uui BCUL Shipptne cata xhifl b« a* a;
except far ^«f» dalirary a* defined in Modification
pecdfiad In tfco BOA.
Affiliate neons may jwnja conducting MRLC Raosaich within tha acopa of a relationship with
EOC which, la i4*tijje*i jjy n rj\fiji-f ^ Koploysunc fln&ncial support umi^onant, or wiltum
cooporativB """"r*h •qr«-Tn^"*l provided:
— Affiliates daitjamtm tn writing- tha nature, scope, and nqoirenianta of tha raaearch acthrtry,
- They agraa to publkh fa open Ufenton the result! of such roeuch.
— They agroa not to oaa (ha data or nnanhaccod products derived from it fcr any conunardal
-------
MRLC Consortium
Documentation Notebook
May 1995
EOSAT agrees to ship all data directly to EDC on 8 mm cartridges and EDC
agrees to return prepaid these 8 nxa cartridges after data ingest/
verification.
EOSAT agrees to best effort for rapid data delivery, with approximately
SO scenes per month provided to EDC.
EDC will prepare and execute an Affiliate User Agreement with the
agencies having access to the data. This agreement will define terms
and conditions regarding data use, publishing requirements, etc..
similar to those contained in the HASA/EOSAT Data Grant Program
Memorandum of Agreement.
At such time as EOSAT issues an across-the-board price reduction from
the published price schedule, whether Federal or commercial, the HRLC
Project shall automatically be offered the same price, for any scene not
yet delivered, if such price is less than the effective price of all
scenes herein offered.
After identification of scene ID by EDC, EOSAT also agrees to allow at no
cost, up to 150 scenes previously purchased to be used 1n this HRLC project,
subject to the same terms and conditions defined herein. All such scenes must
be identified by September 30, 1994. These scenes are not a part of the EOSAT
United States Product Package purchase. _
If these terns and conditions are acceptable, please sign/concur, return the
original letter, and I will forward a copy to your office.
Again, we appreciate EOSAT's help, specifically that of David Edwards and Jin
Love, in bringing this program forward. David and Jim demonstrated a real
comaitflient to the HRLC program and went out of their way to help settle a
number of issues while they were attending the Pecora Conference.
Sincerely,
Donald T. Lauer
Chief, EROS Data Center
Attachments
CONCUR:
-o VfrA^-
DM ARTURO SILFESTRINI
PRESIDENT & C.i.O.
EOSAT
-------
MRLC Consortium
Documentation Notebook
January, 1994
SECTION 3
TM SCENE PURCHASE AGREEMENT
This section contains the final form of the agreement between
the EROS Data Center and EOSAT for the purchase of the Thematic
Mapper scenes. This purchase was implemented as a modification
(Modification 16) to the Basic Ordering Agreement (BOA) between the
EROS Data Center and EOSAT, with an accompanying letter of
concurrence between EROS and EOSAT. Also to be included in this
section is a copy of important clauses within the BOA which apply
to the use and distribution of the TM data purchased for the MRLC
Consortium.
-------
GOT- 6-93 WED 14=56 EOSAT 13017319148 P 0,
.. united States Department of the interior MRLC consortiujn" "
Documentation Manual
GEOLOGICAL SURVEY
EROS
Sioux FaUflS? J 98
January, 1994
September 24. 1993
Dr. Arturo Silvestrini
President, EOSAT
4300 Forbes Boulevard
Lanham, HO 20706
Dear Dr. Silvestrini:
The intent of this letter is to obtain final concurrence on the terms for the
EOSAT United States Product Package, as required by the Federal multi-agency
consortium identified in my letter to you dated August 19, 1993, attached
hereto and incorporated herein.
The specific terms defined herein are in addition to those contained in the
USGS/EOSAT Basic Ordering Agreement (BOA). Modification 16 of the BOA allows
for the purchase of 430 Landsat TM scenes of the conterminous United States at
$3,256 per scene with an option for 100 additional scenes at no cost1 over
the conterminous United States. In the case of inconsistency between the
terms of the BOA and this letter, the terms of the BOA will be controlling,
except those unique provisions defined herein on data use.
The additional provisions of the USGS/EOSAT United States Product Package that
equire our mutual concurrence are:
EDC acknowledges that proprietary ownership of the data delivered to EDC
under this agreement shall remain with EOSAT, until terminated by
contract, agreement, or law, and that these data constitute a special,
valuable, and unique asset of EOSAT. Data use shall be limited to EDC
and its Affiliates'.
AU data shall be used for non-commercial purposes only.
EDC agrees that the initial order will consist of 430 scenes to be
ordered by November 15, 1993. EDC has prepared, for EOSAT, a list of
the 430 Landsat TM scenes, incorporated herein by reference, that
. represents one-time coverage of the conterminous U.S. A subsequent
order for the 100 scenes will be placed prior to September 30, 1994.
1 Scenes at no coat means no per scene costs, except for data delivery as defined in Modification
16 of the BOA. Shipping costs shaH be a* specified in the BOA.
1 Affiliate means any parson conducting MRLC Research within the scope of a relationship with
EDC which is defined by contract, employment, financial support arrangement, or written
cooperative research agreement, provided:
- Affiliates designate tn writing the nature, scope, and requirements of the research activity,
- They agree to publish in open literature the results of such research,
— They agree no( lo use the data or unenhanced products derived from It for any commercial
purpose.
-------
MRLC Consortium
Documentation Manual
January, 1994
C
COSAT agrees to ship all data directly to EDC on 8 mm cartridges and EDC
agrees to return prepaid these fi Dm cartridges after data Ingest/
verification.
EOSAT agrees to best effort for rapid data delivery, with approximately
80 scenes per month provided to EDC.
EDC will prepare'and execute an Affiliate User Agreement with the
agencies having access to the data. This agreement will define terms
and conditions regarding data use, publishing requirements, etc.,
similar to those contained in the NASA/EOSAT Data Grant Program
Memorandum of Agreement.
At such time as EOSAT issues an across-the-board price reduction from
the published price schedule, whether Federal or commercial, the HRLC
Project shall automatically be offered the same price, for any scene not
yet delivered, if such price is less than the effective price of all
scenes herein offered.
After identification of scene ID by EDC, EOSAT also agrees to allow at no
cost, up to 150 scenes previously purchased to be used In this HRLC project,
subject to the same terms and conditions defined herein. All such scenes must
be identified by September 30, 1994. These scenes are not a part of the EOSAT
United States Product Package purchase*
If these terns and conditions are acceptable, please sign/concur, return the
original letter, and I will forward a copy to your office.
Again, we appreciate EOSAT's help, specifically that of David Edwards and Jim
Love, in bringing this program forward. David and Jim demonstrated a real
commitment to the HRLC program and went out of their way to help settle a
number of issues while they were attending the Pecora Conference.
Sincerely,
Donald T. Lauer
Chief, EROS Data Center
Attachments
CONCUR:
JRO SILVESTRINI
PRESIDENT & C.t.O.
EOSAT
-------
-------
OCT- 6.-9S WED 14
EOSAT 1^017-19148
United States Department ol tne interior
P 04
OC 8-18
GEOLOG/CAI SURVEY
EROS Dau Center
Sioux FaUs. South Dakota 57198
'MRLC Consortium
Documentation Manual
January, 1994
August 19. 1993
Dr. Arturo Silvestrini
President. EOSAT
4300 Forbes Boulevard
Unham, HD 20706
Dear Dr. Silvestriniz
It is a pleasure to work with you concerning the development of the EOSAT
United States Product Package.
As you know, six Federal programs having similar remote sensing and research
needs are collaborating on the development of a Multi -Resolution Land
Characteristics (HRLC) Monitoring System which will provide a capability for
broad-based research on existing and future conditions of physical and
biological resources of the United States. The HRLC user group has a need for
Landsat coverage of the conterminous United States, and has agreed that in
exchange for EOS AT 's cooperation in making this data available, that the data
will be used only by the following agencies and for the following programs:
LGEMCV
U.S. Geological Survey (USGS)
U.S. Fish and Wildlife Service
National Oceanic and Atmospheric
"Administration
Environmental Protection Agency (EPA)
PROGRAM
Land Characteristics and National
Water-Quality Assessment (NAWQA)
Gap Analysis (GAP)
CoastWatch Change Analysis (C-CAP)
Environmental Monitoring and Assessment
(EHAP) and North American Landscape
Characterization (NALC)
We appreciate EOSAT 's help in bringing this program to fruition and we are
looking forward to a mutually rewarding relationship in the future.
Sincerely,
DonaTtHrrTauer
Chief. EROS Data Center
-------
OCT- 6-S<3 WEL 14:52 EOSAT
13Q1751S14S
P .
effective Date:
MRLC Consortium
Documentation Manual
iu«ry, 1994
BOA No.
Modification No. 16
' BASIC ORDERING AGREEHENT
BETWEEN
THE U.S. GEOLOGICAL SURVEY
AND
EARTH OBSERVATION SATELLITE COMPANY
parties.
....„,.— .— .,
(MRLC) Letter of Agreement between the
B. The preselected scenes are as specified 1n the HRLC Letter of Agreement
referenced herein.
C, All other terms and conditions contained 1n the original Agreement as
previously modified remain unchanged and 1n full force and effect.
D. No.additional funds are obligated as a result of this modification. Funding
there/pre remains unchanged.
lim COMPANY U.S. GEOLOGICAL SURVEY
By; ^j2^Qt*x
AWURO hlVESTRWl
k C.R.O.
Teresa M.
Title; Contracting Officer
Date: 9/H/93
-------
MRLC Consortium
Documentation Notebook
January, 1994
SECTION 4
TM SCENE SELECTION
This section contains the documentation on the Thematic Mapper
scene selection process. The section is divided into 3
subsections:
4.1 Scene Selection Criteria
4.2 Scene Selection Process Flow Diagram
4.3 Scene Selection and Order Status
TM scene selection has been an extended process within the
MRLC Consortium driven both by the information needs of the
participating programs, and the quality of the available scenes
within the specified time frame of 1992 +/- 1 year. An iterative
process has developed since the initial selection of scenes in June
1993. Numerous lists have been produced and circulated indicating
the status of scene selection at a given time. As scene selection
has remained a highly dynamic process most of these lists are not
valid. The list contained in Section 4.3 is the most up-to-date
list available at the time of preparation. The GAP, NAWQA, and C-
CAP programs have taken the lead in defining scene selection
criteria appropriate to each of the respective programs, and
selecting the actual scenes to be ordered. The EDC has taken a
lead role in evaluating the quality of the selected scenes, and
coordinating the re-selection of scenes for which the original
selections proved unacceptable due to cloud coverage or digital
quality. EOSAT has been cooperating in the efforts to evaluate
selected scenes.
-------
4.1 Scene Selection Criteria
MRLC Consortium
Documentation Notebook
January, 1994
C-CAP
The criteria for scene selection was developed in the
report entitled NOAA CoastWatch Change Analysis Project -
Guidance for Regional Implementation (J.E. Dobson, E.A.
Bright,and others, 1993).
Seasonal criteria for scene selection are region-
specific, with scenes selected to coincide with peak
biomass condition, and other regional considerations.
- C-CAP obtained seasonal preferences from
participating regional experts.
Selected scenes were prescribed to be at the lowest tide
possible, preferably within 2 feet of mean low tide.
GAP
NAWQA
State coordinators were provided with a list of scenes
from 1/91 to 5/92 under the initial scene quality
parameters (see below). Cooperators provided preferred
scenes corresponding to :
a) peak growing season
b) non-peak growing season
These selections were subsequently used to guide seasonal
considerations in scene selection.
The following information was used to identify periods
during the growing season when agricultural land use
could best be discriminated from other land cover:
1. classification of major crop groups based on the
1987 Census of Agriculture,
2. crop phenology information for 10 majo crops,
3. region-specific typical planting and harvesting
dates for major crops as developed by U.S. Dept. of
Agriculture, and
4. 1991-92 bi-monthly map composites of vegetation
greenness derived from AVHRR data.
-------
MRLC Consortium
Documentation Notebook
January, 1994
SCENE QUALITY
o An initial scene list was generated from a data dump of
EOSAT resources, with the following parameters:
1. cloud coverage = 0/1/2
2. no data quality specifications
3. time frame - 1/91 - 5/92 (based on scene
availability from EOSAT)
o Following the first visit by EDC staff to EOSAT to review
initial selected scenes, a second data dump from EOSAT
resources was made with the following parameters
1. cloud coverage «= 0/1
2. data quality -= 8/9
3. time frame -= 1/91 through 1993 (following EOSAT
agreement to accelerate scene availability for MRLC
purchase)
MULTITEMPORAL SCENE SELECTION
o Three types of regions were selected for multitemporal
(2-scene) coverage:
1) eastern deciduous forest biome
2) agricultural regions
3) selected coastal region
o Multitemporal pairs were selected to be in consecutive
seasons.
-------
MRLC Consortium
Documentation Notebook
January, 1994
4.2 Scene Selection Process Flow Diagram
The attached process flow diagram is intended to demonstrate
the nature of the scene selection process as it has been
implemented to date. Because of the dynamic nature of the
selection process, this diagram is intentionally generalized in its
nature, and shows only the primary processes that occurred. A
finalized version of this diagram will be included in this section
once all scenes have been ordered.
-------
MRLC Consortium
Documentation Notebook
January, 1994
CO O
CD "(J
Ill =
(O (O O Q-
LJJ
O
O
DC
QL
O
LU
_J
LU
C/)
CO
0>
CO
eg
c\j
(JO i
OLD LL
-------
MRLC Consortium
Documentation Notebook
January, 1994
Oiu u_
ill
-------
MRLC Consortium
Documentation Notebook
April 1994
4.3 Scene Selection and Order Status
The attached scene selection and ordering list was prepared and provided by Paul
Severson of Hughes STX, EROS Data Center on 3/25/94. This is the most up-to-date listing
available at the time of preparation of this notebook. The reader is directed to Mr. Severson
for information on the status of ordered scenes, including their receipt at the EDC or their
rejection following a quality review of all bands by EOSAT and the EDC, not indicated in the
attached list. This list will be updated on a regular basis until all scenes have been ordered and
received; a final listing will be included in this section at that time.
-------
MRLC Consortium
Documentation Notebook
February 1995
Path
10
11
11
11
11
11
11
11
12
12
12
12
12
12
12
12
12
13
13
13
13
13
13
13
14
14
14
14
14
14
14
14
14
14
14
14
14
15
15
15
15
15
Row
29
27
28
28
29
29
30
31
27
28
28
29
29
30
30
31
31
29
29
30
30
31
31
32
29
29
30
31
32
32
33
34
34
35
35
36
36
29
30
30
31
31
Date
05/26/1992
06/16/1991
07/02/1991
10/11/1993
04/15/1992
10/11/1993
09/04/1991
09/04/1991
06/07/1991
05/22/1991
09/16/1993
06/12/1993
09/16/1993
06/12/1993
08/15/1993
04/25/1993
09/27/1991
07/02/1992
10/06/1992
07/02/1992
10/06/1992
05/02/1993
10/06/1992
09/20/1992
05/09/1993
08/29/1993
05/09/1993
05/09/1993
03/17/1991
05/20/1991
05/04/1991
05/04/1991
08/10/1992
06/23/1992
10/13/1992
06/23/1992
10/13/1992
06/14/1992
05/29/1992
10/18/1991
06/14/1992
10/07/1993
Cloud
1
0
1
0
0
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
0
0
0
1
1
1
0
1
0
1
0
0
0
0
Quality
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
Comments
eosat ? cloud cover-rejected once-needed it for multi temporal-ordered
again 7/14/94
found scene on CORE system. Nice fall scene
per eosat-pixel noise over ocean-told them ok-11/23
Northeast gap-GAP scene unusable-had to order from EOSATbecause
there was nothing else close to adequate
eosat ? cloud cover-told them to proceed-best available
eosat ? cloud cover-told them to proceed-best available
ordered as a replacement for 8/31/93 which was hazy
slight clouds in a&c
data grant scene
mike wanted multi here
data grant scene
mike wanted multi here - 01/19/94
mike wants this one-looks good
mike wants this one - looks good
15 32 06/17/1993 0
m wants this one - ok some snow in ql
only picked one-others looked unusable
no quality rating, no image - eosat sent image looks good
there is no quality rating, so cannot look at image - eosat sent image
looks good-per m try 4/23/93
data grant scene
data grant scene
a few clouds mostly in Canada - only 1 good pick here-eosat
questioned clouds told them to process it
NY gap
had a look at it on ftche - only one i could find that was good
probably only want one here considering available choices
gail pick-good on core-small cloud in nw corner-multitemp with
6/14/92-but GOOD looking scene
very slight clouds in q3-good scene!
-------
MRLC Consortium
Documentation Notebook
February 1995
15
15
15
15
15
15
15
15
15
15
15
15
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
32
33
34
34
35
35
36
37
37
41
42
43
30
31
32
32
33
33
34
34
35
35
36
36
37
37
38
39
40
41
41
42
43
30
30
31
31
32
32
33
33
34
35
35
36
36
37
37
10/20/1992 1
09/16/1991 0
04/11/1992 0
10/18/1991 0
05/16/1993 0
10/20/1992 0
10/18/1991 0
02/04/1991 1
10/18/1991 0
04/14/1993 1
04/14/1993 0
04/14/1993 1
05/20/1992 0
05/20/1992 0
06/24/1993 0
08/24/1992 0
03/01/1992 0
05/20/1992 0
05/20/1992 0
09/28/1993 0
03/01/1992 0
09/28/1993 0
03/01/1992 0
05/04/1992 0
07/07/1992 1
11/26/1991 0
05/07/1993 0
03/04/1993 0
03/17/1992 0
03/17/1992 0
12/14/1992 0
03/17/1992 0
03/17/1992 1
05/30/1993 0
10/02/1992 1
05/11/1992 0
10/02/1992 0
05/14/1993 0
10/02/1992 0
07/17/1993 0
10/02/1992 0
10/02/1992 0
05/11/1992 0
11/03/1992 0
05/11/1992 0
11/03/1992 0
04/28/1993 0
06/20/1992 0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
7
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
try to get another one
gmil pick-ok scene per eosat
looks good
usgs request fl
usgs request fl
usgs request fl eosat ? cloud cover-told them to go ahead - clouds
over water per requestor
only one good pick here
need to verify the quality of this one
had Suzi at EOSAT look at it-cloud free-7/14/94
mike wants to try three here
picked from the images good early summer
bit of clouds-not bad
data grant scene
FL gap- couldn't get GAP scenes-ordered this one
FL gap- couldn't get GAP scenes-ordered this one
data grant scene
FL gap- couldn't get GAP scenes-ordered this one
usgs request fl
Eosat ? pick-minor clouds-told them to proceed 12/02
best choice we have
minor clou in quad b
good looking scene
-------
MRLC Consortium
Documentation Notebook
February 1995
17
17
17
17
17
18
18
18
18
18
18
18
18
18
18
18
18
18
18
18
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
38
38
39
40
41
31
31
32
32
33
34
35
35
36
36
37
37
38
38
39
31
32
32
33
33
34
34
35
35
36
36
37
37
38
38
39
39
29
30
30
31
32
32
33
33
06/20/1992 0
11/22/1993 0
10/16/1991 0
03/24/1992 0
03/24/1992 0
04/22/1994 0
08/09/1993 0
04/22/1994 0
08/06/1992 0
08/06/1992 1
11/29/1993 1
06/06/1993 0
10/25/1992 0
04/19/1993 0
10/28/1993 0
06/06/1993 0
10/12/1993 0
04/16/1992 0
10/25/1992 0
11/29/1993 0
06/13/1993 0
08/29/1992 0
10/03/1993 0
OS/31/1994 0
10/03/1993 0
04/23/1992 0
10/03/1993 0
04/23/1992 0
09/30/1992 0
07/12/1992 0
11/17/1992 0
07/31/1993 0
10/03/1993 0
04/10/1993 0
10/03/1993 0
04/10/1993 0
10/03/1993 0
07/22/1993 0
05/16/1992 1
09/05/1992 0
05/16/1992 0
05/16/1992 0
10/21/1991 0
08/02/1991 0
10/21/1991 0
9
9
9
7
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
FL gap- couldn't get GAP tape-ordered this one
usgs request fl
looks good on core
clouds in quad d
terrible p/r to find coverage-some clouds in ql but best I could find
for any time
good scene
some haze in ql
gail wants this one-per eosat looks good
gail pick-looks pretty good-bit of clouds-eosat ? clouds decided to take
it anyway 04/11
looked very good as an image
Found on core-great looking scene
looked at on core 12/28-good scene
looked at it on core 12/28 slight clouds in q4
picked from images-good scene-ordered when 8/11/91 scenewas
rejected
good looking scene
couldn't see on core-per eosat looks fine 3/18/94
gail pick-looks good per eosat
good scene
good image
gail pick-looks good per eosat
good one
look at this one - good scene
good looking scene
picked from visual look at images, also there is a good 10/24/92 scene
not on list checkitout
check this one out - looks good very slight clouds
no quality rating - eosat sent image - looks good
-------
MRLC Consortium
Documentation Notebook
February 1995
20
20
20
20
20
20
20
20
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
23
34
34
35
36
36
37
38
39
28
29
30
31
31
32
32
33
33
34
34
35
36
36
37
37
38
38
39
39
40
28
28
29
29
30
30
31
32
33
33
34
34
35
36
36
37
38
39
40
28
08/02/1991 1
10/07/1992 0
11/06/1991 0
07/22/1993 0
11/11/1993 0
11/11/1993 0
11/11/1993 0
11/06/1991 0
06/11/1993 0
05/10/1993 0
05/10/1993 0
05/10/1993 0
08/30/1993 0
05/10/1993 0
08/25/1991 0
05/07/1992 0
09/12/1992 0
04/05/1992 0
10/14/1992 0
08/30/1993 0
01/16/1992 0
09/12/1992 0
02/03/1993 0
10/01/1993 0
01/16/1992 0
10/14/1992 0
05/07/1992 0
10/14/1992 0
10/12/1991 0
08/21/1993 0
10/05/1992 0
08/21/1993 0
10/05/1992 0
06/21/1994 0
10/05/1992 0
10/05/1992 0
10/05/1992 0
07/20/1993 0
10/05/1992 0
05/17/1993 0
10/24/1993 0
06/18/1993 0
03/09/1991 0
07/31/1991 0
10/21/1992 0
10/21/1992 0
10/05/1992 0
10/05/1992 0
05/05/1992 0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
need qua! rating - eosat sent image - looks good
look for summer of this pit
good scene a few small clouds
Replacement for 7/22/93 which was hazy
looks ok
minor clouds in al
good scene
gail pick • bit of clouds but ok
good no clouds
IN gap-not recieved checked core-GOOD scene-need it
look at this one per mike-pretty good slight cl
looks better • no cloud-lets go with 91 scene-very little difference
checked on core » looks good
not a bad scene bit of clouds on east edge
looked at all images these 2 look best
good scene
extension of miss delta in quad a-rest is mostly h2o
found this one 7/20/94 - need one more
very slight clouds in q4-s edge
a couple of scan line defects are apparent
luman request-per eosat-Iooks real good
luman request-per eosai-looks ok-few clouds at bottom of q3
good scene
check this one - get it if good- its good
good one
good scene
look at this one if ok change-its good
looks good on image-nothing else we saw is good at all
-------
MRLC Consortium
Documentation Notebook
February 1995
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
28
29
29
30
31
32
33
34
34
35
35
35
35
36
36
36
36
37
37
38
38
39
27
27
28
28
29
29
30
30
31
31
32
32
33
33
34
35
35
35
36
36
36
37
37
37
38
07/24/1992 0
05/05/1992 0
07/24/1992 1
05/05/1992 0
09/10/1992 0
10/12/1992 0
10/1271992 0
07/06/1991 0
10/15/1993 0
04/03/1992 0
04/22/1993 0
07/27/1993 0
10/12/1992
04/22/1993 0
05/05/1992
07/27/1993 0
10/12/1992 0
04/22/1993 0
09/29/1993 0
04/22/1993 0
09/29/1993 0
09/29/1993 0
07/13/1991 1
10/03/1992 0
07/31/1992 0
10/03/1992 0
07/31/1992 0
10/03/1992 0
07/31/1992 0
10/03/1992 0
05/15/1993 0
10/03/1992 0
06/13/1992 0
10/03/1992 0
08/14/1991 0
10/17/1991 0
10/03/1992 0
04/24/1991 2
06/16/1993 0
10/03/1992 0
04/26/1992 0
10/19/1992 0
12/25/1993 0
04/26/1992 0
10/19/1992 0
12/25/1993 0
04/26/1992 0
9
9
9
9
9
9
9
9
9
9
9
9
8
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
checked core-looks good
this was our original pick, ordered now because of rejection of
10/12 scene- 02/04/94
ARgap
gail pick-looks ok per eosat
ARgap
good scene
ARgap
few clouds in q4
ARgap
good looking scene
great image!
picked sub from core search
a few clouds on land area-not too bad-Eosat ? pick-
proceed-12/02
a few clouds scattered not bad acouple of bad lines
checked core-looks good
little bit of clouds central area
checked core-looks good
perhaps a couple of bad lines lower half
checked core-good scene
mike & gail wanted this on instead of 4/24/91
mike & gail wanted this one instead of 10/17/91
checked core-bit of clouds in q24 but not bad
luman request-per eosat-looks good
AR gap- in and ok
good scene
AR gap - in and ok
AR gap - in and ok
AR gap - in and ok
AR gap - in and ok
AR gap - in and ok
couldn't see on core-per eosat looks fine-3/18/94
told them to
-------
MRLC Consortium
Documentation Notebook
February 1995
24
24
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
38
39
27
28
28
29
29
30
30
31
31
32
32
33
33
34
34
35
35
35
36
36
36
37
37
38
39
39
40
26
27
27
28
28
29
29
30
31
32
32
33
34
34
35
35
35
36
36
10/19/1992 0
10/19/1992 0
09/24/1992 0
05/06/1993 0
09/08/1992 0
05/19/1992 0
09/08/1992 0
05/19/1992 0
09/08/1992 0
05/03/1992 0
09/08/1992 0
05/03/1992 0
09/24/1992 0
05/03/1992 0
09/24/1992 0
05/03/1992 0
09/24/1992 0
05/03/1992 0
09/24/1992 0
09/27/1993 0
04/15/1991 1
07/25/1993 0
10/05/1990 0
04/15/1991 1
07/25/1993 0
07/06/1992 0
02/10/1991 1
09/22/1991 0
07/06/1992 0
05/10/1992 0
05/10/1992 0
08/28/1991 0
05/13/1993 0
10/01/1992 0
05/10/1992 0
10/01/1992 0
05/13/1993 0
10/01/1992 0
03/23/1992 0
10/01/1992 0
04/06/1991 0
03/23/1992 0
08/14/1992 0
04/27/1993 0
07/13/1992 0
10/01/1992 0
03/07/1992 0
07/13/1992 0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
0
9
9
9
9
see if we have enough land area to be worth it -it would be worth it
few clouds q34 not very bad
data gramt scene-good looking scene-we want it
see what it looks like-looks good
would realy like to see this one -can't - checked mss fiche - good
cannot see image - looked at mss fiche - good
AR gap - in and ok
AR gap - in and ok
AR gap - in and ok
good scene
check this one out-looks really good
AR gap - in and ok
fairly good size cloud in quad b
some clouds in quad c-not too bad
look at it - looks great
look at this - looks great!
found on core when 8/14/91 scene was rejected
good scene
look at this one - looks great
checked core-looks good
look at it-looks good
good scene
change to this one of ok - its fine
change to this if ok-this one is good
cannot see image - mss fiche • good scene
AR gap - in and ok
picked from visual look at all available images
AR gap - looked at it good scene - in and ok
-------
MRLC Consortium
Documentation Notebook
February 1995
26
26
26
26
26
26
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
37
38
39
40
41
42
26
27
27
28
28
29
30
30
31
32
32
33
33
34
34
35
35
36
37
38
38
39
40
41
41
42
26
26
27
27
28
29
30
31
32
33
33
34
35
36
37
38
03/07/1992 0
11/0271992 0
11/02/1992 0
11/02/1992 0
11/02/1992 0
11/02/1992 0
08/19/1991 0
06/16/1991 1
09/20/1991 0
06/16/1991 0
09/20/1991 0
09/20/1991 0
05/01/1992 0
09/22/1992 0
09/04/1991 0
05/01/1992 0
08/21/1992 0
03/14/1992 0
08/21/1992 0
03/14/1992 0
09/22/1992 0
03/14/1992 0
08/21/1992 0
03/14/1992 0
03/14/1992 0
03/14/1992 0
08/05/1992 1
10/24/1992 0
12/14/1993 0
04/18/1993 0
07/23/1993 0
07/23/1993 0
05/11/1993 0
08/10/1991 1
05/11/1993 0
07/30/1993 0
10/02/1993 0
08/26/1991 0
08/26/1991 0
08/26/1991 0
08/26/1991 0
04/09/1993 0
08/26/1991 0
07/30/1993 0
07/09/1991 0
05/08/1992 0
08/15/1993 0
05/11/1993 0
9
9
9
9
9
9
9
9
9
8
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
looked it toss fiche - good scene
the only other one i like here is the 10/1 scene & it appears that there
would be very little difference
data grant scene-good looking scene!
looked at fiche - good scene
checked core-looks good
need to see the image • looked at tm fiche - good
checked core-looks good
look at this - good one
bit of popcorn in quad d
this looks good
a few clouds but pretty good summer scene
slight clouds in corner of q2
good scene
good scene
eosat screened scene &. provided an image - looks good
good looking scene
good scene-out some clouds q2
good scene - pre flood
data grant scene
good i
no clouds at all oa this-its better than 91 scene
good scene
good scene
-------
MRLC Consortium
Documentation Notebook
February 1995
28
28
28
29
29
29
29
29
29
29
29
29
29
29
29
29
29
29
29
29
29
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
31
31
31
31
31
31
31
31
31
31
31
39
40
41
26
26
27
27
28
29
29
30
31
32
33
34
35
36
37
38
39
40
26
27
27
28
29
30
31
32
33
34
35
36
37
38
39
40
26
27
28
29
30
31
32
33
34
35
36
03/08/1993 0
05/11/1993 0
08/10/1991 0
05/15/1992 0
09/04/1992 1
06/06/1994 0
09/23/1993 0
08/17/1991 0
05/15/1992 0
08/19/1992 0
08/19/1992 0
08/19/1992 0
08/19/1992 0
08/01/1991 0
08/22/1993 0
08/22/1993 0
08/22/1993 0
08/22/1993 0
08/22/1993 0
09/04/1992 1
08/06/1993 0
05/06/1992 1
05/06/1992 0
08/10/1992 0
08/10/1992 0
05/25/1993 0
07/28/1993 0
07/28/1993 0
07/28/1993 0
07/25/1992 0
07/28/1993 0
07/28/1993 0
08/26/1992 1
08/13/1993 0
07/28/1993 0
07/28/1993 0
07/28/1993 0
07/14/1991 1
06/12/1991 0
06/12/1991 0
07/14/1991 0
07/14/1991 0
07/14/1991 0
07/14/1991 0
07/30/1991 0
07/30/1991 0
06/12/1991 0
09/21/1993 0
9
8
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
no clouds
good alternate
looks good
if ok add-Iooks good
Thi« should be the last of the 530 scenes! -pas 8/3/94
looked at it on core 12728 - good scene
looked si mss fiche - bit of clouds in quad a
gall pick - ok per eosat
gail pick - ok per eosat
gail pick - ok per eosat
looks good
eosat sent bitmap disk-looks good-slight clouds in quad 1
eosat sent bitmap-looks very good
looked at all i could this and 12/92 all we had
if ok add-looks good-digital problems with 7/23/91 scene so requested
this as a sub
looked at all images of area-tbis one is best
real good scene
good scene
need to find it
some clouds through quads b&d
looks good
looks good
looks good
no rating - no image - received image-lookd good
no rating, no image - received image-looks good
-------
MRLC Consortium
Documentation Notebook
February 1995
31
31
31
31
32
32
32
32
32
32
32
32
32
32
32
32
32
32
33
33
33
33
33
33
33
33
33
33
33
33
33
33
33
34
34
34
34
34
34
34
34
34
34
34
34
34
35
37
38
39
40
26
27
28
29
30
31
32
33
34
35
36
37
38
39
26
27
28
29
30
31
32
33
34
35
36
36
37
37
38
26
27
28
29
30
31
32
33
34
35
36
37
38
26
09/21/1993 0
04/27/1992 0
04/27/1992 0
04/27/1992 0
07/23/1992 1
08/08/1992 1
08/27/1993 1
09/09/1992 0
09/09/1992 0
09/09/1992 0
09/09/1992 0
08/06/1991 1
08/11/1993 0
08/11/1993 0
09/28/1993 0
09/28/1993 0
03/31/1991 0
03/31/1991 0
07/17/1993 0
05/27/1992 0
08/18/1993 0
08/15/1992 0
08/15/1992 0
08/15/1992 0
08/15/1992 0
07/06/1992 0
09/03/1993 0
09/03/1993 0
04/09/1992 0
09/03/1993 0
07/06/1992 1
09/03/1993 0
09/03/1993 0
08/09/1993 0
09/23/1992 0
08/09/1993 0
08/09/1993 0
08/09/1993 0
09/21/1991 0
09/23/1992 0
09/23/1992 0
09/26/1993 0
09/26/1993 0
09/26/1993 0
09/26/1993 0
06/11/1992 0
09/30/1992 0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
we had digital problems with it-remade and sent back by EOSAT on
7/29/94
little popcorn in quad b
good scene summer in ND some clouds west ride
not bad but there is some popcorn
good scene, there is an 8/6/91 i would like to see its not on this list
or no quick look for it
replacemant for 8/6/91 which had a bit to many clouds
eoamt sent bitmap-looks real good
looked at it on core 12/28-good scene
check clouds compare crop patterns with 8/22- clouds aren't too bad
good looking image
good looking scene
a few popcorn clouds-not bad at all though
best looking of all i could look at of the area
very slight clouds q34
good looking scene-no clouds at all
data grant scene
good scene
this looks really good
good scene
good scene
mis and the 9/23/92 scene are good-went with 91
checked images this one looks best
looked on core 12/28- real good scene
data grant scene
-------
MRLC Consortium
Documentation Notebook
February 1995
35
35
35
35
35
35
35
35
35
35
35
35
35
36
36
36
36
36
36
36
36
36
36
36
36
36
36
37
37
37
37
37
37
37
37
37
37
37
37
37
38
38
38
38
38
38
38
27
28
29
30
31
32
33
34
35
36
37
38
38
26
27
28
29
30
31
32
33
34
35
36
37
37
38
26
27
28
29
30
31
32
33
34
35
36
37
38
26
27
28
29
30
31
32
08/16/1993 0
08/16/1993 0
08/16/1993 0
08/16/1993 0
08/16/1993 0
08/11/1991 0
06/18/1992 0
06/13/1993 0
06/13/1993 0
06/13/1993 0
06/13/1993 0
05/23/1991 0
06/13/1993 0
08/07/1993 0
08/07/1993 0
08/07/1993 0
08/04/1992 0
08/23/1993 0
08/23/1993 0
08/23/1993 0
08/23/1993 0
08/23/1993 0
08/23/1993 0
08/23/1993 0
07/01/1991 0
10/26/1993 0
05/14/1991 0
08/14/1993 0
08/14/1993 0
08/27/1992 0
09/28/1992 0
08/27/1992 0
08/14/1993 0
08/14/1993 0
06/27/1993 0
06/22/1991 0
06/22/1991 0
06/22/1991 0
06/22/1991 0
06/22/1991 0
09/22/1993 0
09/22/1993 0
09/22/1993 0
09/22/1993 0
07/20/1993 0
07/20/1993 0
07/17/1992 0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
better -no clouds
compare with 9/28-this one looks good
good sceoe
good scene
need to look at this-did. its good
need to look at this-did. its good
looks good
looks good
very good looking
very good
good scene
data grant scene
Eosat ? cloud cover-check core again-its not to bad-best available-told
them to proceed - 2/17/94
looked at image-great looking scene
good scene slight cloudsl
good scene-very slight clouds
I had plugged in the wrong eosat id ordered 8/14/93. Found when
eosat ? the cloud cover of 8/14/image.
looks good
•ee what this looks like on this date-looks ok
good scene-slight ctouds on east edge
-------
MRLC Consortium
Documentation Notebook
February 1995
38
38
38
38
38
39
39
39
39
39
39
39
39
39
39
39
39
39
40
40
40
40
40
40
40
40
40
40
40
40
40
41
41
41
41
41
41
41
41
41
41
41
41
42
42
42
42
42
33
34
35
36
37
26
27
28
29
30
30
31
32
33
34
35
36
37
26
27
28
29
30
30
31
32
33
34
35
36
37
26
27
28
29
30
30
31
32
33
34
35
36
26
27
28
29
29
07/20/1993 0
09/22/1993 0
06/18/1993 0
06/18/1993 0
05/28/1991 0
09/10/1992 0
09/10/1992 0
09/10/1992 0
09/10/1992 0
05/05/1992 0
07/24/1992 0
07/27/1993 0
07/27/1993 0
07/27/1993 0
07/27/1993 0
07/27/1993 0
07/27/1993 0
07/27/1993 0
08/19/1993 0
07/31/1992 0
08/19/1993 0
07/31/1992 0
07/31/1992 0
08/03/1993 0
08/19/1993 1
08/19/1993 0
08/19/1993 0
08/19/1993 0
08/19/1993 0
08/19/1993 0
04/26/1992 0
09/27/1993 0
09/27/1993 0
09/27/1993 0
08/10/1993 0
06/04/1992 0
09/27/1993 0
08/26/1993 0
08/26/1993 0
08/26/1993 0
08/10/1993 0
08/10/1993 0
08/26/1993 0
10/04/1993 0
10/04/1993 0
06/14/1993 0
07/27/1991 0
07/29/1992 0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
looked at core 12/28-best scene available here
looked a core 12/28-slight clouds but good sceoel
Id gap- UTM - in and ok
MT gap-GAP scene unusable-bad to order new from EOSAT
because there wasn't anything else good enough
good scene
MT gap-unusable as gap scene-ordered from eosat-best available
ID gap - UTM - in and ok
a few clouds in q2-not bad
looks good either this or the 6/13/92 are fine
checked on core-looks good
ID gap - UTM - in and ok
looked at it on core-good scene
very alight clouds
good scene • better of two here in 93
good looking scene-minimal clouds
checked core-pretty good-small cloud right in middle
checked core-good scene
alight clouds
ID gap - UTM - in and ok
-------
MRLC Consortium
Documentation Notebook
February 1995
42
42
42
42
42
42
42
42
43
43
43
43
43
43
43
43
43
43
43
44
44
44
44
44
44
44
44
44
44
44
45
45
45
45
45
45
45
45
45
46
46
46
46
46
46
46
46
46
30
30
31
32
33
34
35
36
26
27
28
29
30
31
32
33
34
34
35
26
27
28
29
30
31
31
32
33
34
34
26
27
28
29
30
31
31
32
33
26
27
28
29
30
30
31
31
32
05/13/1993 0
07/29/1992 0
06/14/1993 0
06/14/1993 0
08/01/1993 0
07/16/1993 0
08/17/1993 0
05/29/1993 0
09/25/1993 0
08/08/1993 0
09/25/1993 0
07/02/1991 0
08/08/1993 0
08/08/1993 0
08/08/1993 0
08/24/1993 0
05/01/1992 0
07/20/1992 0
07/20/1992 0
05/11/1993 0
05/11/1993 0
05/11/1993 0
06/09/1992 0
10/13/1991 0
06/12/1993 0
10/13/1991 0
07/30/1993 0
10/15/1992 0
01/03/1993 0
06/12/1993 0
10/25/1993 0
09/18/1991 0
10/25/1993 0
08/03/1992 0
08/19/1992 0
08/19/1992 0
10/25/1993 0
08/06/1993 0
08/22/1993 0
08/10/1992 0
08/29/1993 0
08/10/1992 0
08/10/1992 0
07/09/1992 0
08/29/1993 0
07/09/1992 0
10/11/1991 0
07/12/1993 0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
ID gap - UTM - in and ok
gall pick - go with it
gall pick - per eocat looks ok
ice in high mnts
gail pick - per eosat-looks ok
We had digital problems with 9/09 scene this is a sub
looks good
Digital problems with 9/09 scene this is a sub
looks good
good scene-bit of ice comer of q2—
mike wants these two over the area-looks good
clouds around monterey bay
repl for GAP - best on core system-abit of ice
repl for GAP scene - best on core system
repl for GAP scene - best on core system
we were looking at June-its cloudy-this oct scene looks really good
looked at all images, oct is away from desired season but best looking
scene available
gail pick-per eosat looks ok
mike wants this one-again some clouds-not terrible
January but really clear over san francisco-thats rare
this one is good
repl for GAP scene - best on core system
repl for GAP scene - best on core system
found on core after data grant scene turned out cloudy
data grant scene
data grant scene
gail pick - per eosat looks ok
gail pick - per eosat looks ok
data grant scene-looks good
repl for GAP scene - best on core system
data grant scene
data grant scene
data grant scene
we need this one as of 12/14
data grant scene
good scene -no clouds at all
-------
MRLC Consortium
Documentation Notebook
February 1995
47
47
47
47
47
48
48
26
27
28
29
30
26
27
08/04/1993 0
09/16/1991 0
09/16/1991 0
07/30/1991 1
07/30/1991 0
09/09/1992 0
09/09/1992 0
9
9
9
9
9
9
good scene-eosat ? pick-a bit of clouds but I think wewant it anyway
we have one-look at this anyway-looks very good
•elected as a nib for rejected may 93 scene - 3/14/94
it for GAP-best on core system 2/23
check out this one-looks great
-------
MRLC Consortium
Documentation Notebook
January, 1994
SECTION 5
TM SCENE ACQUISITION AND ARCHIVING
This section contains information on the archiving of TM data
received from EOSAT. The section is divided into two sub-sections:
5.1 system flowsheet for archiving data upon receipt at EROS
Data Center
5.2 status sheet of scenes received from EOSAT
-------
5.1 EDC Scene Archive Flowsheet
MRLC Consortium
Documentation Notebook
January, 1994
The flowsheet: included in this section was prepared by the
EROS Data Center, and illustrate the process by which the TM scenes
ordered for the Consortium are received and processed for archiving
at the EDC.
-------
g
—SI
MRLC Consortium
Documentation Manual
January, 1994
si
CO i—M
co
CO
CO
&a
ea-
CO •
CO
CO
CO
OsS
-------
MRLC Consortium
Documentation Notebook
April 1994
5.2 TM Scene Acquisition and Archive Status
The attached sheets were prepared by the EDC and represent the status of all TM scenes
received at the EDC, including GAP holdings, effective 4/1/94. This list will be updated on a
regular basis until all scenes have been received and archived.
-------
MRLC Consortium
Documentation Notebook
April 1994
MRLC Consortium TM Scenes
EROS Data Center Archive Update
Effective Date: 4/1/94
NOTES
ACQ_DATE «= The original acquisition date for this path row.
RECEIVED •= The date this media was received at EROS Data Center.
CONTRIB - Agency that sent the media:
ny-st : New York State
new eng : University of Massachusetts
eosat : EOSAT Corp.
wyom gap : University of Wymoing
louisana : Louisana
Idaho : University of Idaho
ark : University of Arkansas
TYPE * Level of processing on the image:
p : basic data, no alterations
s : systematic corrections on image
t: terrain corrections on image
u : unuseable
ARCHIVE - The date the copy was generated and archived at EROS Data Center.
RETURN - The date the original was returned to the Agency that sent it
PATH ROW ACQ_DATE RECEVIED MEDIA CONTRIB TYPE ARCHIVE RETURN
10
10
10
11
11
11
11
11
11
11
11
11
11
11
11
11
12
12
12
12
12
28
28
29
27
27
27
28
28
28
29
29
29
29
29
30
31
27
27
27
28
28
19910625
19910625
19920526
19900816
19900816
19910616
19910616
19910616
19931011
19910702
19910702
19910718
19910718
19931011
19910904
19910904
19910607
19910607
19910607
19910607
19910607
19931220
19931228
19940214
19931220
19931228
19940214
19931220
19931228
19940311
19931220
19931228
19931220
19931228
19940311
19931202
19931202
19931220
19931228
19940318
19931220
19931228
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
mi^l
H1FII
mm
mm
pun
mm
flU&
fT^t|}
mm
fnni
nun
wnni
mm
mm
mm
mm
mm
linn
mm
mm
mm
new eng
eosat
eosat
new eng
eosat
eosat
new eng
eosat
eosat
new eng
eosat
new eng
eosat
eosat
eosat
eosat
new eng
eosat
eosat
new eng
eosat
8
8
8
S
8
»
8
8
S
S
s
s
s
s
p
p
p
p
p
p
p
19931230
19940107
19940304
19931230
19940107
19940107
19931230
19940107
19940314
19931230
19940107
19931230
19940107
19940314
19931203
19931203
19931230
19940107
19940323
19931230
19940107
19940103
19940110
19940310
19940103
19940110
19940110
19940103
19940110
19940328
19940103
19940110
19940103
19940110
19940328
19931203
19931203
19940103
19940110
19940328
19940103
19940110
-------
MRLC Cttuortium
meaUtion Notebook
April 1994
12
12
12
12
12
12
12
12
12
12
12
13
13
13
13
13
13
13
14
14
14
14
14
14
14
14
14
14
14
14
14
14
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
28
29
29
29
29
30
30
30
30
31
31
29
30
31
31
31
31
32
29
29
29
30
30
31
31
32
32
33
34
35
36
36
29
29
30
30
31
31
32
32
32
33
34
34
35
36
37
37
41
42
19930916
19910607
19910607
19930612
19930916
19910607
19910607
19930612
19930831
19910607
19910927
19921006
19921006
19910513
19910817
19910817
19921006
19920920
19910520
19930509
19930829
19910520
19930509
19910520
19930509
19910317
19910520
19910504
19910504
19921013
19920623
19921013
19920529
19920614
19911018
19920529
19910511
19920614
19921020
19921020
19930617
19910916
19911018
19920411
19921020
19911018
19910204
19911018
19930414
19930414
19940225
19931220
19931228
19940214
19940225
19931220
19931228
19940214
19940318
19940131
19931202
19940225
19940225
19940131
19911220
19931228
19931101
19931217
19940131
19940225
19940401
19940131
19940225
19940131
19940225
19931112
19931112
19931112
19931210
19940401
19931228
19931228
19940131
19931130
19931112
19940131
19940131
19931119
19931101
19940201
19931112
19931130
19931101
19931101
19940121
19940121
19931228
19940311
19940204
19940204
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
IfUtt
OHO
fpyyi
iyiti|
joni
ODD
XDID
IHtff^
mm
Hym
^QQJ
ftim
mm
moi
moi
ffltn
mm
BDfll
QJOl
mm
JOfll
mm
mm
CDfll
mm
OBfll
mm
mm
moi
flUD
Olfll
JDfll
om^
flOD
OUfl
flCUD
mm
mm
yum
XDDA
OUpi
mm
HUH
mm
mm
nun
mm
mm
mm
mm
eosat
new eng
eosat
eosat
eosat
new cog
eosat
eosat
eosat
ny-st
eosat
eosat
eosat
ny-st
neweng
eosat
eosat
eosat
ny-st
eosat
eosat
ny-st
eosat
ny-st
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
ny-st
eosat
eosat
ny-st
ny-st
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat .
eosat
s
s
s
8
8
S
S
P
S
8
8
P
S
8
8
P
8
8
P
8
P
8
8
S
8
8
8
8
8
P
8
8
P
P
8
8
8
8
S
8
S
8
S
S
S
s
s
19940228
p 19931230
19940107
19940308
19940228
p 19931230
19940107
19940307
19940320
19940202
19931203
19940227
19940303
19940202
p 19931230
19940107
19931111
19931228
19940202
19940303
19940406
19940202
19940303
19940202
19940227
19931116
19931118
19931116
19931213
19940406
19931229
19940103
19940202
19931130
19931116
19940202
19940202
19931119
19931111
19940202
19931118
19931130
19931111
19931111
19940202
19940122
19940103
19940325
19940210
19940307
19940303
19940103
19940110
19940328
19940328
19940103
19940110
19940328
19940328
19940203
19931203
19940228
19940328
19940203
19940103
19940110
19931119
19931228
19940203
19940328
19940408
19940203
19940328
19940203
19940228
19931119
19931119
19931119
19931227
19940408
19940103
19940104
19940203
19931203
19931119
19940203
19940203
19931119
19931119
19940209
19931119
19931203
19931119
19931119
19940209
19940303
19940104
19940328
19940303
19940328
-------
MRLC CoMoftnim
Documenutioo Notebook
April 1994
15
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
17
18
18
18
18
18
18
18
18
18
18
19
19
19
19
19
43
30
30
31
31
32
33
33
34
34
35
36
36
37
38
40
41
42
43
30
30
31
31
32
33
34
35
35
36
36
37
38
39
40
41
31
32
33
34
35
36
36
37
38
39
31
32
33
34
35
19930414
19910822
19920520
19920520
19930624
19930624
19920301
19920520
19920520
19930928
19920301
19920301
19920504
19911126
19930507
19920317
19920317
19920317
19920317
19920612
19921002
19920511
19921002
19930514
19930717
19921002
19920511
19921103
19920511
19921103
19930428
19920620
19911016
19920324
19920324
19930809
19920806
19920806
19931129
19930606
19930419
19931028
19930606
19920416
19931129
19930613
19931003
19931003
19920423
19920930
19940214
19940131
19931112
19940311
19940131
19940113
19931130
19931130
19940113
19940401
19931228
19931228
19931230
19931228
19940121
19940325
19940325
19940325
19940128
19940131
19931210
19931130
19931210
19940113
19940110
19931210
19931228
19931228
19931228
19931228
19940121
19931230
19931217
19940325
19940214
19940121
19931130
19931130
19940214
19940113
19931228
19940311
19940113
19940113
19940325
19940128
19940128
19940128
19940225
19931217
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
f|U)fl
mm
mmi
yiwyi
DUO
DUD
HJfll
DUD
Q]Q2
HUD
Dini
1D1D
DUD
mm
pirn
fyiyii
ID^Q
OBOE
]Q]Q
mflj
flD^D
^002
^DDl
JQQ2
miD
mm
O1D1
OUD1
nun
DUD
mm
tMtn
iBim
yum
mm
Wim
mm
mm
mm
mm
mm
mm
mm
mm
mm
trim
mm
mm
mm
mm
eosat
ny-st ]
eosat
eosat
ny-st j
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
ny«t
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat t
eosat t
eosat t
eosat f
eosat i
eosat s
eosat s
eosat s
eosat G
s 19940221
9 19940202
s 19931116
s 19940317
» 19940202
19940212
19931130
19931202
19940215
19940406
19940103
19940103
19940105
19931229
19940202
19940327
19940327
19940327
19940304
19940202
19931213
19931202
19931213
19940214
19940127
19931213
19931229
19931231
19931231
19931228
19940216
19940105
19931228
19940327
19940226
19940208
19931130
19931130
19940303
19940202
19940103
19940322
19940211
19940203
19940326
19940220
19940220
19940221
19940228
19931228
19940228
19940203
19931119
19940328
19940203
19940303
19931203
19931203
19940303
19940408
19940104
19940104
19940110
19940103
19940209
19940328
19940328
19940328
19940328
19940203
19931227
19931203
19931227
19940303
19940209
19931227
19940103
19940104
19940104
19931228
19940303
19940110
19931228
19940328
19940328
19940228
19931203
19931203
19940328
19940209
19940104
19940328
19940227
19940209
19940328
19940303
19940303
19940303
19940303
19931228
-------
MRLC Consortium
Documentation Notebook
April 1994
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
22
22
22
22
22
22
36
37
37
38
38
39
39
29
30
30
31
32
33
33
34
34
35
36
36
37
38
38
39
28
29
30
31
31
32
32
33
34
34
35
36
36
37
37
38
38
39
39
39
40
28
29
30
31
32
33
19920712
19930731
19931003
19930410
19931003
19930410
19931003
19930722
19920516
19920905
19920516
19920516
19910802
19911021
19910802
19921007
19911106
19930722
19931111
19931111
19930722
19931111
19911106
19930611
19930510
19930510
19930510
19930830
19910825
19930510
19920912
19920405
19921014
19930830
19920116
19920912
19930203
19931001
19920116
19921014
19920507
19921014
19930307
19911012
19930821
19921005
19921005
19921005
19921005
19921005
19940311
19940121
19940311
19940121
19940311
19940121
19940318
19931130
19931130
19931130
19931130
19940121
19931230
19931230
19940113
19940401
19931228
19940121
19940318
19940318
19940121
19940325
19931228
19931217
19940110
19931119
19940110
19940318
19931112
19931119
19931130
19931228
19931228
19940318
19931228
19931228
19940214
19940318
19940128
19940401
19931228
19931228
19940111
19931210
19931130
19931112
19931130
19931130
19940110
19931112
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
mm
mm
mni
Q]2D
mm
oim
DUD
fim|
ttitn
OBflB
flflttl
wtiitl
^tym
iiHH
mtn
BUD
OUD
mm
OttD
mm
OUD
OUD
IBOOi
DUD
ODOA
OVD
nun
OUD
mm
mm
OUD
nmy
mm
mm
mm
OUD
OUD
OUD
flUpi
mm
OUD
OUD
mm
mm
mm
mm
mm
nun
mm
mm
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat i
eosat i
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat i
eosat i
eosat i
louisaaa
eosat i
eosat i
eosat i
e tsat i
eosat i
eosat t
eosat i
s 19940317
s 19940203
s 19940317
s 19940224
5 19940319
5 19940202
s 19940319
s 19931201
s 19931201
i 19931202
s 19931201
19940216
19940105
19940105
19940211
19940405
19940103
19940217
19940320
19940320
19940207
19940327
19940103
19931228
19940127
19931119
19940115
19940321
19931118
19931119
19931201
19940103
19931231
19940321
19940103
19931231
19940227
19940320
19940214
; 19940405
s 19931231
5 19931228
p 19940112
i 19931213
; 19931201
i 19931116
i 19931202
; 19931202
i 19940118
i 19931116
19940328
19940209
19940328
19940228
19940328
19940209
19940328
19931203
19931203
19931203
19931203
19940303
19940110
19940110
19940303
19940408
19940104
19940303
19940328
19940328
19940303
19940328
19940104
19931228
19940209
19931119
19940209
19940328
19931119
19931119
19931203
19940104
19940104
19940328
19940104
19940104
19940303
19940328
19940303
19940408
19940104
19931228
19940114
19931227
19931203
19931119
19931203
19931203
19940303
19931119
-------
MRLC Conioctium
Documenutioo Notebook
April 1994
22
22
22
22
22
22
22
22
22
22
22
22
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
24
24
24
24
24
24
24
24
24
24
24
24
24
33
34
34
35
36
36
37
38
38
39
39
40
28
29
30
31
31
32
33
34
35
35
35
35
36
36
36
36
37
37
37
37
37
38
38
39
39
27
28
29
30
31
31
32
33
33
34
35
35
35
19930420
19930517
19931024
19930618
19910309
19910731
19921021
19921021
19930125
19921005
19930125
19921005
19920505
19920505
19920505
19920910
19920910
19921012
19921012
19910706
19920403
19921012
19930422
19930727
19920505
19921012
19930422
19930727
19920505
19921012
19930201
19930422
19930929
19930201
19930422
19921129
19930929
19910713
19920731
19920731
19920731
19921003
19930515
19921003
19910814
19911017
19921003
19910424
19921003
19930616
19931112
19940318
19940110
19940318
19931228
19940401
19940113
19940113
19940111
19940110
19940111
19931130
19931130
19931130
19931130
19940113
19940216
19940113
19940113
19931230
19940110
19940110
19940110
19931112
19940110
19940110
19931119
19931112
19940110
19940110
19940111
19931119
19940401
19940111
19940307
19940111
19940225
19931210
19940121
19940113
19940113
19931217
19931217
19940113
19931202
19931210
19931112
19940110
19940110
19931112
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
mm
QQQQ
m^Q
QUO
JQIQ
IHlfl
null
JQ]Q
HUH
IHlft
JBflA
]Q]Q
HUP
]QQJ
ffilD
QJQJ
1D1D
fllfll
moot
QUQ
mm
QJQJ
flQQJ
BQ1D
moj
m^Q
m^Q
QJQJ
QIQJ
BB1D
BJQ2
XD1D
B21Q
DUO
turn
OUD
iiiiii
HHII
Bftm
nun
QIIQ
fll^p
miy^
mm
mm
mm
mm
mm
mm
mm
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
louisana
eosat
louisao*
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
uof aik
uof ark
eosat
eosat
uof ark
oof uk
eosat
eosat
uof ark
uof ark
louisan*
eosat
eosat
louisana
eosat
louisan*
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
u of ark
u of ark
eosat
s
s
s
s
s
s
s
s
p
8
P
S
s
8
8
8
8
8
8
S
8
8
8
8
8
S
8
8
8
8
P
8
8
P
S
P
S
s
8
S
8
8
8
S
8
S
S
s
s
s
19931118
19940325
19940114
19940322
19931231
19940405
19940203
19940203
19940112
19940117
19940112
19931202
19931201
19931201
19931201
19940126
19940227
19940202
19940116
19940105
19940111
19940111
19940127
19931118
19940111
19940111
19931119
19931117
19940111
19940111
19940112
19931119
19940405
19940112
19940310
19940112
19940306
19931213
19940129
19940304
19940203
19931228
19931228
19940122
19931203
19931213
19931117
19940111
19940111
19931117
19931119
19940328
19940209
19940328
19940104
19940408
19940209
19940209
19940114
19940209
19940114
19931203
19931203
19931203
19931203
19940303
19940303
19940209
19940303
19940110
19940114
19940114
19940209
19931119
19940114
19940114
19931119
19931119
19940114
19940114
19940114
19931119
19940408
19940114
19940328
19940114
19940328
19931227
19940303
19940328
19940209
19931228
19931228
19940303
19931203
19931227
19931119
19940114
19940114
19931119
-------
MRLC Consortium
Documentation Notebook
April 1994
24
24
24
24
24
24
24
24
24
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
25
26
26
26
26
26
26
26
26
26
26
26
26
36
36
37
37
37
38
38
39
39
27
28
29
29
30
30
31
31
32
32
33
33
33
34
34
35
35
36
36
36
36
37
37
37
37
38
39
39
40
26
27
28
28
29
30
31
32
32
33
34
34
19920426
19921019
19920426
19921019
19930208
19921019
19930208
19921019
19930208
19920924
19930506
19920519
19920908
19920519
19920908
19920503
19920908
19920503
19920924
19920503
19920924
19920924
19920503
19920924
19920503
19920924
19901005
19901005
19910415
19930725
19901005
19910415
19930130
19930725
19920706
19910210
19910922
19920706
19920510
19920510
19921001
19930513
19921001
19930513
19921001
19920323
19921001
19910406
19920323
19920814
19940110
19940110
19940110
19940110
19940111
19931130
19940111
19931130
19940111
19940113
19931119
19931217
19931130
19931217
19931202
19931228
19931228
19931228
19931228
19931130
19931217
19940127
19931112
19931112
19940110
19940110
19931112
19940110
19940110
19931112
19940110
19940110
19940111
19931119
19940128
19931228
19931230
19940128
19931210
19940121
19931112
19931119
19940113
19940113
19940113
19931202
19931217
19931210
19931101
19931112
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
fptn
flMH
fnm
HUB
OUD
mm
IBID
mm
IDfll
JJJ^Q
nun
DUD
2Bflft
XDflA
QUQ,
OU&
flll&
fiym
O1D1
IDflA
Boot
1DO&
HUD
mm
DUfi
moA
O3E01
DUD
mm
nun
jnm
nnD
nun
nun
nun
nuD
nun
nun
mm
mm
mm
mm
mm
mm
mm
mm
mm
mm
mm
mm
u of ark
u of ark
u of ark
u of ark
louisana
eosat
louisana
meat
louisana
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
nof ark
uofaik
eosat
uofaik
u of ark
eosat
u of ark
uofaik
louisana
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
s
s
s
s
p
s
p
s
p
s
s
s
s
s
8
s
8
8
8
8
8
8
8
8
8
8
8
n
8
8
8
8
P
8
8
S
8
8
8
S
S
S
S
S
S
8
S
6
S
S
19940111
19940111
19940111
19940111
19940112
19931202
19940112
19931202
19940112
19940203
19931119
19931228
19931202
19931228
19931203
19931231
19931228
19931231
19931228
19931202
19931228
19940216
19931117
19931116
19940111
19940111
19931118
19940111
19940111
19931117
19940111
19940111
19940112
19931119
19940307
19931229
19940105
19940206
19931213
19940215
19931118
19931119
19940117
19940202
19940122
19931203
19931228
19931213
19931111
19931117
19940114
19940114
19940114
19940114
19940114
19931203
19940114
19931203
19940114
19940209
19931119
19931228
19931203
19931228
19931203
19940104
19931228
19940104
19931228
19931203
19931228
19940303
19931119
19931119
19940114
19940114
19931119
19940114
19940114
19931119
19940114
19940114
19940114
19931119
19940328
19940103
19940110
19940303
19931227
19940303
19931119
19931119
19940209
19940209
19940209
19931203
19931228
19931227
19931119
19931119
-------
MRLC Consortium
Docuiiifmtiflti Notebook
April 1994
26
26
26
26
26
26
26
26
26
26
26
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
28
35
35
35
36
36
37
38
39
40
41
42
27
28
29
30
30
31
32
32
33
33
34
34
35
35
36
37
38
38
39
40
41
41
42
26
26
27
27
28
29
30
31
32
33
34
35
36
37
38
39
19920713
19921001
19930427
19920307
19920713
19920307
19921102
19921102
19921102
19921102
19921102
19910616
19910616
19910920
19920501
19920922
19910904
19920501
19920821
19920314
19920821
19920314
19920922
19920314
19920821
19920314
19920314
19920314
19920805
19921024
19931214
19930418
19930723
19930723
19910810
19930511
19930511
19930730
19931002
19910826
19910826
19910826
19910826
19330409
19930730
19910709
19920508
19930815
19930511
19930308
19931217
19940110
19940110
19931217
19931217
19931217
19940121
19940121
19940214
19940121
19940121
19940113
19940118
19931217
19931228
19931228
19940113
19931228
19931217
19931228
19931228
19931228
19931228
19931228
19931202
19931130
19931130
19931130
19931217
19940204
19940318
19931230
19931230
19931230
19931112
19931112
19931230
19931228
19940225
19940113
19940118
19940113
19940113
19931230
19931209
19940214
19931217
19931217
19931217
19931217
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
Tiim
1H1P
]Q2D
imp
BfllH
mfp
fimi
OUD
SD^D
mm
BBD1
mm
QJJQ
Q]]Q
O3JD
tMlft
QJQJ
o^m
OXD}
DBfll
XDDA
QJQJ
JQQJ
DUD
DUD
BBBP
mm
QJUQ
Hjm
mo2
IQIQ
mjQ
miQ
2Qfl2
JQQ2
JQQJ
DJDI
Q2|Q
IQIQ
lyitn
IflHl
mm
mm
mm
mm
mm
mm
mm
mm
mm
eosat
11 of ark
u of ark
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat i
eosat i
eosat i
eosat i
eosat i
eosat i
eosat t
eosat c
eosat t
eosat i
eosat . i
eosat t
s 19931228
s 19940111
s 19940111
s 19931223
s 19931222
s 19931223
s 19940307
s 19940202
19940303
19940303
19940307
19940228
19940203
19931228
19931229
19931231
19940307
19931229
19931228
19931229
19931228
19940103
19931228
19940103
19931203
19931130
19931130
19931130
19931228
19940215
19940323
19940105
19940105
19940105
19931118
19931117
19940105
19931228
i 19940304
i 19940304
i 19940304
i 19940203
, 19940202
i 19940107
, 19931210
, 19940227
: 19931222
19931222
19931222
19931223
19931228
19940114
19940114
19931227
19931227
19931227
19940328
19940209
19940303
19940303
19940328
19940303
19940209
19931228
19940103
19940104
19940328
19940103
19931228
19940103
19931228
19940104
19931228
19940104
19931203
19931203
19931203
19931203
19931228
19940303
19940328
19940110
19940110
19940110
19931119
19931119
19940110
19931228
19940328
19940328
19940328
19940209
19940209
19940110
19931213
19940303
19931227
19931227
19931227
19931227
-------
MRLC Coowrtium
DocumeotatioQ Notebook
April 1994
28
28
29
29
29
29
29
29
29
29
29
29
29
29
29
29
29
29
29
29
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
31
31
31
31
31
31
31
31
31
31
31
31
31
31
40
41
26
26
27
28
29
29
30
31
32
33
34
34
35
36
37
38
39
40
26
27
27
28
29
30
31
32
33
34
35
36
37
38
39
40
26
27
28
29
30
31
32
33
34
35
36
37
38
39
19930511
19910810
19920515
19920904
19930923
19910817
19920515
19920819
19920819
19920819
19920819
19910801
19930822
19930822
19930822
19930822
19930822
19930822
19920904
19930806
19910723
19920506
19920810
19920810
19930525
19930728
19930728
19930728
19920725
19930728
19930728
19920826
19930813
19930728
19930728
19930728
19910714
19910612
19910612
19910714
19910714
19910714
19910714
19910730
19910730
19910612
19930921
19930921
19920427
19920427
19931217
19931217
19931217
19931217
19940214
19940110
19931230
19940110
19940110
19940110
19940110
19931230
19940204
19940204
19940128
19931130
19931130
19931130
19931217
19931217
19940121
19931217
19931217
19940113
19931230
19931230
19931230
19931228
19940113
19931217
19931217
19931217
19931130
19931130
19931130
19931217
19931230
19940110
19931230
19931230
19931230
19931228
19931230
19931230
19931217
19931217
19940401
19940401
19931130
19931217
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
IttfTj
HMIf
"fptP
2UD
fH^n
mm
IBID
gtmi
ttllB
BUD
Olfll
flXOD
OQflA
mjjj
OUD
mpi
CDfl)
mpi
mm
mfl?
mm
mm
XDfll
BVD
mm
mm
mm
OUD
DB01
IDO1
OUD
BUD
flUD
fll^D
1OO1
mm
mt|i
mm
BIDS
B1O1
OUD
mm
Him
mm
TIlBft
mm
Tnm
mm
mm
mm
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat i
eosat i
eosat f
eosat i
eosat t
eosat i
eosat (
eosat i
eosat i
s 19931223
s 19931223
s 19931222
s 19931222
s 19940304
19940117
19940106
19940127
19940202
19940202
19940117
19940107
19940216
19940216
19940216
19931130
19931130
19931130
19931222
19931222
19940128
19931222
19931222
19940205
19940106
19940106
19940106
19931231
19940301
19931221
19931221
19931221
19931130
19931130
19931130
19931220
19940106
19940127
19940106
19940106
19940106
i 19931231
i 19940106
; 19940106
i 19931222
i 19931222
5 19940405
i 19940405
s 19931202
i 19931222
19931227
19931227
19931227
19931227
19940328
19940209
19940110
19940209
19940209
19940209
19940303
19940110
19940303
19940303
19940303
19931203
19931203
19931203
19931227
19931227
19940209
19931227
19931227
19940209
19940110
19940110
19940110
19940104
19940303
19931227
19931227
19931227
19931203
19931203
19931203
19931227
19940110
19940209
19940110
19940110
19940110
19940104
19940110
19940110
19931227
19931227
19940408
19940408
19931203
19931227
-------
MRLC Consortium
Documenutioo Notebook
April 1994
31 40
32 26
32 27
32 28
32 29
32 30
32 31
32 32
32 33
32 34
32 35
32 36
32 37
32 38
32 39
33 26
33 27
33 28
33 29
33 30
33 31
33 33
33 34
33 35
33 36
33 37
33 38
34 26
34 27
34 28
34 29
34 30
34 31
34 31
34 32
34 33
34 34
34 36
34 37
35 26
35 27
35 28
35 29
35 30
35 30
35 31
35 31
35 32
35 33
35 34
19920427
19920723
19920808
19930827
19920909
19910806
19920909
19920909
19910806
19930811
19930811
19930928
19930928
19910331
19910331
19930717
19920527
19930818
19920815
19920815
19920815
19920706
19930903
19930903
19930903
19930903
19930903
19930809
19920923
19930809
19930809
19930809
19910617
19910921
19920923
19920923
19930926
19930926
19930926
19920930
19930816
19930816
19930816
19910624
19930816
19920624
19930816
19910811
19920618
19930613
19931230
19940110
19940110
19940118
19931230
19931230
19940204
19940128
19931217
19940325
19940325
19940401
19940401
19931217
19931217
19940121
19931230
19931230
19931230
19940110
19931217
19931217
19940401
19940401
19940401
19940401
19940401
19931230
19931230
19931230
19940214
19931217
19940209
19931217
19931228
19940121
19940401
19940401
19940401
19940210
19940125
19940214
19931217
19940228
19931217
19931122
19931217
19931228
19931228
19940325
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
84k__M_ A««^M6
imn ouaU
8 Him flOUt
8 mm eosat
8WM_^B_ A«^MAA
mill QO8B4
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
9trk wyom gap
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
9 tr wyom gap
8 mm eosat
9trk wyom gap
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
s 19940106
s 19940127
s 19940127
s 19940202
s 19940106
s 19940106
s 19940215
8 19940217
19931221
19940327
19940327
19940405
19940405
19931221
19931221
19940213
19940106
19940107
19940107
19940127
19931222
19931222
19940404
19940404
19940404
19940404
19940404
19940107
19940107
19940107
19940227
19931222
p 19940209
19931222
19940103
19940228
19940404
19940404
19940404
19940218
19940203
19940218
s 19931220
t 19940228
8 19931220
u 19931209
s 19931220
s 19940103
s 19940103
s 19940326
19940110
19940209
19940209
19940209
19940110
19940110
19940303
19940303
19931227
19940329
19940329
19940408
19940408
19931227
19931227
19940303
19940110
19940110
19940110
19940209
19931227
19931227
19940408
19940408
19940408
19940408
19940408
19940110
19940110
19940110
19940303
19931227
19940210
19931227
19940104
19940303
19940408
19940408
19940408
19940303
19940209
19940303
19931227
19940301
19931227
19931227
19940104
19940104
-------
MRLC Coniortjum
Documentation Notebook
April 1994
35 35
35 36
35 37
35 38
36 26
36 27
36 28
36 29
36 29
36 30
36 31
36 31
36 32
36 33
36 34
36 35
36 36
36 38
37 26
37 27
37 28
37 29
37 30
37 31
37 32
37 33
37 34
37 35
37 36
37 37
37 38
38 26
38 27
38 28
38 29
38 29
38 30
38 31
38 32
38 33
38 34
38 35
38 36
38 37
39 26
39 27
39 28
39 29
39 31
39 32
19930613
19930613
19930613
19930613
19930807
19930807
19930807
19910818
19920804
19930823
19920812
19930823
19930823
19930823
19930823
19930823
19930823
19910514
19930814
19930814
19920827
19920928
19920827
19930814
19930814
19930627
19910622
19910622
19910622
19910622
19910622
19930922
19930922
19930922
19910731
19930922
19930720
19930720
19920717
19930720
19930922
19930618
19930618
19910528
19920910
19920910
19920910
19920910
19930727
19930727
19940325
19940325
19931230
19931230
19940110
19940110
19940214
19931108
19940214
19931217
19931108
19931217
19931217
19931217
19931217
19940110
19940110
19940110
19931228
19940214
19940222
19940225
19931228
19931228
19931228
19940113
19931112
19940113
19940121
19940121
19940121
19940214
19940201
19940201
19940302
19940214
19931210
19931210
19931228
19931112
19940128
19940214
19940311
19940110
19940311
19940311
19940311
19940311
19931112
19931217
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
9 tr wyomgap
8 mm eosat
8 mm eosat
9tr wyomgap
8 mm eosat
8mm eosat
8mm eosat
8 mm eosat
8mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8«»MM mnnmt
mill BOSK
8^MMB* *
mm eosai
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8iff" eosat
8 mm eosat
8 mm eosat
8 mm eosat
9 tr wyomgap
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eostt
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
8 mm eosat
6
S
8
a
a
s
s
P
s
s
u
P
8
S
s
s
s
s
s
s
s
s
19940326
19940326
19940107
19940107
19940211
19940118
19940221
19931214
19940307
19931221
19931209
19931221
19931221
19931221
19931221
19940127
19940116
19940117
19931231
19940304
19940301
19940228
19940103
19940103
19940103
19940203
19931118
19940203
19940301
19940202
19940301
19940226
19940219
19940211
19940303
19940217
19931213
19931213
19940103
19931116
19940210
19940220
19940318
19940220
19940318
19940318
19940319
19940319
19931118
19931221
19940110
19940110
19940303
19940209
19940303
19940328
19931227
19931227
19931227
19931227
19931227
19940209
19940303
19940303
19940104
19940328
19940328
19940303
19940104
19940104
19940104
19940209
19931119
19940209
19940303
19940209
19940303
19940303
19940303
19940303
19940303
19940303
19931227
19931227
19940104
19931119
19940303
19940303
19940328
19940303
19940328
19940328
19940328
19940328
19931119
19931227
-------
MRLC Consortium
Documenutioa Notebook
April 1994
39
39
39
39
39
40
40
40
40
40
40
40
40
40
40
40
40
41
41
41
41
41
41
41
41
41
41
42
42
42
42
42
42
42
42
42
42
43
43
43
43
43
43
43
43
43
43
43
43
43
33
34
35
36
37
26
27
27
28
29
31
32
33
34
35
36
37
26
27
28
29
31
32
33
34
35
36
26
27
28
30
31
32
33
34
35
36
26
26
27
27
28
28
29
30
31
32
33
34
34
19930727
19930727
19930727
19930727
19930727
19930819
19920731
19920731
19930819
19920731
19930819
19930819
19930819
19930819
19930819
19930819
19920426
19930927
19930927
19930927
19930810
19930826
19930826
19930826
19930810
19930810
19930826
19931004
19931004
19930614
19920729
19930614
19930614
19930801
19930716
19930817
19930529
19910803
19930909
19910702
19930808
19910702
19930909
19910702
19930808
19930808
19930808
19930824
19920501
19920720
19931217
19931217
19931217
19931217
19931217
19940225
19940225
19940318
19940110
19940225
19931130
19931217
19931217
19931217
19931130
19931217
19931112
19940225
19940225
19940225
19940318
19931210
19931209
19931228
19931217
19931217
19931112
19940318
19940318
19940110
19940125
19931228
19940113
19940128
19940121
19940121
19931119
19940217
19940225
19940217
19931210
19940217
19940318
19931210
19931209
19931217
19940121
19940121
19931217
19931228
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
9
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
nun
mm
mm
IT^tB
HUP
Q3fll
ODflA
OBfll
fllfll
Hflffi
nini
moi
flUD
flUD
JDID
mm
fnfyi
OD1&
QQ^Q
fllfll
Dl^Q
flUB
OEUB
HUfi
flUD
mm
mm
nun
flim
flUD
tr k
flSID
mm
DMD
lUDtt
HMft
nun
JO]]}
filial
OUHi
mm
Bum
mm
mm
mm
mm
mm
mm
mm
mm
eosat t
eosat t
eosat t
eosat t
eosat i
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat -
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
eosat
labo s
eosat
eosat
eosat
eosat
eosat
eosat
wash
eosat
wash
eosat s
wash
eosat s
eosat s
eosat s
eosat s
eosat s
eosat s
eosat s
eosat s
i 19931221
, 19931221
. 19931221
19931221
19931221
19940228
19940228
19940326
19940116
19940228
19931202
19931220
19931220
19931220
19931130
19931220
19931118
19940301
19940301
19940301
19940303
19931213
19931210
19940103
19931220
19931220
19931117
19940325
19940325
19940119
19940203 1
19940103
19940203
19940203
19940202
19940203
19931119
19940218
19940301
19940218
19931213
B 19940218
19940322
19931213
19931210
19931220
19940301
19940202
19931220
19940103
19931227
19931227
19931227
19931227
19931227
19940303
19940303
19940328
19940209
19940303
19931203
19931227
19931227
19931227
19931203
19931227
19931119
19940303
19940303
19940303
19940328
19931227
19931213
19940104
19931227
19931227
19931119
19940328
19940328
19940303
9940205
19940104
19940209
19940209
19940209
19940209
19931119
19940223
19940303
19940223
19931227
19940223
19940328
19931227
19931213
19931227
19940303
19940209
19931227
19940104
-------
MRLC Consortium
Documentation Notebook
April 1994
SECTION 6
TM SCENE PREPROCESSING
This section contains documentation, protocols, and process flows being implemented at
the EROS Data Center to handle the preprocessing of the image data including radiometric
correction, geometric rectification and multitemporal image registration, terrain corrections, and
spectral clustering.
-------
-------
MRLC Consortium
Documentation Notebook
April 1994
6.1 TM Scene Preprocessing Overview
The attached document was prepared by the EROS Data Center as an overview of the
preprocessing flow being implemented at EDC for the Landsat Thematic Mapper scenes
purchased by the MRLC Consortium.
-------
-------
MRLC Consortium
Documentation Notebook
PROCESSING FLOW **", "94
i
INGEST IMAGE - Load the 6 TM bands
PERFORM DATA QUALITY CHECK -
Debanding - Attenuate the banding pattern found in TM imagery
Detector-to-Detector Noise - Research is currently being done to
develop techniques to perform detector-to-detector noise removal
Reduce Image Size - Down sample the image by a factor of four.
GPYRAMID calculates the average value of a 4x4 pixel area and
writes it to the output image.
Contrast Enhancement - Apply a contrast enhancement to the
filtered image
Concat - Combine the enhanced bands into a single output image
Print Image - Write the preview image to the quick-look printer.
PREVIEW THE QUICK-LOOK PRINT TO DETERMINE IF FURTHER
NOISE REMOVAL IS NECESSARY
ARCHIVE PREPROCESSED IMAGE - The preprocessed image will be
archived to tape. The MRLC Data Base will be updated to contain the pre-
processed image archive location and appropriate metadata
-------
MRLC Consortium
Documentation Notebook
April, 1994
PROCESSING FLOW CONT.
IMAGE REGISTRATION
Control Point Selection - Control points will be selected using the
technique determined to be most efficient and meet registration
requirements i.e. 7.5-minute topographic maps, DLG's, or GCP librar-
ies
Image Resampling - Images will be ortho-rectified to the UTM projec-
tion, NAD83 Datum, 30 meter pixels.
Image Verification - Images will be verified against 1:24,000-scale
topographic maps. All images failing to meet the requirement of geo-
metric accuracy of between -1 and +1 pixel will be rejected and new
control selected.
For multi-temporal coverage the most recent image will be registered
to a map-base and used as the reference image for subsequent regis-
trations.
Image-to-image - If multi-temporal data has been requested for a par-
ticular path/row the most recent image will be registered to a map-
base and used as the reference image for subsequent registrations.
Verification - Images will be verified against 1:24,000-scale topo-
graphic maps. All images failing to meet the requirement of geometric
accuracy of between -1 and +1 pixel will be rejected and new control
selected.
ARCHIVE CONTROL POINT FILES - The control point files will be written
to a special CTP.ARCHIVE directory. The appropriate control point informa-
tion will be passed to Archive Management Section for inclusion in the
COMPLEX data base. ;
ARCHIVE REGISTERED IMAGE - The registered image and correspond-
ing DEM data will be archived to tape. The MRLC Data Base will be updated
to contain the registered image archive location and appropriate metadata.
-------
-------
MRLC Consortium
Documentation Notebook
April, 1994
PROCESSING FLOW CONT.
CLUSTER REGISTERED IMAGE - Where single date data are available
the six ortho-rectified TM images will be input to the Clustering algorithm. If
multitemporal data exists over the path/row the 12 resampled preprocessed
TM bands will be input to the clustering algorithm,
ARCHIVE CLUSTERED IMAGE - The clustered image and associated sta-
tistics files will be archived to tape. The MRLC Data Base will be updated to
contain the clustered image archive location and appropriate metadata.
ARCHIVE WORKING DIRECTORY- Upon successful registration pertinent
processing information will be archived on digital tape.
-------
-------
MRLC Consortium
Documentation Notebook
May 1995
6.2 Online Map of Scene Processing Status
The EROS Data Center is maintaining a graphical map showing the current status of
acquisition, preprocessing, and archiving of TM scenes for the MRLC. The map can be
viewed online through the MRLC WWW exhibit or directly from the EROS Data Center's
server. The URLs are:
http://www.epa.gov/grd/inrlc
http://edcftp.cr.usgs.gov/pub/sarges/mrlc.gif
-------
MRLC Consortium
Documentation Notebook
April 1994
From mail Thu Mar 31 11:06 EST 1994
To: MRLC Consortium Members
From: Joy Hood, EROS Data Center
Subject: MRLC TM Processing Status;
The following scenes over EPA regions 2 and 3 have been geometrically
registered as of 3/31/93 (* indicates scenes completed since 3/18/94):
Path/Row
014/032
015/032
018/033
018/032
017/034
015/030
013/031
015/031
015/033
014/033
015/034
017/030
013/032
012/031
017/035
014/036
019/035
039/031
026/040
Acquisition Data
05/20/91
03/17/91
06/17/93
10/20/92
08/06/92
08/06/92
10/02/92
10/18/91
10/06/92
06/14/92
09/16/91
05/04/91
10/18/91
04/11/92
10/02/92
09/20/92
09/27/91
11/03/92
05/11/92
10/13/92 *
06/23/92 «
09/30/92 «
07/27/93 *
11/02/92 *
The following scenes are scheduled for completion within the next
three weeks:
Path/Row
016/033
016/031
016/030
015/029
Acquisition Data
05/20/92
03/01/92
05/20/92
05/20/92
06/14/92
Path/Row
017/031
026/028
026/028
Acquisition Data
10/02/92
05/11/92
10/01/92
05/13/92
-------
-------
MRLC Consortium
Documentation Notebook
April 1994
6.3 Preprocessing Protocol Documentation
This section included documentation prepared by the EROS Data Center to support the
preprocessing of the TM scenes.
o Debanding and noise removal will be performed using a custom LAS .pdf
program called preview.pdf, a copy of which is included in this section. This
.pdf will also produce a quick-look image for review by EDC staff to indicate
whether additional radiometric corrections are necessary prior to further
processing.
o EDC staff have been conducting research on geometric rectification and terrain
corrections. The LAS software system has full geometric rectification capability,
and the documentation is included in this section.
o The EDC is currently conducting research on the effects of using 1:100,000
Digital Line Graphs for the selection and application of control points. Several
documents relating to this research are included in this section.
o The EDC has also been integrating the capability to perform terrain corrections
into the LAS software. Included in this section is a memorandum describing the
terrain correction approach being implemented.
o The final documents in this section include sample output on rectification and
terrain correction that will be available on request for each of the processed
scenes. These output products included graphical displays of control point
displacement and statistical summaries.
-------
-------
MRLC Consortium
Documentation Manual
January, 1994
PREVIEW.PDF
(1) FSTFMTIN - Load the 6 TM bands
(2) DEBAND - Attenuate the banding pattern found in TM imagery
(3) Detector-to-Detector noise removal - Research is currently
being done to develop techniques to perform detector-to-detector
noise removal
(4) GPYRAMID - Down sample the image by a factor of two. GPYRA-
MID calculates the average value of a 4x4 pixel area and writes it to the
output image.
(5) GPYRAMID - Down sample the previously down-sampled image
by 2.
(6) FILTERJ-HGH - Apply a high-pass filter to the down sampled
image created in step 5. above
(7) REDIST2 - Apply a contrast enhancement to the filtered image
(8) CONCAT - Combine the enhanced bands into a single output
image
(9) QLP_ADD - Write the preview image to the quick-look printer.
-------
MRLC Consortium
Documentation Manual
January, 1994
Overview of the
Geometric Manipulation Package
September 1992
-------
-------
MRLC Consortium
Documentation Manual
January, 1994
TABLE OF CONTENTS
1. The Geometric Manipulation Package 1-1
1.1 Control/Tie Point Selection & Refining 1-2
1.2 Calculation of Mapping Parameters 1-4
1.3 Image Transformation 1-5
2. Processing Scenarios 2-1
2.1 Image-to-Map Rectification 2-1
2.1.1 Image Ingest and Preparation 2-1
2.1.2 Control Point Selection 2-1
2.1.3 Coordinate Transformations & Framing of an Output Space 2-2
2.1.4 Point Modeling and Grid Generation 2-3
2.1.5 Geometric Transformation • 2-3
2.2 Image-to-image Registration 2-5
2.2.1 Image Ingest & Preparation 2-5
2.2.2 Control Point Selection 2-5
2.2.3 Refine Tie Points 2-6
2.2.4 Point Modeling and Grid Generation 2-6
2.23 Geometric Transformation 2-6
2.3 Changing Image Projections 2-8
APPENDIX A A-l
Representation of Image Geometry in the DDR A-l
APPENDIX B B-l
Registering Maps to Digitizers B-l
APPENDIX C C-l
Subpixd Accuracy '. C-l
APPENDIX D D-l
Framing of an Output Space D-l
APPENDIX E E-l
Projection Transformation Package E-l
APPENDIX F F-l
Modeling of Data F-l
APPENDIX G G-l
Gridding Proccu G-l
APPENDIX H H-l
Geometric Transformations H-l
APPENDIX I 1-1
-------
-------
MRLC Consortium
Documentation Manual
January, 1994
Geometric Manipulation Package
I. The Geometric Manipulation Package
The Geometric Manipulation Package is a set of modules designed to rectify remotely sensed data to a
user-selected reference frame. This reference frame may be a map projection, the geometry of another
image, or some other user-defined space.
The geometric rectification process involves many steps-often iterations of those steps-to register an
image. The rectification process can vary from project to project, depending upon the quality and
availability of reference maps, image quality and resolution, the type of registration (image-to-image or
image-to-map), and user preference of the control/tie point selection flow. Therefore, the Geometric
Manipulation Package has been implemented as a set of tools to satisfy a diverse set of requirements.
Since the package has been implemented with the tool approach, a TAE procedure, register, exists which
combines modules to perform image-to-image and image-to-map registrations. Thus, it is not necessary for
occasional users to understand individual modules or intermediate file structures in the package; only an
understanding of the driver procedures is needed.
The Geometric Manipulation Package is divided into three areas of processing: control/tie point selection
and refining, calculation of mapping parameters, and image transformation.
Projection
Definition
mm
Geometric
Mapping
Grid File
Registered
imoge
September 1992
1-1
-------
Geometric Manipulation Package
1.1 Control/Tie Point Selection & Refining
There are many different methods available in LAS 5.1 for the selection or origination of tie points. Points
may be selected from maps and images either interactively or separately, they may be entered from other
processing systems: or they may be entered manually. Once entered into the system, they may be edited.
displayed, or transformed into other coordinate spaces.
Because of this diverse set of requirements, a file format (the tie point selection file) was selected to store
points for a single image. Two tie point selection flies, one for the search image and one for the reference
image or geographic data, must be merged to specify an image-to-image or image-to-map registration.
Although merging the files requires an extra processing step, the initial generation of two separate data sets
permits measurement of map coordinates in a separate step from tie .point selection and also facilitates
re-using a set of reference image tie points for image-to-image registration of a series of images.
Tie/control points in LAS 5.1 may be entered in image, projection, geographic, or user-defined coordinates.
Image coordinates start in the upper-left corner of the image with a value of (1,1). Projection coordinates
may be entered in meters or feet; geographic coordinates may be entered in radians, degrees, seconds, or
in a packed degrees, minutes, seconds format; and user-defined coordinates may be given in almost any
unit of measure although transformations are not supported (see Appendix A). Projection and geographic
coordinates reside at the center of a pixel (see Appendix C).
The LAS 5.1 modules that provide tor point selection or ingest are tiepts, mappts, and tab2tu. Ttepts selects
ground control or tie points for image-to-map or image-to-image geometric rectification. Points may be
selected from multiple scenes/maps simultaneously or from single scenes. Mappts is a ground control point
collection module which creates a tie point selection file consisting of latitude, longitude or user-defined
coordinates (x,y) from a map. Input is through a digitizer or manually through the user's terminal Both
tiepts and mappts read and write the map identifier file which stores information needed to register a map
1-2
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Geometric Manipulation Package
to a digitizer. Each tie point in a tie point selection file contains the map identifier so that each point may
he referenced back to the map source. Tab2tie reads coordinates from a labeled table file and places them
tn a tie point selection file. Conversely, ne2tab reads the contents of a tie point selection or tie point
location file and places them in a labeled table file.
Editne edits the contents of existing tie point selection, merged tie point, or tie point location files. It also
creates new tie point selection, merged tie point, and tie point location files.
Dsprie displays tie point selection files, merged tie point files, and tie point location files to the line printer,
the user's terminal, or a text file.
Trancoord converts coordinates from one coordinate system to another. Coordinate conversions may be
performed between any of the projections supported by the LAS projection transformation package or with
a polynomial (up to fourth order). An option exists to grid the resulting coordinates to image coordinates
given a set of framing parameters (see Appendix D). Input and output projection spaces are described in
the projection definition file, which is created by the projprm module. Refer to Appendix E for more
information on the projection transformations supported by LAS.
Tiemerge merges two tie point selection files to form either a merged tie point file for use in
image-to-image correlation or a tie point location file for use in the tie point modeling process.
Correlate refines tie points for image-to-image registration using either gray-level correlation or edge
correlation. Input is a merged tie point file; output is a tie point location file.
September 1992 1-3
-------
Geometric Manipulation Package
1.2 Calculation of Mapping Parameters
The calculation of mapping (transformation) parameters is necessary prior to the transformation of an
image to a given geometric reference frame. Modules in this category allow the creation and editing of a
geometric mapping grid file given a set of tie points, a linear transformation, or a map projection change.
Appendix G describes the gridding process.
For the user with a set of tie points, pofyftt takes a tie point location file containing tie point pairs and
derives bivariate polynomial coefficients which define the transformation from one coordinate system to
another. The polynomial is used to create the geometric mapping grid of linear segments that
approximates this transformation. Rotmscl is used to generate a geometric mapping grid which corresponds
to a linear transformation representing a combination of rotation, translation, and geometric scaling;
whereas, the remap module generates a grid to change the projection of a previously registered image.
Various geometric mapping grid utility modules also exist: Gridfom converts geometric mapping grids in
other (old) formats to the current format, enabling the user to display and edit these grids as well as apply
them. Editgrid allows the user to change or originate the polynomial and projection information contained
in a grid file. Gridgcn is then used to regenerate the mapping grid portion of the file. Dspgrid is used to
display the contents of a geometric mapping grid file to the line printer, the user's terminal, or a text file.
The geometric mapping grid fife contains a geometric mapping grid, information needed to fill the output
space image's DDR, and miscellaneous mapping grid parameters and statistics.
tto »«*n
1-4
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Geometric Manipulation Package
1.3 Image Transformation
An image is geometrically rectified (transformed) according to the contents of the geometric mapping grid
file and other user-entered parameters.
The %eom module performs the geometric rectification of an image as specified by a previously generated
mapping grid using a one-pass, two-dimensional or a three-pass, one-dimensional algorithm. Resampling is
accomplished to 1/32 of a pixel using nearest neighbor interpolation, parametric cubic convolution
interpolation, bilinear interpolation, or a user-entered table of resampling weights.
If the user wants a resampling method other than nearest neighbor, cubic convolution, or bilinear
interpolation-a user-entered table of resampling weights can be generated with the nable module and
displayed with the dsprwt module. Rtable may be used to convert inverse point spread functions used in the
restoration process into a resampling weight table file or to create sine function resampling kernels of
various sizes.
The resampling weight table files contain up to three N by 33 matrices of separable interpolation weights to
be used in brightness level resampling. N is the resampling kernel dimension (2 to 16) and 33 identifies the
32 subdivisions between pixel values, including both endpoints.
The inverse point spread function file is an ASCII file which contains an inverse point spread function for a
given resampling method. This is the manner in which resampling weight tables for the restoration process
are entered into LAS; other resampling methods may be entered in this manner as well
Refer to Appendix H and the geom user's guide for more information on geometric transformations and
resampling.
PMM «p».»«
September 1992 1-5
-------
MRLC Consortium
Documentation Manual
January, 1994
Processing Scenarios
2. Processing Scenarios
This section provides descriptions of the modes of processing data geometrically. This includes the process
of rectifying an image to a map base, registering an image to another image, and changing projections of an
image. These examples are provided to guide the first-time or occasional user through the registration
process. Many other combinations of modules are possible; the most direct method for each type of
registration is given.
2.1 Image-to-Map Rectification
The process of rectifying an uncorrected image (the search image) to a map base (the reference space)
proceeds as follows.
2.1.1 Image Ingest and Preparation
LAS 5.1 contains various modules for ingesting image data. Many of these modules place nominal location
and projection information in the image's DDR. This information is very useful during the registration
process, even if it is not exact Examples of these modules are ccttipsp for TM-P data, dementer for DEM
(elevation) data, latin for AVHRR data, and edipsin and/riemer for MSS data.
Image preparation refers to the fixing of line drops (ficlin) or other image imperfections. It may also
include contrast stretches (map) to make the image data more manageable for the tie point selection
process.
2.1.2 Control Point Selection
The interactive selection of ground control point data from a map is accomplished using the tiepts module.
Output consists of two tie point selection fifes-reftps, containing geographic coordinates from the map(s),
and srch;tps, containing the search image coordinates of the corresponding points. Since this is an
image-to-map rectification, the input image fife is specified as ("jrch), meaning there is no reference
image; the search image name is snh. Note Appendix B describes the digitizer interface and Appendix C
describes subpixel image coordinates.
LAS> T!SmiK<<^£K^
September 1992 2-1
-------
MRLC Consortium
Documentation Manual
January, 1994
Processing Scenarios
2.1.3 Coordinate Transformations & Framing of an Output Space
The geographic coordinates in ref-.tps must be converted to line and sample locations in the output image
space. To accomplish this, three modules are used. First, the output image projection parameters are
established using the projprm module. In this example, the output image will be registered to a Transverse
Mercator projection with a Clark 1866 ellipsoid, a scale factor at the central meridian of 0.9996. a central
meridian of 103 degrees west, and with a latitude of origin of zero. The unit of measure of the projection
is meters.
LAS> PROJPRM-TM OUTPROJ-EXAMFLE ?ROJKEY«TM" +
SCALFACT«.»9W CENTMER»-I«1* ORIGIN*** +
GEOUNTTS«DEG DATUM** SMAJORAX«- ECSQVAL««
FALEAST-0 FALNORTH-* PROJUNIT«MET
Next, the output image space frame must be established and the geographic coordinates in ref;tps gridded
to output image space coordinates. This is accomplished using the trancoord module and the -grid
subcommand. The output space is framed using the Transverse Mercator projection previously described,
which includes the area bounded by 4135 north, 107.75 west and 37.90 north and 100.10 west Each pixel
is SO meters square (in the projection distance). Refer to Appendix D for more information on the
framing process.
LAS>
INTS-RET +
INPROJ-EXAMPLE PROJKY.- OPlOJKEf »«TM"
2-2
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Processing Scenarios
Finally, the two tie point selection files, ref.mapped;tps and srch;tps, are merged into a tie point location file,
example;tpl. for use in the modeling process.
LAS> TIEMERGE-NOCORK INTS«(REF.MAPPED,SRCH} -f
OUTTL.EXAMPLE CONFLG«NO
2.1.4 Point Modeling and Grid Generation
Example;tpl contains line and sample coordinates for both the search image and the resulting output image
space. These coordinate pairs are now used to derive a pair of second order teast squares regression
equations that predict input coordinates from output coordinates. Results of the predictions are displayed,
including point identifiers, coordinates, residuals of the transformation, and RMS residual values.
The user may remove or reinstate points and refit until a satisfactory fit is obtained. When a satisfactory fit
has been obtained, a geometric mapping grid is created.
LAS> POLYflT INTL-EXAMPLE WINIKW**
2.1.5 Geometric Transformation
The input image, srth-img, is now rectified to the map base defined by the geometric mapping grid gridgrid.
Cubic convolution with an alpha parameter of -0.5 is the resampling method used. The output image size
was determined during the process of framing the output image space and is contained in the geometric
mapping grid.
LAS> GEOM
September 1992
2-3
-------
MRLC Consortium
Documentation Manual
January/ 1994
Processing Scenarios
An alternate way to perform this rectification combines the steps of coordinate transformation and framing,
point modeling and grid generation, and geometric transformation steps into one module--register. This
method is less flexible than running the functions individually but is useful for the occasional user. This is
accomplished by running nepts. projprm, and register.
LAS> HEFTS IN*r»S«CH) OUTTS«(REF,SttCH)
LAS> PROJPRM-TM OUTPROJ-EXAMPLE PROJKEY-TM* 4-
SCALFACT»J996 CENTMER—183.9 ORIGIN-** +
GEOUNITS-DEG DATUM-8 SMAJORAX«~ ECSQVAL-**
FALEAST^ FALNORTH^ PROJIWIT-MET
LAS> REGISTER-IMG2MAP 1N«SRCH INtB*(1lEF^RCH)
1NPROJ-EXAMPLE OUT*IMAGE,OUT IPBOJKEY--
OPROJKEV-TM
cooRiwrr-T>Ec« POSGKEE-I Jus4Mp»cc
CONFLG-NO FBJNT*LF
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Processing Scenarios
2.2 Image-to-Image Registration
The process of registering an image (the search image) to another image (the reference image) proceeds as
follows.
2.2.1 Image Ingest & Preparation
LAS 5.1 contains various modules for ingesting image data. Many of these modules place nominal location
and projection information in the image's DDR. This information is very useful during the registration
process, even if it is not exact Examples of these modules are ccnipsp for TM-P data, dementer for DEM
(elevation) data, lactn for AVHRR data, and edipsin and planter for MSS data.
Image preparation refers to the fixing of line drops (/zr/m) or other image imperfections. It may also
include contrast stretches (map) to make the image data more manageable for the tie point selection
process.
2.2.2 Control Point Selection
The interactive selection of tie point data is accomplished using the ciepts module. Refer to the riepts user
guide for a description of the point selection process. Points may be selected manually or automatically
after a number of manual points have been selected. Output consists of two tie point selection files, one
containing reference image coordinates, nrf^pj, and the other containing search image coordinates, srch;tps.
LAS> TffiPTS IN«(BEF,8RCH}
Next, the two tie point selection files (rcfzps and srchyps) are merged into either a merged tie point file for
use in the correlation process or a tie point location file for use in the modeling process.
If image-to-image correlation is not toad, create the tie point location file exampleypl.
OUTTL«EXAMPLE
September 1992 2-5
-------
MRLC Consortium
Documentation Manual
January, 1994
Processing Scenarios
Otherwise, create the merged tie point file aample;mtp for use in the correlation process.
LAS> TIEMERGE-PRECORR INTS«(REF,5RCH) + OUTMT«EXAMPLE
CONFLG=NO
2.2.3 Refine Tie Points
The correlation process refines tie points using either grey level or edge correlation. This step may be
skipped if the user thinks tie points were selected with satisfactory accuracy. If tie points were collected in
automatic mode with tiepts, correlation is recommended. Output from the correlation process is a tie point
location file, example;tpl.
LAS> CORREIATE-GREY tN«(KET,SBCB) INMT-EXAMPLE
OUTTLvEXAMPLE MIN€ORR»*
FTTMETHmPARAB CONFLG-KO
2.2.4 Point Modeling and Grid Generation
Example;tpl contains line and sample coordinates for both the search and reference images. These
coordinate pairs are now used to derive a pair of second order least squares regression equations that
predict input coordinates from output coordinates. Results of the predictions are displayed, including point
identifiers, coordinates, residuals of the transformation, and RMS residual values.
The user may remove or reinstate points and refit until a satisfactory fit is obtained. When a satisfactory fit
has been obtained, a geometric mapping grid is created.
LAS> POLYFTT
oinri
2.2.5 Geometric Tranflfonnitioa
The input image, srdijmg, is now registered to the reference image space defined by the geometric
mapping grid gridgrid. Cubic convolution with an alpha parameter of -0.5 is the resampling method used.
The output image size was determined during the modeling process (pofyfit module/window parameter) and
is contained in the geometric mapping grid.
LAS> GEOM IN«S*CH
RESAMP-CC
2-6
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Processing Scenarios
An alternate way to perform this registration combines the steps of tie point merging, correlation, point
modeling and grid generation, and geometric transformation steps into one module, register. This method is
less flexible than running the functions individually but is useful for the occasional user. The user thus runs
nepts and register.
LAS> TIEPTS IN*(KEF,SRCH} OUTTS*(REF,SRCH)
LAS> REG1STER-IMG2IMG IN«SRCH INTS«(REF,SRCH) +
1NPROJ«EXAMPLE OUT«IMAGE>OUT -f
WlNDOW»(l,U«flV«W) CORRMODB«GRET ^
BANDS«(U) PDEGttEE-2 RESAMP«CC CONTLG-NO *
PRINT-LP
September 1992
2-7
-------
MRLC Consortium
Documentation Manual
January, 1994
Processing Scenarios
2.3 Changing Image Projections
The projection of an image is changed using the projprm, remap, and geom functions. Remap operates in
two modes-one mode reads input (existing) projection information from the input image DDR and the
other obtains it manually from parameter input In this example, a DEM image with valid projection and
location information in its DDR is being reprojected to match the image rectified in the image-to-map
example above. Projprm defines the output projection system, remap determines image frame and creates a
geometric mapping grid, and geom remaps (transforms) the input image using the geometric mapping grid
and bilinear interpolation resampling.
LAS> PROJPRAMM OUTPROJ.EXAMPLE PROJKEY-TST +
SCALFACT*J99C CENTMER—Ittt ORIGIN-** *;
GEOUNITS-DEG DATUM «• SMAJORAX— BCSQVAL-*
FALEAST-0 FALNOR1H«* PRQJUNIT-MET
LAS> REMAP IN«DEM INTitOJ«EXAMFLE OCTGRID«GRID
OPROJKEY-TW PIXS!Z-(5MI> ^
COORUNIT»'DEG* ROTANG-t*
LAS> GEOM IN-DEM
RESAMP«BI
OUT-DEM.OUT
2-8
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Processing Scenarios
Another scenario involving the changing of image projection is the combination (side-by-side) of two data
sets which are registered to a map projection but not to the same map projection. As an example, consider
two DEM images with valid DDRs, one registered to UTM zone 11 and one registered to UTM zone 12.
Before these two images can be combined, they must be in a common projection system. In this example,
the DEM registered to UTM zone 12 will be remapped to UTM zone 11. The two images may then be
combined using the concat function:
LAS> PROJPRM-OTM OUTPROJ -EXAMPLE +
PROJKE¥**UTMir ZONENUM-ll
LAS> REMAP IN-DEM.12 INPROJ«EXAMPLE
OUTGRID«GJHD#4-
OPROJKEY-1JTM11* PiXSlZ«<38^»)
LAS> GEOM IN«D£tt4i IKGRID«GRII> OUT-DEM.12 11
RESAMP-BI ". .: •.;••-.., • . ..... . -... ..-.-
LAS>
II)
*•••. • *
September 1992
2-9
-------
MRLC Consortium
Documentation Manual
January, 1994
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix A
APPENDIX A
Representation of Image Geometry in the DDR
Information regarding image geometry and location is contained in the image's data descriptor record
(DDR). Functions in LAS 5.1 utilize and update this information while functions in the Geometric
Manipulation Package often determine image geometry and originate this data location information.
Tracking location and projection related information in the DDR helps to automate the mosaicking,
combining, or overlaying of images in like projections. It assists in the verification of the registration
process and allows the user the freedom to select any pixel at some point in the processing of an image
and determine that pixel's projection coordinates. Once projection coordinates have been found, latitude
and longitude may be calculated using the appropriate function.
The coordinate system of the projection plane is a cartesian coordinate system-the projection coordinate is
designated by its distance from two perpendicular axes. The point at which these axes cross is the
projection origin. Terms commonly used to name these coordinates are: projection coordinates, northings
and eastings, (U,V) and (X,Y).
Image coordinates (line, sample) are also in a cartesian coordinate system. In many cases, the axes of the
projection coordinate system and the axes of the image coordinate system are parallel If this is so,
translating and scaling is all that is needed to convert between the two systems. Projection coordinates may
be converted to image coordinates when one common point in the two coordinate systems and the pixel
size in projection units is known.
September 1992 A-l
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix A
In Figure A-l. if the upper-left image coordinate and upper-left projection coordinate is known, the
projection coordinate of a window beginning at image coordinate (1000,1000) may be found as follows:
proj_y = upper_left_proj_y - ((line - I) * pixel_size_y)
proj_x = upperjeft_proj_x + ((sample - 1) • pixel_size_x)
and in this example:
proj_y - 77900.0 - ((1000 -1) * 25.0) - 52925.0
proj_x « -83575 + ((1000 - 1) • 25.0) « -58600.0
[-•3373,77900]
Ptx«l (1.1)
A
(US) •-> imago
[x.y] --> •rolooMon
Coordinator
1 plxol • 23.0 motor*
In both x * y
Wxot (1000.1000)
[-58600.92929]
312
S12
[0.0]
(3117.3344)
(4444.1)
(1.3BS9)
Origin
V
(444«,3B59)
[12679.-33229J
Figure A^
A-2
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix A
In most cases the projection coordinate system is parallel to the image coordinate system. It is possible,
however, to have a projection coordinate system which is rotated from the image coordinate system (Figure
A-2). Converting between the two systems requires rotation in addition to translating and scaling. An
example of this condition is processed TM data (TM-P) as read by ccrripsp. The DDR update routines
used bv all LAS modules handle this condition.
A
IMAGE
Image Axis
Projection
Axis
V
FlforeA-C
September 1992
A-3
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix A
A list and brief description of each projection related field stored in the DDR follows.
Projection Code Numeric code of projection system used. Refer to Appendix E for a list
of supported projections.
Zone Code Numeric code of projection zone (UTM or State Plane Coordinate
System). Appendix E contains tables listing zones and their corresponding
area of coverage.
Datum Code Numeric code for datum used. Refer to Appendix E for a list of datums
supported.
Projection Units This is an ASCII field containing the unit of measure for the projection
system. The Projection Transformation Package supports degrees,
seconds, radians, feet, meters, and a packed degrees/ minutes/seconds
format. This field is ASCII so units other than these types may be used;
however, the Projection Transformation Package will work only with the
units it supports.
Projection Parameters An array of IS projection parameters containing parameters specific to a
given projection. A table describing the contents of this array for a given
projection is given in Appendix E.
Upper-Left Corner The projection coordinate at pixel (1,1).
Lower-Left Corner The projection coordinate at pixel (NL.1), where NL is the number of
image lines.
Upper-Right Corner The projection coordinate at pixel (ItNS), where NS is the number of
image samples.
Lower-Right Corner The projection coordinate at pixel (NUNS), where NL is the number of
image lines and NS is the number of image samples.
Projection Distance The amount of projection distance in X and in Y that one image pixel
covers (also referred to as pixel size).
Refer to the LAS 5.1 Programmer's Manual for a description of the DDR and the manner in which each
LAS module updates the DDR.
A A September 1992
-------
KKLC Consortium
Documentation Manual
January, 1994
Appendix B
APPENDIX B
Registering Maps to Digitizers
The process of registering a map to a digitizer proceeds as follows:
The user is prompted for map parameters. Refer to the tiepts or mappa user's guides for a description of
the user interface and the type of data to be entered.
• The user sparks points to register the map to the digitizer. The number of points used
depends upon the map being used. The map coordinates entered by the user are fit to the
corresponding digitizer coordinates for a given point If six or fewer points are sparked, a
first order least squares fit is used. Otherwise, a second order least squares fit is used. If
the four map comers are not square (Le., the digitizer to map fit cannot be determined
with a rotation, X and Y translation, and X and Y scaling), more than four points are
required. Either latitude/longitude or projection (XY) coordinates may be used in
defining the map coordinates. The user should abo keep in mind errors due to map
shrinkage and expansion if paper maps are utilized.
• A verification point is sparked. This point is not used in the digitizer-to-map fit calculation
and gives an added indication of map registration accuracy. The verification point is a
known coordinate marked on the map-a tick mark or some other point whose coordinate
is known.
September 1992 B-l
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix B
Digitizer Interface
The digitizer is connected between the user's terminal and the
host computer system (as shown below).
Host
Computer
Digitizer
Terminal
Figure B>7
With this configuration, the digitizer coordinates usually echo on the user's terminal Currently, ALTEC
and COMPLOT 7000 digitizers are supported. This interface should also work with a variety of other
digitizers.
The routines which interface with the digitizer expect the following format:
B XXXXX YYYYY
where
B is the digitizer button pushed,
XXXXX is the X digitizer coordinate,
YYYYY is the Y digitizer coordinate,
is a carriage return (one character),
and a blank is a space.
The button numbers Cor various digitizer (unctions are described in the digimer.H include file. This file also
contains the digitizer format The active button numbers may be changed in the include file and become
valid when the modules which interface to the digitizer are rebuilt. The digitizer format is not as easily
changeable at this time.
B-2
September 1992
-------
MKLC Consortium
Documentation Manual
January, 1994
Appendix C
APPENDIX C
Subpixei Accuracy
In the LAS 5.1, the location of a projection or other (X,Y) coordinate is defined to be located at the center
of a pixel. The diagram that follows describes coordinate locations when dividing a pixel into subpixel
increments.
Y - o.s
X - 0.5
X - 0.25
—H
-X +,0.25
0.5
Y - 0.25.
0.25.
September 1992
C-l
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix C
To calculate subpixel values from an enlarged (interpolated) image window (as in tiepts), the following
equation is applied:
x' = 0.5 +
y' - 0.5 +
2 * zoom
2 * zoom
(x-l)
zoom
(y-i)
zoom
where:
xandy
x' and y*
zoom
are integer image coordinates in the enlarged image
are noninteger image coordinates in the original image
is the enlargement factor
C-2
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix D
APPENDIX D
Framing of an Output Space
Three different methods of specifying the output space frame are provided in LAS 5.1 (modules trancoord
and remap).
Method 1:
The user defines the upper-left and lower-right corner coordinates of the area of interest in geographic
coordinates. This forms a rectangle in geographic coordinates.
LATiLONOi UTiLOMGi
LATiLQNGi
UTiLONGi
LATiLONGi
UTiLONGi
Figure D-l
This rectangular space is projected into the output projection coordinate system. This usually results in a
nonrectangular area.
LATiLONGi LATiLONG.
LAT«
LONO«
Figure D-2
September 1992
D-l
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix D
The boundaries of this projected area are searched for minimum and maximum X and Y coordinates in the
output projection coordinate system. This is performed by stepping along the output space frame at a
given interval, sometimes resulting in approximate minimums and maximums. In most cases, the difference
between the absolute and approximated minimum/maximums is negligible. This forms a minimum
bounding rectangle in the output space of the area of interest
D-2
September 1992
-------
MRL.C Consortium
Documentation Manual
January, 1994
Appendix D
The upper-left corner (the minimum X, maximum Y) is adjusted outward to the next X and Y multiples of
the pixel size. This is an arbitrary step which simplifies the combining of images with different scales. It
does not alter the internal image geometry, it only slightly adjusts the size of the image frame and the
location of pixel (1,1).
The upper-left corner projection coordinate is assigned to image coordinate (1,1) in the output image.
Figure D-4
September 1992
D-3
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix D
Method 2:
The user defines:
• An output space projection coordinate at some image coordinate. When the image
coordinate defaults to (1,1), the output space coordinate given is the upper-left corner of
the output space.
• The lower-right corner in the output projection space.
The first coordinate given is adjusted to image coordinate (1,1) to get the minimum X and the maximum Y
projection coordinate. The second coordinate pair given is the maximum X and the minimum Y
coordinate. The output space is now defined. The pixel size is applied to the projection minimums and
maximums to determine the number of lines and samples in the output space.
;MINX
UNI
UtCORNCR
(MAXX.MINV)
Figure D-5
D-4
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix D
Method 3:
• The user defines an output space projection coordinate at some image coordinate as in Method 2.
The user enters the number of output image lines and samples.
The coordinate entered is adjusted to image coordinate (1,1) to get the minimum X and the maximum Y
projection coordinates.
The maximum X and minimum Y coordinates are calculated using the pixel size and the number of lines
and samples from the minimum X and maximum Y projection coordinate.
Note that only the upper-left corner (the minimum X, maximum Y) of the output space is needed to grid
the area to image coordinates. However, the lower-right corner (maximum X, minimum Y) is necessary to
initialize framing information that is needed later in the process of registering an image.
September 1992 D-5
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix E
APPENDIX E
Projection Transformation Package
The LAS 5.1 Projection Transformation Package is a system of subroutines designed to permit the
transformation of coordinate pairs from one map projection to another. The primary "workhorse" of this
package is the U.S. Geological Survey's General Cartographic Transformation Package (GCTP), although
it has been buffered from the application routines to minimize changes to calling programs when the GCTP
is changed or added to. A good reference on map projections is:
Snyder, John P., Map Projections-A Working Manual: U.S. Geological Survey Professional Paper 1395,
United States Government Printing Office, Washington D.C., 1987.
Supported Projections
The following projections are supported by the Projection Transformation Package. All are available in the
spherical form; many are available in the ellipsoidal form.
Albers Conical Equal Area
Azimuthal Equidistant
Equidistant Conic
Equirectangular
General Vertical Near-Side Perspective
Geographic
Gnomonic
Hammer
Lambert Azimuthal Equal Area
Lambert Conformal Conk
Mercator
Miller Cylindrical
Oblique Mercator (Hotine)
Orthographic
Polar Stereographk
Poryconk
Robinson
Sinusoidal
Space Oblique Mercator (SOM)
State Plane Coordinate System
Stereographk;
Transverse Mercator
Universal Transverse Mercator (UTM)
Van Der Grinten
September 1992 E-l
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix E
Projection Units
A list of supported units follows this paragraph. Meters are most commonly used for most projections;
however, exceptions are radians, degrees, seconds, and packed degrees/ minutes/ seconds for geographic
and a common use of feet for the State Plane Coordinate System.
Radians
Feet
Meters
Seconds of arc
Degrees of arc
Packed Degrees Minutes Seconds (DDDMMMSSS.SS)
£.2 September 1992
-------
KRLC Consortium
Documentation Manual
January, 1994
Appendix E
Datum
The semimajor axis and semiminor axis for a given datum may be selected via a datum code in projprm. If
the datum to be used is not supported via menu selection, projprm allows the user to manually enter the
values. NOTE: The Projection Transformation Package will not currently convert between datums!
Supported datums and their values: (in meters)
Datum
Clarke 1866
Clarke 1880
Bessel
International 1967
International 1909
WGS72
Everest
WGS66
GRS 1980
Airy
Modified Everest
Modified Airy
Walbeck
Southeast Asia
Australian National
Krassovsky
Hough
Mercury 1960
Modified Mercury 1968
Sphere of Radius
Semi-maior Axis
6378206.4
6378249.145
6377397.155
6378157.5
6378388.0
6378135.0
6377276.3452
6378145.0
6378137.0
6377563396
6377304.063
6377341.89
6376896.0
6378155.0
6378160.0
6378245.0
6378270.0
6378166.0
6378150.0
6370997.0
Semi-Minor Axis
6356583.8
6356514.86955
6356078.96284
6356772.2
6356911.94613
6356750.519915
6356075.4133
6356759.769356
635675231414
6356256.91
6356103.039
6356036.143
6355834.8467
63567733205
6356774.719
6356863.0188
6356794343479
6356784.283666
6356768.337303
6370997.0
September 1992
E-3
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix E
Zone codes
The Universal Transverse Mercator (UTM) and the State Plane Coordinate System use zone codes instead
of specific projection parameters. The two tables which follow in Figure E-l and E-3 list UTM and State
Plane zone codes as used by the LAS 5.1 Projection Transformation Package. Figure E-2 shows UTM
zones plotted on a world map (Figure E-2 source: Map Projections-A Working Manual: U.S. Geological
Survey Professional Paper 1395). Equations for each State Plane zone are given in:
Clarie, Charles N, State Plane Coordinates by Automatic Data Processing, Publication 62-4, U.S. Department
of Commerce, Environmental Science Services Administration, Coast and Geodetic Survey, United States
Government Printing Office, Publication 62-4, 1973.
£.4 .September 1992
-------
MR1.C Consortium
Documentation Manual
January, 1994
Appendix E
Figure E-l
UTM Zone Codes, Locations, and Central Meridians
Zone
01
02
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
CM.
Range
Zone
177W
171W
153W
147W
141W
135W
129W
123W
117W
111W
105W
099W
093W
087W
081W
075W
069W
063W
057W
05 1W
045W
039W
033W
027W
021W
015W
009W
003W
180W-174W
174W-168W
156W-150W
150W-144W
144W-138W
138W-132W
132W-126W
126W-120W
120W-114W
114W-108W
108W-102W
102W-096W
096W-090W
090W-084W
084W-078W
078W-072W
072W-066W
066W-060W
060W-054W
054W-048W
048W-042W
042W-036W
036W-030W
030W-024W
024W-018W
018W-012W
012W-006W
006W-OOOE
31
32
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CM.
003E
009E
027E
033E
039E
045E
051E
057E
063E
069E
075E
081E
087E
093E
099E
105E
HIE
117E
123E
129E
135E
141E
147E
153E
159E
165E
171E
177E
Range
OOOE-006E
006E-012E
024E-030E
030E-036E
036E-042E
042E-048E
048E-054E
054E-060E
060E-066E
066E-072E
072E-078E
078E-084E
084E-090E
090E-096E
096E-102E
102E-108E
108E-114E
114E-120E
120E-126E
126E-132E
132E-138E
138E-144E
144E-150E
150E-156E
156E-162E
162E-168E
168E-174E
174E-180W
September 1992
E-5
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix E
Figure E.2 Universal Transverse Mertator (UTM) grid zone designations for the
world shown on a horizontally expanded Equtsdistant Clyilndrical
projection index map.
E-6
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix E
Figure E-3
Jurisdictions. State Plane Coordinate Systems.
and Zone Representations
Jurisdiction Zone Jurisdiction
Zone name or number Code Zone name or number Code
Alabama Indiana
East 0101 East 1301
West 0102 West 1302
Alaska Iowa
01 through 10 5001 North 1401
thru 5010 South 1402
Arizona Kansas
East 0201 North 1501
Central 0202 South 1502
West 0203 Kentucky
Arkansas North 1601
North 0301 South 1602
South 0302 Louisiana
California North 1701
01 through 07 0401 South 1702
thru 0407 Offshore 1703
Colorado Maine
North 0501 East 1801
Central 0502 West 1802
South 0503 Maryland 1900
Connecticut 0600 Massachusetts
Delaware 0700 Mainland 2001
District of Columbia 1900 Island 2002
Florida Michigan
East 0901 East (Trans Merc) 2101
West 0902 Central (Trans Merc) 2102
North 0903 West (Trans Merc) 2103
Georgia North (Lambert) 2111
East 1001 Central (Lambert) 2112
West 1002 South (Lambert) 2113
Hawaii Minnesota
01 through 05 5101 North 2201
thru 5105 Central 2202
Idaho South 2203
East 1101 Mississippi
Central 1102 East 2301
West 1103 West 2302
Illinois
East 1201
West 1202
September 1992 E-7
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix E
Figure E-3 fcont)
Jurisdictions. State Plane Coordinate Systems.
and Zone Representations
Jurisdiction Zone Jurisdiction Zone
Zone name or number Code Zone name or number Code
Missouri Pennsylvania
East 2401 North 3701
Central 2402 South 3702
West 2403 Rhode Island 3800
Montana South Carolina
North 2501 North 3901
Central 2502 South 3902
South 2503 South Dakota
Nebraska North 4001
North 2601 South 4002
South 2602 Tennessee 4100
Nevada Texas
East 2701 North 4201
Central 2702 North Central 4202
West 2703 Central 4203
New Hampshire 2800 South Central 4204
New Jersey 2900 South 4205
New Mexico Utah
East 3001 North 4301
Central 3002 Central 4302
West 3003 South 4303
New York Vermont 4400
East 3101 Virginia
Central 3102 North 4501
West 3103 South 4502
Long Island 3104 Washington
North Carolina 3200 North 4601
North Dakota South 4602
North 3301 West Virginia
South 3302 North 4701
Ohio South 4702
North 3401 Wisconsin
South 3402 North 4801
Oklahoma Central 4802
North 3501 South 4803
South 3502 Wyoming
Oregon East (01) 4901
North 3601 East Central (02) 4902
South 3602 West Central (03) 4903
West (04) 4904
September 1992
-------
KRLC Consortium
Documentation Manual
January, 1994
Appendix E
Figure E-3 fcont)
Jurisdictions. State Plane Coordinate Systems.
and Zone Representations
Jurisdiction Zone Jurisdiction Zone
Zone name or number Code Zone name or number Code
Puerto Rico 5201 SL Croix 5202
Virgin Islands American Samoa 5300
St. John, SL Thomas 5201 Guam 5400
Obtained from Software Documentation for GCTP General Cartographic Transformation Package:
National Mapping Program Technical Instructions, U.S. Geological Survey, National Mapping Division,
May 1982, Appendix B, Table 1.
September 1992 E-9
-------
KRLC Consort ium
Documentation Manual
January, 1994
Appendix E
Projection Parameters
The LAS 5.1 Projection Transformation Package uses a double precision array of fifteen elements to send
projection specific parameters to the Projection Transformation Package. Normally, projection parameters
are entered through the projprm module, which doesn't require the user to know the array location and
format of each projection parameter. However, the locations of each projection parameter need to be
known when looking at a projection definition file (output from projprm), when editing a geometric
mapping grid file (module editgrid), when editing a tie point file (module edittie), or when editing an image
DDR (module editddr).
A table of projection parameters is given on the following page.
The following notes apply to the Space Oblique Mercator projection:
• A portion of Landsat rows 1 and 2 may also be seen as parts of rows 246 or 247. To place
these locations at rows 246 or 247, set the end of path flag (parameter 11) to 1-end of
path. This flag defaults to zero.
• When Landsat-1,2, 3 orbits are being used, use the following values for the specified
parameters:
Parameter 4 099005031.2
Parameter 5 128.87 degrees - (360/251 * path number)
in packed DMS format
Parameter 9 103.2669323
Parameter 10 0.5201613
• When Landsat-4,5 orbits are being used, use the following values for the specified
parameters:
Parameter 4 098012000.0
Parameter 5 1293 degrees • (360/233 • path number)
in packed DMS format
Parameter 98.884119
Parameter i: 0.5201613
E-10
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix E
Projection Transformation Package Projection Parameters
Array Element
Code & Projection Id
0 Geographic
1 UTM
2 State Plane
3 Albers Equal Area
4 Lambert Conform* I C
5 Mercator
6 Polar Stereographic
7 Polyconic
8 Equid. Conic A
Equfd. Conic 6
9 Transverse Mercator
10 Stereographic
11 Lambert Azimuthal
12 Azimuthal
13 Gnomon ic
14 Orthographic
15 Gen. Vert. Near Per
16 Sinusoidal
17 Equi rectangular
18 Miller cylindrical
19 Van der Grinten
20 Oblique Mercator A
Oblique Mercator B
21 Space Oblique Merc
22 Hammer
23 Robinson
1
Lon/Z
SMajor
SMajor
SMajor
SMajor
SMajor
SMajor
SMajor
SMajor
Sphere
Sphere
Sphere
Sphere
Sphere
Sphere
Sphere
Sphere
Sphere
Sphere
SMajor
SMajor
SMajor
Sphere
Sphere
2
Lat/Z
SMinor
SMinor
SMinor
SMinor
SMinor
SMinor
SMinor
SMinor
SMI nor
SMinor
SMinor
3
STDPR1
STDPR1
STOPAR
STOPR1
Factor
Height
Factor
Factor
4
STDPR2
STDPR2
STDPR2
AzIAng
IncAng
5
CentMer
CentMer
CentMer
LongPo I
CentMer
CentMer
CentMer
CentMer
CentLon
CentLon
CentLon
CentLon
CentLon
CentLon
CentMer
CentMer
CentMer
CentMer
AzmthPt
AscLong
CentMer
CentMer
6
OriginLat
OriginLat
TrueScale
TrueScale
OriginLat
OriginLat
OriginLat-
OriginLat
CenterLat
CenterLat
CenterLat
CenterLat
CenterLat
CenterLat
TrueSca le
OriginLat
OriginLat
OriginLat
7
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
FE
8
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
FN
9
Longl
PSRev
10
Latl
LRat
11
Long2
PFlag
12
Lat2
13
one
where
Lon/Z Longitude of any point In the UTM zone or zero. If zero, a zone code must be specified.
Lat/Z Latitude of any point In the UTM zone or zero. If zero, a zone code must be specified.
SMajor Semi-major axis of ellipsoid. If zero, Clarke 1866 in Meters is assumed.
SMinor Eccentricity squared of the ellipsoid if less than zero, if zero, a spherical form is
assumed, or if greater than zero, the semi-major axis of ellipsoid.
Sphere Radius of reference sphere. If zero, 6370997 meters is used.
STOPAR Latitude of the standard parallel
STDPR1 Latitude of the first standard parallel
STDPR2 Latitude of the second standard parallel
CentMer Longitude of the central Meridian
OriginLat Latitude of the projection origin
FE False easting in the same units as the semi-major axis
FN False northing In the same units as the semi-major axis
TrueScale Latitude of true scale
LongPoI Longitude down below pole of map
Factor Scale factor at central meridian (Transverse Mercator) or center of projection (Oblique
Mercator)
CentLon Longitude of center of projection
CenterLat Latitude of center of projection
Height Height of perspective point
Longl Longitude of first point on center line (Oblique Mercator, format A)
Long2 Longitude of second point on center line (Oblique Mercator, format A)
Latl Latitude of first point on center line (Oblique Mercator, format A)
Lat2 Latitude of second point on center line (Oblique Mercator, format A)
AziAng Azimuth angle east of north of center line (Oblique Mercator, format B)
AzmthPt Longitude of point on central meridian where azimuth occurs (Oblique Mercator, format B)
IncAng Inclination of orbit at ascending node, counter-clockwise from equator (SOU)
AscLong Longitude of ascending orbit at equator (SOM)
PSRev Period of satellite revolution In minutes (SOM)
LRat Landsat ratio to compensate for confusion at northern end of orbit (SOM ~ use 0.5201613)
PFlag End of path flag for Landsat: 0 - start of path, 1 * end of path
September 1992
E-ll
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix E
Notes: Array elements 14 and 15 are set to zero
All array elements with blank fields are set to zero
All angles (latitudes, longitudes, azimuths, etc.) are entered in packed degrees/ minutes/ seconds
(DDDttffSSS.SS) format
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix F
APPENDIX F
Modeling of Data
The LAS 5.1 module which fits two sets of tie points is pofyfit. Pofyftt uses a tie point location file
containing tie point pairs which specify a mapping from the output (reference) coordinates to the input
(search) coordinates to derive a pair of brvariate polynomials. The polynomials are calculated using least
squares regression analysis in a forward stepping procedure (stepwise linear regression). The user controls
the regression fitting by specification of the polynomial degree and statistical significance levels (alpha)
parameters and can remove or reinstate tie point pairs from the fitting process.
The degree of the polynomial is restricted to a maximum of 4. The actual degree may be less than that
specified, depending on the number and distribution of the tie point pairs. (If a set of tie points describes
a lower order transformation than the degree entered, the stepwise regression should eliminate polynomial
terms which are not significant.) The following table gives the minimum number of tie point pairs required
for each degree:
Number of
Degree Points
1 3
2 6
3 10
4 15
The stepwise linear regression procedure described in the following paragraphs is used by pofyfit to fit tie
point pairs, by the digitizer interface to compute the digitizer to map fit, and by riepts to compute rough
transformations and tie point residuals.
Stepwise Linear Regression
Stepwise linear regression examines variables incorporated in the model at every stage of the regression. A
variable which may have been the best choice to enter the model at an early stage may later be
nonsignificant because of the relationships between it and other variables now in the regression. Once a
variable is proven to be nonsignificant, it is removed from the model This process continues until no more
variables can be accepted and no more can be rejected.
The process of determining whether or not a variable is significant is based on the F-statistic and a user-
entered statistical significance level for variables entering the model and for those exiting the model
(commonly referred to as alpha). For example, if the user enters an alpha of (0.05,0.05), a point with a
September 1992 F-l
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix F
statistical significance in the upper 95 percent of the F-distribution is entered into the model; points in the
model may stay in the model if their statistical significance level is in the upper 95 percent of the
F-distribution.
Alpha values near 1.0 allow variables to enter and remain in the model that do not significantly help define
the relations between the tie point pairs. Alpha values near 0.0 may prevent any variables from entering
the model and/or will cause variables to be quickly removed after other variables are entered.
Alpha values of (0.05,0.05) are recommended. Alpha values of (0.999,0.999) closely approximate a
common least squares fit
This procedure is modeled after Draper & Smith's Applied Regression Analysis, Section 6.8 entitled
"Computational Method for Stepwise Regression.* This stepwise procedure was originated by Efroymson.
Residual Errors
A residual error is the difference between the actual value and the value calculated by the polynomial
resulting from the modeling process. The magnitude of the residual is the square root of the sum of the
square of the error in the X direction and the square of the error in the Y direction.
p_2 September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix G
APPENDIX G
Gridding Process
The process of generating a geometric mapping grid can be broken into four steps: (1) establishing a
transformation (usually a polynomial), (2) testing the one-to-one property of the transformation. (3)
gridding and grid reduction, and (4) the calculation of geometric errors due to the gridding process.
The process of establishing a transformation usually means a bivariate polynomial has been modeled from
a set of tie point pairs (pofyfit) or has been manually entered by the user. Other methods include a finite
element model used in the Large Area Mosaicking System and the Projection Transformation Package as
used by the remap module.
If the transformation is to be successfully applied in geom, using a geometric mapping grid, the
transformation must be one-to-one. This means that each point in the output space can map to only one
point in the input space and vice versa. Testing the one-to-one property is performed by the Jacobian
(functional determinate) of the transformation. This test is performed in the gridgen module.
When a valid transformation has been established, it is applied to create the geometric mapping grid. The
user can control the density of the mapping grid in two ways-with a default technique or by user
specification.
Default Technique
The default technique is to impose a 127 x 127 grid over the output space. This maximum grid density may
then be reduced using the tolval parameter. The tolval parameter contains the maximum amount (in
pixels) that the linear approximated X,Y coordinates can differ from the actual grid XY coordinates. The
grid density reduction is performed by discarding rows and columns whose grid points all have values within
the tolerance value. An example using a 5 x 5 grid rather than a 127 x 127 grid follows.
September 1992 G-l
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix G
Figure G-l illustrates the rectangular output image space mapped into the corresponding distorted input
image space. Figure G-2 illustrates the linear segments that approximate the same mapping of output to
input.
Figure G-3
Figure G-4
The elimination process for vertical lines (columns) and horizontal lines (rows) is done in the same way.
Elimination of vertical lines is shown in Figure G-3.
G-2
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix G
Vj Vj V4 V»
Output image in
output coordinates
Output image in
input coordinates
Figure G*3
Steps:
First, test if vertical line V2 can be eliminated by drawing straight lines from the points of VI' to the points
of V3' as shown. Then, according to the spacing of vertical lines VI, V2, and V3, linearly map the points
on V2 (V marks on the straight ones).
Next, check the X and Y distances between the x-marked point and its corresponding point on V2'. The
maximum of j X-X'j and j Y-V| is used to test against the tolerance value, where:
X, Y are coordinates of x-mark points and X', Y* are coordinates of points on V2'.
If all five maximum distances are less than or equal to the tolerance value, then drop V2 and check V3 as
in Figure G-4.
September 1992
G-3
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix G
Vt'
V3f V/ Vs'
Figure G-4
As before, connect points on VI* and V4* by straight lines and compute intermediate points. Compute
distances between points on the straight line and points on V2' and V3' and check the maximum distance.
If all 10 of the maximum distances are less than or equal to the tolerance value, drop V3, check V4, and so
on.
Going back to Figure G-3, if any of the five distances between the points on V2* and the points on the
straight lines (connecting points on VI' and V3') is greater than the tolerance value, retain vertical line V2
and continue the process from vertical lines V2 through V5. Test V3 in the same manner as V2 was
tested, and continue the process. Whenever any vertical line is to be retained, the process continues from
that line.
Vt'
V*' V»'
Figure G-5
September 1992
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix G
As an example, if tolval is 0.1, then the distance of the linear-approximated coordinates must be within 0.1
pixel distance of the transformation-derived coordinates. When the original transformation is linear, only
vertical lines VI and V5 are needed. On the other hand, if the transformation is highly nonlinear, no
eliminations may be possible. As tolval increases, more lines can be eliminated until eventually there will
be two lines (VI and V5); however, there will be noticeable geometric distortions when geom is run and
resampling generates an output image. Conversely, should tolval become smaller, more lines would be
retained. If the transformation is linear, any small value for tolval gives only two vertical lines and the rest
is dropped.
It should be noted that the grid expression for the transformation function is an approximation (locally
linear) and reducing vertical and horizontal lines is another approximation. Also, the example given above
is a simplified version of the process. Before the grid is actually reduced, geom buffering requirements are
taken into consideration, often resulting in a grid which is denser than the transformation equations and the
tolval parameter require it to be.
User Specification
The user has the option to specify the number of lines and samples in each grid cell From these, the
number of rows and columns in the geometric mapping grid are calculated. If the number of rows and
columns exceed the maximum size of 127 x 127, the number of lines and samples in each grid cell are
adjusted to fit the maximum grid size. The user is then informed of the adjustment Using this option, it is
assumed the user is aware of potential gridding errors due to grid density; grid reduction techniques are not
applied.
Calculation of the errors due to the gridding process
Sixteen points (4 in X by 4 in Y) located 3/126, 43/126,83/126, and 123/126 of the distance from one edge
of the mapping grid (output space) to the other edge in both x and y dimensions are checked for errors.
First, the input space coordinates of these 16 points are determined using the mapping grid. Next, the
input space coordinates are recalculated using the transformation equations which were used to create the
geometric mapping grid. Residual errors are then calculated between the true points (transformation
derived) and the approximated points (grid derived). These residuals are written to the geometric mapping
grid file and are an indication of the errors which occurred when the original transformation mapping
output space to input space was gridded.
The output geometric mapping grid ffle contains the mapping grid point values, projection and framing
information needed by geom to fill the output image's DDR (if it was available at the time of grid
generation), polynomial transformation information (if a polynomial transformation was applied), as well as
statistical information regarding the errors resulting from the gridding process.
September 1992 G-5
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix H
APPENDIX H
Geometric Transformations
Geometric Transformations involve spatial transformations to modify image geometry and grey level
interpolation to assign pixel values in the spatially transformed image. The spatial transformation is
defined by a geometric mapping grid (see Appendix G). Within each grid cell, a pair of bilinear equations
maps from the spatially transformed space to the original (input) space:
Xoriginal = aO + alx + a2y + a3xy
Yoriginal = bO + blx + b2y + b3xy
where x and y are pixel locations in the spatially transformed (output) space and aO, al, a2, a3 and bO, bl,
b2, b3 are coefficients defined by each geometric mapping grid cell
The geom function applies this transformation both directly using a one-pass, two-dimensional algorithm
and also with a multi-pass method involving three one-dimensional passes through the data.
Pixel values in the original image are defined at integer locations. Since the pair of bilinear equations were
often given map to noninteger locations in the original image, a method to determine pixel grey-level values
at noninteger locations, based on surrounding pixel values, is needed. Resampling methods available in the
LAS 5.1 include nearest neighbor, bilinear interpolation (a 2x2 kernel), parametric cubic convolution, (a 4
x 4 kernel) and a user-entered resampling weight table with a kernel up to 16 x 16. Resampling methods
are accurate to 1/32 of a pixel, except nearest neighbor, which is accurate to 1/2 a pixel
Resampling kernels which are separable into horizontal and vertical dimensions can be represented in the
user-entered resampling weight table fite. Each kernel dimension has 33 entries, corresponding to the 32
increments between two pixels and both end points. Resampling weight table files are generated by the
rtable module. Rtable generates a sin(x)/x kernel of user-specified dimensions and also generates
resampling kernels from an inverse point spread function. Presently, LAS cannot generate an inverse point
spread function; it must be done externally. This is the manner in which resampling kernels for the
restoration process are generated.
Park, S. FC, and R. A, Schowengerdt, 'Image Reconstruction by Parametric Cubic Convolution," Computer
Vision, Graphics, and Image Processing 23, Academic Press Inc., 1983 pp., 258-271
Friedman, D. E., "Two Dimensional Resampling of Line Scan Imagery by One-Dimensional Processing,"
Photogrammetric Engineering and Remote Sensing, Vol 47, No 10, pp. 1459-1467, October 1981.
September 1992 H-l
-------
-------
MRLC Consortium
Documentation Manual
January, 1994
Appendix I
APPENDIX I
References
Map Projections:
Clarie, Charles N., State Plane Coordinates by Automatic Data Processing, Publication 62-4, U.S.
Department of Commerce, Environmental Science Services Administration., Coast and Geodetic Survey,
Washington D.C., 1973.
Synder, John P., Map Projections-A Working Manual: U.S. Geological Survey Professional Paper 1395,
(Supersedes USGS Bulletin 1532), United States Government Printing Office, Washington D.C., 1987.
Stepwise Linear Regression:
Draper, N.R., and H. Smith, Applied Regression Analysis, John Wiley & Sons, Inc., New York, 1966, pp.
171-172 & pp. 178-195, and portions of Chapters 1 and 1
Draper, N.R., and H. Smith, Applied Regression Analysis, 2d ed, John Wiley & Sons, Inc., New York, 1981,
pp. 307-311
Geometric Transformation and Resampling:
Friedman, D.E., Two Dimensional Resampling of Line Scan Imagery by One-Dimensional Processing,"
Photogrammetric Engineering and Remote Sensing, Vol 47, No 10, American Society for Photogrammetric
and Remote Sensing, Falls Church, Virginia, October 1981, pp. 1459-1467.
Park, S.K., and R.A. Schowengerdt, Image Reconstruction by Parametric Cubic Convolution," Computer
Vision, Graphics, and Image Processing 23, Academic Press Inc, Duluth, Minnesota, 1983, pp. 258-272.
General:
Gonzalez, Rafael G, and Paul Wirtz, Digital Image Processing, 2d ed, Addison-Westey Publishing Co.,
Reading, Massachusetts, 1987.
Press, William H., et al, Numerical Recipes—The Art of Scientific Computing, Cambridge University Press,
New York, 1986.
Quirk, Bruce, et at, Selected Annotated Bibliographies for Image Mapping: Geometric Registration,
Resampling, Contrast Enhancement, Spatial Filtering, and Color Calibration, USGS Open-file Report 85-51,
EROS Data Center, Sioux Falls, S.D., 1985.
September 1992 1-1
-------
Interoffice Memorandum
DATE: August 31, 1993
MRLC Consortium
Documei
HUGHES
HUGHES STX CORPORATION
A Subsidiary of Hugh** Aircraft Company
TO: John Dwyer, NALC Project Manager
THROUGH: Supervisor, Digital Data Production
FROM: Digital Data Systems Analyst
SUBJECT: Use of DIG for NALC Image to Map Registrations in the U.S.
The NALC project uses MSS data for image registrations. The final product is
a 3-date data set (triplicate) that should be generated in an 8 hour work day.
The project is experiencing some difficulty with timeliness and one of the
problems encountered is map handling.
Map handling causes problems for several reasons:
1) There is a shortage of personnel for pulling and filing maps, therefore
the geometric registration people must get their own maps and this adds
time to the processing flow.
2) The map file program selects maps based on a number to select, not taking
the map distribution or feature availability into consideration. This
often results in either too many or too few maps being pulled initially,
both wasting time.
3) Using maps as a source for ground control point selection is very time
consuming. First the map is pulled from the warehouse and delivered to
the geometric registration personnel. The map is then registered to the
table, a point is selected and the operator will determine if the same
point is visible in the image. If the point is visible the operator will
select the point on the image that she believes corresponds to the
selected point on the map. This can be difficult when the display and the
digitizing table are several feet apart. If the point is not visible on
the image the map is removed and the process is repeated until a good
point is found.
The l:100,000-scale DLG is available over the conterminous U.S. This is an
easily accessible source material for use in geometric registration. The
source of the DLG is the l:100.000-scale topo maps whose accuracy is
~50 meters. There are no claims for the accuracy of the DLG itself. While
the accuracy of DLG is not as good as the l:24,000-scale maps the techniques
used in the registration process compensate for the accuracy of the source
material.
The technique used in image to vector registration is an area overlay
technique. The operator determines the best fit of an area of vector data to
an area of the image. When the best fit is achieved the system picks the
point by matching the node closest to the center of the window to the pixel
-------
MRLC Consortium
Documentation Manual
January, 1994
immediately beneath it. It is much easier for the operator to determine if
enough points have been selected and if she has an appropriate distribution
because the overlays will not need adjustment when this has happened. The
elevation values can be obtained from the OEM data so relief corrections can
still be applied. The final verification will be done with the l:24,000-scale
maps. This will still substantially reduce the number of maps that need to be
pulled for each registration. The image overlay technique is also much faster
than the image to map registration.
The image to vector technique using l:100,000-scale DIG should not affect the
accuracy of the final product. The final verification RMSE on a test data set
was .81 pixels for image to map registration and .89 pixels for the image to
vector registration using the same verify points and a 60 meter output pixel
size.
Using image to vector (DIG) should enhance turnaround times and the accuracy
will always be checked using an image to map verification so I am recommending
that the l:100,000-scale DLG's be used whenever possible in NALC image
registrations.
Brenda Jones
Concurrence: Concurrence:
Glenn Kelly Daniel Stelnwand
Cartographer Senior Scientist
-------
MRLC Consortium
Documentation Manual
January, 1994
Status of using DLG as a source material
Have tested mainly MSS data
Results have shown that the technique used in the registration -
vector overlay - compensates for extra error in the source material.
Have results on one TM scene that shows that the same will
probably hold true for TM data.
Image to map registrations using 1:24000 scale USGS topographic maps
and image to vector registrations using 1:100,000 scale DLG result
in the same overall RMSE on the verification using 1:24,000 scale maps
as the verification source.
-------
L
I
N
£
S
MRLC Consortium
Documentation Manual
January, 1994
85074116415xO_029_040_registration
Pt_16«
500
1000
1500
2000
2500 -
A: •Pt-12«r.-vvr. adpt o: -
pt_l«
Pt_9V.
pt_15.
pt_8«
pt_14>(
500 1000 1500 2000 2500 3000
SAMPLES (SCALE = 400 RMSE = 0.859371)
-------
L
I
N
£
S
MRLC Consortium
Documentation Manual
January, 1994
85074116415xO_029_040_verify
a.dp.t .<>.-.
1000
2000
3000P
4000
5000
6000
1000 2000 3000 4000 5000 6000 7000
SAMPLES (SCALE = 400 RMSE = 4.849487)
-------
MKLC Consortium
Documentation Manual
January, 1994
XXX
o oo
aaa
OOO
xxx
• •DM
vet.
0.1. V It
v o **
» c •> x w
OH- 3 • •
OOO
mmm
mm xc
•»• A O
o f
ooo
IMIMIM
OOO
u u u
B B B
c ce
c.
•
a
e.
o
•
u
B a.
w i. v
U V W
o • u
4.3 B
a i
XV
OLD L.
• *"* O f •—•-" I.
V »*--> f m v_< o
• C U • il-D B*.
« ll_i -> 003 W
*•<. a« <. « OMC.O
«•• O4J • C II t- (. •
c • i- • t. aw • «
•^v • • 3 o j: -o
O OIC 3 O U OWO O
av-^ • c v c « w
•w «) f X C — O**
e ti t)c«*j— _>cat»
^-^- l.->-r- O ••»• tl •
O*< B.O • L. XL. *O U U_<
a •>• H-J3 O O 3 U*-
•O C t. » l~ BT)B
w w • own t. i w
v m
mm-
• c *
c •
** O O • •> !>•*• I
•• B+'O^B*- I «•<-•• *•
• TJ «. U l>— "D 3 3*Jt)
tl V*Jt * 00 0 X C •
«i—• c oo o o t> u B w—• aa-c
«"•• oo ov avaa*
f-u. w oo L. x a » v -»
u. _• OO O-w-iC «•-••- •
C— mm B x-< O XOfl 3«^> B
C Oik OO —' «» B-r- L. >B U
— *J-O OO Xt- «i-«J • X *>--*>-< tl O
4-> BI- ti-/«—'-D o>c a o-* «-j"->
• U (. >-• a W— C O E C • E C II •
OOO •• f- « E u «! «> O) 3O31-I)
O-J —" *• 3 O UJ3 1- • « C C«-> C w
_l a 4J— u C4J O 3*"<-i O « 3 • X«
*• C I*** < • 3 U w «V*"r- E • B—> U
*J c-r- «« CBIIII 4-1 I I I »i-
C— Q 4-> U »- I I I I Bill 3-D
•r- O O. +* o *'*'*'*'*'*' -J*>«J«OC
OO. a L. Q.QOOOO t> O o U"""-
a. E vw cwwtiwww t- ti t) t) >
.Jl) wi- o O O'l'i—vi^ U ^ UX
ex -^oovuvv owvvi-
3 ae xx
aw o aa
C 3 V —•-•
-1O19 «
a
X
•er-ooo-x
-------
MRLC Consortium
Documentation Manual
January, 1994
c i
I
o-o i i I i I i I i I i i I I I
I 0.0.0.0.0.0.0.0.0.0.0.0.0.
II
o 10000000000000
— ''O I fwOK(«-J>-K»OO«--*«-
• a i ^«or-«-iArwir»cor-OO>a>r-
3** I .............
-o«- i oo«-voaaao*-oao
i- C I
• at i
« • i
a i oaooaooooaooo
••11 it it it
_> • i
• •0 I
i »-<»»^oor-cocoo>f\iinr>..«-,-
I O><»*-Or»O
I Om«Om«M'»«-OKO>«r>O>fwir>
• I
u
•
a
3
O
O. (M
C
O. C <
« o
• ••- +
4J * X >•
C -r- ^N ^
4) X •* in
i- o ^> v
u <. < m
•^ o.
»- a + +
• a
U*
0.10
*>
3
a
CO
M c
i «-r-OKiOT-OK>
I «f>fUi» Kl •«-!>.•»
I «(\jm«-KlO«l«r-«--*.»Kl>OtO
..............
I fM«Mir>«o
i o«n«-incwnj<\n-<-
«-«- IM«-
M «M
C «
e >••
ux>-x>-
«M»-^- mr-iftK>«-eoo>O«0
«M
«o
0>
CIMMin m
~»OOO>
O>
in OK) to
o*-O> A
mnio •»
ION-IO «-
x i «-omcomcoco«MO»-K>wco
C I «*O.»Ul«1«M«-«-«-«ftO«OB>«-
O
< CD
e <-
OOl XX
OOOOO
+ i i i i
x
^
a
C
C
v
I
O O
L. L.
Will
OOO
OOO
OOO 0DO>
OOO «»«M
• • • oo
OOO OO
go
o
— do*
• •O
•DM .. «
i- II _>3
MOC «TI
II 9>^
oe «
K>mfOKtin
oor»mojoO
-O C C
E T3
L. v
o *•
--
** +* CpjOOOJr-
•»m-«-f')
.
>- i .rcMO>oooeD<»cx<-c\j«-K>o mm co <-CM
*N i ««O>m^^l^«N • * • O-^
*• i »o«*r* o^r
_l I ^> OCX
i »-rj•
3 ~- s-^« ~-
1C X IIU. u.
O « O
u. X U
rviioo^O
.....
•OCKi-rj'-
I i I
<-> it w
CUV
o*-
L. ii u
II l> II
.O I. I.
S 0>D)
3IIII
ZOO
3
T)
«--* TIM
—ii ............. -- 1>
II O. i m >*-•* in nr>m^ MK
u E i «-«-K>OK>mocO'OO»OKiao 11
•• i »» ON. *-<*««- too «MCJOirv K ti
bin i «-»- ro IMT-IMK>IM(M(M — < -•
10 i it au
' ce«o
•o
II
ii»r» «aw MCC
(.«•» C •« V
UCOO t- « Bl>
KI-* ^ o
I.C • O f
ot-« a w
3
O. I CO COO COO -COCO Or-tO OKI
wiii .............
3 c i -•-«• m-»o>c\jr- K> 00
Of I mO«*MT>m»w
WO.
ii ce
X
•
••-33 C
aaa M
a 3 3 -o
zoo -r-
L.
O •*-•»- O
f O O
U II
w u I. O>
II II II •
mjajo L.
O E E II
II 3 3 >
wzz <
O
a
w
X
-------
MRLC Consortium
Documentation Manual
January, 1994
*•»"»•
i- XX
• 'o'o
«- a. a
o'o'o'
XXX
mm
^^nw
•*
^•P**-'
••«•••»
t-r-r-
OOO
minin
CO 00 CO
OOO
•**^
000
•- 3 II •
• • •
*• xc c
i- A O •
o i- a
X:WDW
u ii • c
t.f «- u o
ii a**-^ *•
• c u •
ii u-- ->
•>- L. a* t. ii
c • t- «• i. atis«
-^11 ••SO .C T>
o e»c 3 o u owo o
a*o-^ • c v c ii w
*> l> ^- £ C «- O*'
c ii iicB+>f- ->caa
T"T- t-->-r- O ttf II •
O«"«^B>-X<- V (. u— <
a ii • H-.O o o 3 o»-
•D c i. » i. m-o «
ii ii • ova L. i »
T- *• o a— > c«ii(.iit.ii>
w U O • II !>•*" I ll£ II <- —
u
c
• a
VI. tl
u w u
o •» u
t. a •
a i
xii
ota I.
C
• iiw o
«T>*«-
U O 3 II
II •
• C II «lt) U u II^--O D 3WU
C • IIDMJC Jf U O U X CD
II • tJ** X •.* • • «*• tl.O «*•
0-.C 00 O O II II • l>~> Q.Q-C f O->
—•»• oo OB an aa« • « x** aa
.^>kii oo ux a. H ^i- m+> x— i xi>^-
C-~ mm • x-< O XO"O D^Jl a ^3 II D
coo. oo -j t> m-r- 1. mm u w-
OO
not.
U OU>
0-1
X>»- •»-*• <0 X*'-'*"-* II O
ti— < w^-o ac a o— ' B— "^>
>_< • ««- c O E C c fj a w «
"- • E (_ tn * » 3 O 3 L. II
*• 3 o ujj i. a w c c** c +>
o C w O 3 *»^£ O « 3 • X«
< e 3 u u wo w^ E B E— ' u
E 8 I I I I *J I I I !••<-
*- i i i i a i i i D'D
on. B t. o.o.uuuu ii o u o-<-~-
O. E !>*> CllVlllllltl L.UIIW>
II *"^ O U Oi-w^»^»S U ••-"•'-"••-HI X
ex •^uuiiviiu o t> v t> i.
ii ii ^«a:oea:K ucEcecea
*J 3 II « •
sac ^Ji
a*-, o a a
CSV —'—•
a
x
•oi^coox
-------
MRLC Consortium
Documentation Manual
January, 1994
c
f
OT>
f i «-
•o
o
I I I I I I I I I I I I I I
I *l *l *• *> 4-1 *> «•> *• W 4-1 4-1 *« *l
i aaaaa.aaaao.aaa
looooooooooooo
i co«-«co»rcBOrwa>«-
1 10 O>KI «<>««••* »O OB in « 10.* h-
•O-- I OO«-«-OOOOO«-«-«-0
••- C I
•Id I
•EK I
*l
"a i dd«-«-dddo'ddddo'
t m i
IM I
J I
i ca »-«-
i
c ioo
f i > i i i iii
• a i ^«-
WM I •»O>f».«-O> c «
o >•«
ux>>x>-
3
a
o
OOO
+ 1 +
e
o
(M
• X «• m
•r- o +* *-
u b < m
••- a
*- a * +
o
i
«
o
»
carom m
IW«*IM «
OIMIM m
O b
u o
mo>o «- <-
IM^v- IM
I I I
II
X
IM IM
< a
c b
o a xx
— D *% ~*
v -a «- «-
a >» »^
B -o < a
b II
o *- + +
IMOv
II
a
WO O
c<
«
Ul
in
tiu. u.
o
u
llf -r-
3 • B
o c
*•" IIII
C II II
OfVIOOO Ow-<^
H- O O
o
M W
b IIII
IIIIII
Abb
3 II II
zoo
a
II
oaooooo«O>OO«»«IMO«-09«MO>
I
X i ^Omoom«<»iMO«-ptiM«B
'V I IM<*m<*(MO>OOKIOB'*»tfkO
a i moo<«oO««
C I •*O.«MIT>«M«-^«-mOOOfl»^-
O I «MKlK>«raM«M
^ I
I
I
I
•*••• «-mo
IMK.O
f +*
ooo
-< I I
••c
3
•o
•e« •• •
-> I -^11 —"3
Ii a i r-«-h-o.«-««-ift«-cM«»ioK>inO>IOOlO I) 3t-
10 i ii aui HOC
i c Bin ii
uaao -r-
K>< -IIAK —
ii « a
— C CM
•^ •• O
••II V
•o •>-> «
•»- v a —i
L. C • O
CPV • a
3 I CMO>*IO«fM*>h.r-lA*-(MIO
a i »-^ IM^
I
I
-lino: ->
ii a
II C B
a f •
b
II X
e
o
*» o
c a
ana ~ u
Q4-« II
• 9 3 -D *-•
x o o ••- c
(. M
O^-t- (9
•^ O O , -O
L. II f
*• t- «- a <-
II V II • (P
SJDJJ (.
o e • it x
II 3 3 > •
-------
ui i r>-o<*O*--*
I «-«-MOIOK>O«0«OK>K>CO
I «*Okl^*-O>4>*-ltOKt I OOOOOOOio
OO T3
go co 3
i-li-i O1
OO
OO ^ «
II
OO o«
u u k (i li i eg »-^^(MIO«-T- «M^
— '_ O "DO
• • L.
e c t. •• c cs •
• • O ••••- •» •• «- UI I tl —I
coo w o. i
33 V-t-'—'CVV C I lAOlOinO (>>«--«••* CD ?<-!«•*
££ » V Qrr- U • < I «-»-|OOrOKlOa>«OlOKlB>
i_u_i cc o »-r»«-»H><-KioiM(UOm
e • a *i c 3 i «-«- KI (M<-ruKirururg
•w C tl V^ O. I
i- Ci- Z I
•• II *• O>C*> *" I
v-j O*JH- ^- M .. ui i r>- f- nj r- rvjru pj ^ CM ro r». -r- c £<-"»- •«- H z i cocooco-MOO
4J f I) C ^ +•» —* 4-1 I IA O **" ^- ITMA N. «O N- »"*ft CM «>J
4-> c x li— 'i- v « i >- ii ac
lit- C t-
-D— ' U ft- t. I
O— • O Q—> O • >- ' O»-IM-*ir»O
30. uui»i»c«i3 Zi *-IMI»><*K.OBO>^-«-»-^^^
Q.U .wccaC4-> —i I I I I I I I I I I I I I
C3 £ • II II U II • O I 4J 4J 4J 4J 4J 4J *J *J 4J *J *J *J *J
•-•o »- to »> to in u> o a. i ao.aa.a a.G.aa.a.o.aa
-------
MRLC Consortium
Documentation Manual
January, 1994
a
•>*•
a.—
*• u
ti i.
oo
x x
mm
OO
00
00
•O
00
O>
MM
00
o o
c c
JSX1
.0 JO
33
OO
00
00
•O
0>O
L. C.
O O
o o
u u
o o
00
00
00
00
00
OO
OO
MM
UIK -N
Will X
act- -
o>r» ••-
4J U CO
U tl HI
—•II <
II M Z
M .. M
*• C O
4-1 C O CC
C-«- f O
f- O 4^ O
o a u u
a ti
u —i o
f-4J L. Z
4-> O. O.
*j 3 4J a:
3 O. 3 19
a*- a o
c s c ui
MO '-' (B
4^4^ O **
u a
a.
• «
4-> •
• C
••••D
U u
tl tl
O O
U. (.
aa
c
o
u
tl
o
fc.
a.
3
O
a.
z<-
o
I-
o ti
wo
-> o
OO
4J tl
u u
•o «
ti a
C E
33 t- c
XX C C
>*> u O U
3 O O
«a o u «
34< U 3
as w a
4J O 4J 3 W
3 3 a 3
o e a«> o
i- C 3
ti
M-I ti ti
ti a t L.
C E 33
II tl
H N »-«^
--••- o o
3 3
E E
a ii —
c -r-~ v w
K o c e xx
3 icx a. a
ii ii
m E
•o
o o
«-4- C
'c'c "o
33 a
«M m CK to cooeo»«r ~ ~ "
«-MMIMM^«-MMCM«-MM
Kl IO Kl tO IOIO »O Kl Kl IOIOIO Kl
r- CM 10 •* MO O"- •-<-»-»•»-
I I I I I I I I I I I I I
aaaaaaaaaaaaa.
-------
MRLC Consortium
Documentation Manual
January, 1994
« iw i «-
•O I I
II I
> II I
O~> I
e a. i
v m i «-<*
Ml .......
a i r^«o IOOOOOOOD
II .......
K 000000-*
OOOmOOO
> I
(I I
woo
> aa
odd
M XXX
*• mmm
C «-w
f •»**•
a »-2l?
O
I
_« I
• I
*•»« I
in o i oniioo>**O>-««
k II ^ v* I
UI €> I I I
e i
t- i
-I I •» «-»•«- O-T--
o->»«-«ir»i«
oo«*-oooo
oocxoooo
*• minm
• ooeooo
U r-ir-ii-i OOO>*OOr-O
f • 000 O v - -
«- « *••»•* O
i- O OOO m
k a o>O>O> o
ii • IVNIM N-
>•*• OOO
u u u ••
•O*«
II 3
•«- o
"a*
c e e
3
er
333 -< •
^J=£ -> II
I-UJ
IUI-
XIU
ton
010
OCO
mo.
ui
com
'D-' I OCX
i) a i ae
> • i
— c i
bio i ....... KM
k i o.^-«-minm t-uiwoj
• in
«ii k
_._, o
LIU
ll»-
4^UI
ii a:
O C
k •
k •
O « -iff
Ok i- O O
• c c
k ••
o*
tl II
•o ii
k
k •• c a
o if- u •• c
«k -Out Ct-
C II II II --
II*'-' C II II
• II OH- k II N
• C O Nf-
•t-fr- m a 3 li ••- •
Wfcl *-H- w C VO M
C C C»-»- •»• C-«- —•
•It) OWII w O > C —i tl
EB f P O O*"^ i- •• II X
II II *J U U C £«-"*• w X"-
ou a DaO^k-^a
BB o wo '^T-V ti ii a
aa H-k • j: iiw^ ti v o
• ^- ci 3 Kl f-liaEDa
k > k *—*> kf- • E
II C O w O*- II C B-r-
>-*• ^3kk^—*.
tl tl *•*'*' O- aw
tin ccc caiitiu«C3
ee.ce. «M» k- COm
i o>«Kto>m
^ i inT-
3 i co«»-O>eD«r>-
CUIIUO -OKKKIM
XKIt CIUUIUKM
h- UICO Ml-l-f-in
• UIUIIU •
minm
« I
IONIA
•(>>«- «
— •
Nino li
i m • x
O) i «- cor- «-««-
C
O I
«- i i o a
omO I r-O»-
f- I I
v> i o> -» •» o IA r^ r-
L. L. L. L. IIUIUI L. L.
ii li O O i. into o O
• k k
li «
t-cca: k k
O v w
_< i
M M K> ro K> Kl Kl
•O I «-
"i*~iT"i r\**\
w IWWW*'*'*'*'
aiaaaaaaa
k a a o k • •
O *^ ** > O *^ **
kkOO OkkOO
k ii
k »•*-
v ti o o
ii aiuui
c Bin in
li k ii
k k *-*•
II « O O
k -*
ii« atuui
«j c Bin in
.JMOEK <_IMKK
-------
MRLC consortium
Documentation Manual
January, 1994
*• u
— c
II t.
oo
x x
mm
°.§
OO
•o •
00
« C
4J C O
ct- f- o
«- O 4J O
o a u u
a ii
II •*"» O
Hi- O •*
f-4- U X
4-> a. a.
4-1 <
4- 3 4-> a
3 a 3 u
0.4-1 Q. o
C 3 C UI
MO «- 19
c
o
o o
L. b
o.a
3 3
e c
•*•*
U O
II 11 I
O U l>
1. UIT3
a. -i o
ou E •
4- K 33
3 O. II E E
a c i—f
4-< X O C C
O 3*" XX
O4-
f O
4- II
U i.
«-XJ
TJ II
») D.
C E
C C
II II
II II
X X
O.O.
3
4- a
as
*> o
3
O C
ii a
C E
O O
&. u
IIII
.0.0
E E
3 3
C C T9
f D O
•B L. O
l_ O U
O O
O U *-
U 3
w a
4-3 4->
3 a s
O.4J O
C 3
i- O
II II
t. L.
3 3
M (•
• C
II II
E E
tJ
<•-«*• *^
O O
V
4-* 4-1 C
C C O
as o.
«-o«-ooo
-------
MRLC Consortium
Documentation Manual
Interoffice Memorandum
HUGHES
HUGHES 6TX CORPORATION
A Subsidiary of Hugh** Aircraft Company
RE: OAB8-21 August 12, 1993
TO: Joy 0. Hood, MRLC Task Leader for Landsat TM Processing
FROM: Senior Scientists, Sensor Systems
SUBJECT: Landsat Thematic Mapper Terrain Corrections in LAS
The purpose of this memo is to provide a status of the on-going investigations
being conducting into the geometric correction of Landsat Thematic Mapper data
for the MRLC project.
A realistic product for this project is a map-projected, terrain-corrected
image with ground control points applied. Expected accuracies from this type
of product should be in the neighborhood of 3/4 pixel (provided the resulting
pixel size is close to the original 28.5 meter pixel), but will vary with the
conditions of the ground control point reference material, the availability of
selectable image features, and the accuracies of the digital elevation models
used. End-product pixel size and map projection are not yet defined.
Figure 1 contains a plot of the displacement of a pixel in the sample
direction (y-axis) as a function of scan angle (x-axis) for given elevation
errors. These elevation errors may be caused by errors in the DEM or in the
case of this memo, due to the image being mapped to an ellipsoid (i.e., when
the image is not terrain corrected). The figure shows that a elevation error
of just 200 meters at the end of a scan-line causes about a 1-pixel location
error (~30 meters). In this figure, the relation between the elevation error
and displacement error can be scaled by multiplying the displacement error by
the same amount that the elevation error is multiplied (i.e., at end of scan a
4,000 meter elevation error causes a 600 meter displacement error). In the
Rocky Mountains it is not unusual for elevations to be above 4,000 meters. At
that elevation the displacement error at end of scan would be about 600 meters
or about 20 pixels. Clearly this is not acceptable.
Ideally, this type of processing would be handled by EDC's NLAPS system, which
is to be installed late in 1994. NLAPS operates from raw TM data and will
produce map-registered, terrain-corrected products. However, due to the need
to start processing MRLC TM data soon, an alternate method of processing is
needed.
The Landsat Thematic Mapper sensor is a complex sensor to model correctly.
Raw (level-0) TM data is generally considered not useable by end-users due to
-------
MRLC Consortium
Documentation Manual
January, 1994
a number of geometric distortions. It is for this reason that TM level-0 data
are not standardly offered to users. The lowest level of processing offered
as a standard product by EOSAT is referred to as a "P" product. This is a
"systematic" product in that known geometric distortions due to the sensor
have been corrected, earth rotation applied, and is map-projected (either UTM
or SOM). These data have been resampled in this process. Although internally
very accurate, these data have not been referenced to the ground so a shift of
sometimes many pixels may be required. These data have not been terrain
corrected.
This "P" product above is the input to EDO's TM processing. EOC does not
currently terrain-correct TM and all TM imagery processed at EDC is currently
resampled for the second time when ground control is applied. EDC
investigated the purchase of software for terrain correction of raw TM
imagery, but the cost was prohibitive in light of the NLAPS system coming on-
line in a short time. We therefore looked to methods of modifying EDC's
current processing of "P"-level data to provide a higher degree of accuracy by
terrain correcting it.
Based on investigations and details learned during the development of
processing prototypes for the North American Landscape Characterization (NALC)
project and the Humid Tropical Forest Inventory Project (HTFIP), we found that
a Landsat "P" product in the Space Oblique Mercator (SOM) projection is very
close to a raw image with sensor artifacts removed. This results in an image
with scanlines roughly aligned in the sample direction, and this is an
assumption made by our processing prototype. In addition, this imagery has
been corrected for earth rotational effects. This results in an image where
nadir Is not down the center of the scene. We are working from SOM projected
"P" products resampled with cubic convolution resampling. Cubic convolution
was chosen since many of the systematic corrections contain sub-pixel shifts
which would not be fully corrected using nearest neighbor. (Other resampling
methods are not currently offered).
Due to the total amount of terrain-correction required being dependent on the
sample's distance from nadir, the location of the nadir pixel in the image
must be calculated. The earth rotation correction causes the nadir pixel to
move from right to left as a function of line number for descending passes and
from left to right for ascending passes. The total number of pixels the nadir
point moves is found using the equation below:
shift-We*RE*cos(latitude) * pass_tine * cos(heading) / pixel_size
Where (We) is the sidereal rotation rate of the earth, (RE) is the earth's
radius, (latitude) is the scene center latitude, (pass time) is the total time
of the scene in seconds, (heading) is the angle between north and the
direction the satellite is moving and (pixel size) is the size of the pixel in
the along scan direction. The (heading) angle is found using the equation
below:
-------
MRLC Consortium
Documentation Manual
January, 1994
heading » asin(cos(inclination) / cos(latitude))
Where (inclination) is the inclination of the satellite's orbit. The standard
SON P-product image is 6967 samples wide by 5965 lines and the valid data in
the image are 6458 samples wide. Figure 2 shows how the valid image data are
positioned in a standard SON P-product image. The offset in Figure 2 is found
by:
offset*(total image samples - shift - total valid samples) / 2.0
The slope of the nadir pixel shift is:
slope « -shift / number of lines
and the intercept is:
intercept * offset + shift + total valid samples / 2.0 - 1.0
Thus, the nadir pixel location is:
nadir location * intercept + slope * line number
Once the nadir pixel location is found, the ground-range distance
(S) (see figure 3) which is related to the scan angle (d) is calculated from:
S • pixel_size * (sample - nadir location)
and (d) is found using the equations below;
s - S / RE
LOS-sqrt(RE*2 + (RE + ALT)*2 - 2.0 * RE * (RE + ALT) * cos(s))
d - asin(RE / LOS * sin(s))
The change in the scan angle (dd) due to terrain must be found using the
elevation (h) for the pixel in question. The following three equations are
solutions to find (dd). The first equation uses the small angle approximation
of sin(x) « x. The second equation uses the small angle approximation of
sin(x) « x and cos(x) « 1 - x*2 / 2. The third equation is the exact form.
dd - (RE + ALT) * h * d / (RE * (ALT - h))
dd « (RE + ALT) * h * d / (RE * (ALT - h + (RE+ALT) / 2 * dA2 *
((RE + h) * (RE + ALT) / RE~2 - 1)))
dd « atan((RE+ALT) * sin(d) * (1 - (RE + H) / RE) /
((RE + h) * sqrtfl - (RE + ALT)*2 * sin(d)^2 / RE"2)
- (RE + ALT) * cos(d)))
-------
MRLC Consortiura
Documentation Manual
January, 1994
The change in scan angle (dd) is then related to the change in the earth
central angle (ds) which is then related to the change in the ground-range
(dS) which is related to the change in the pixel location. Or in equation
form:
2" « asin((RE + ALT) * sin(d + dd) / RE)
ds - z" - s - d - dd
dS - RE * ds
change in sample « dS / pixel_size
Thus, the resampled brightness value located at input sample plus the change
in sample is used instead of the value located at the input sample in the GEOM
process.
For an implementation of this algorithm, we have extended the LAS Geometric
Manipulation Package to perform terrain correction of Landsat TM-P data. This
involved writing an all-new GEOM function named GEOMTP (GEOM Tm, terrain
corrected P data). GEOMTP currently resides outside of LAS, but will require
little effort to integrate into the system (and will be integrated before
testing progress much further). GEOMTP requires a TM-P scene in the SON
projection, a DEM, a processing "mapgrid" file (in place of the geometric
mapping grid file) and an output file name. GEOMTP currently implements
Parametric Cubic Convolution resampling only, but could easily be extended to
other table-based resampling methods, including restoration resampling. The
performance of this function is currently comparable to the existing LAS GEOM;
no optimization of the code has been done yet (there are a lot of those pesky
trig-functions in the inner loops to calculate scan angles, etc.). The GEOMTP
algorithm is described in the attached data-flow diagram (figure 4).
Prior to running GEOMTP, ground control points are selected in the usual
fashion. These points are framed to the desired output space which sets up
the output-to-input space relationship for the modeling process. Prior to
modeling, however, the input image coordinates must be adjusted for relief
effects, as they were selected from the terrain-distorted input image. This
is accomplished by running RELIEFTM, another extension to the LAS Geometric
Manipulation Package developed for this prototype. (During this study, we
found that the assumptions made in the current RELIEF function are not
adequate. RELIEF assumes nadir to be straight down the center of the scene
which is not the case. This can result in a correction error of more than a
pixel in a high elevation, off-nadir feature.) After the points are adjusted
for the effects of relief, they are modeled with a first-order polynomial.
The use of a first-order polynomial simply applies transnational, slight
rotational, and slight scaling errors as described by the ground control
points. Higher order distortions due to sensor characteristics were
previously corrected for in the TM-P product and distortions due to terrain
are corrected for in GEOMTP. This first-order polynomial also incorporates a
projection change which may or may not be a first-order transformation. This
issue is currently under investigation. A data-flow diagram for this
correction process is given in figure 5.
-------
MRI.C Consort ium
Documentation Manual
January, 1994
To date, we have worked with a scene of TM from path 40, row 35. This area
was chosen because it contains features below sea-level as well as mountains
of significant elevation. It was also available immediately and ground
control points had already been selected by OOPS. We took the ground control,
framed it (TRANCOORD), adjusted the control points for the effects of relief
(RELIEFTM), and created a first-order polynomial model (POLYFIT or EDITCORR-
HAX). OOPS supplied us with a 3 arc-second OEM of the area, which we framed
and resampled to our output space at 30 meters. We then terrain corrected the
image with GEOMTP. We had OOPS do a verify, which came to an error of
0.8 pixels RMS. Subsequent refinement of our algorithm enabled us to obtain
0.75 pixels RMS. This same scene was registered in the traditional, non-
terrain corrected method and resulted in an RMS of 2.04 pixels before the
relief of the verify points was taken into account and 1.5 pixels RMS after
they had. Our results look encouraging, but we feel we must test these
procedures on a number of different scenes before these algorithms are final.
To further test our results, the following tasks are planned. First, the
scene used above is being terrain rectified on two other systems. Glenn Kelly
and Dean Gesch have "volunteered* to run this on the HELAVA stereo
workstation. In addition, Brenda Jones will test the scene on the PCI system,
of which we have an evaluation copy. Both of these systems use a
photogrammetric approach. The results should prove interesting. Another
planned test is over path 33 row 33, a scene which contains Pike's peak—an
interesting feature for terrain correction purposes. We ordered a "P" scene
>from EOSAT, resampled with cubic convolution and projected to the SOM
projection. We also ordered an EOSAT terrain corrected, UTM product for
comparison purposes. The "P" scene has been delivered, the terrain corrected
scene has not.
The process of collecting ground control for this project will be an enormous
task. We have recently been looking at the GSFC Ground Control Point Library
(GCPLIB) and think that it might have an application in this project, as well
as in the NALC. MSS has excellent coverage over the 50 states, but TM is much
. more limited. We have plotted the locations of the 213 path/rows of TM over
the lower 48 states where the GCPLIB has control points on a Landsat WRS map.
In general, the east coast and midwest are well covered, and the western U.S.
is not. In the next few weeks we will be investigating the feasibility of
using these GCP's—the biggest question is their accuracy.
Another issue which needs to be addressed is the double resampling issue. All
three systems being investigated (Our LAS mods, the HELAVA, and PCI) start
with TM-P data which has already been resampled once and must be resampled
again. This double resampling is probably unavoidable, but the question which
comes to mind is what's the best method to use for the next resampling. In
the mid-1980's, EDC funded research with the University of Arizona into
restoration/resampling. We think this may be a good option, but there are
some problems with running restoration in a batch-like mode. There was also a
study done in the late 1970's by Benner and Young at IBM on twice resampled
MSS. We plan to do a brief investigation into this topic by reviewing these
«dies with the MRLC in mind. Although this double resampling issue poses a
blem, it does provide the user with two levels of useable products—the
P scene and the fully rectified version. Although already processed, the
-------
MRLC Consortium
Documentation Manual
January, 1994
TM-P scene is probably as "raw" as end users would wish to work with, given
the highly non-linear errors associated with the raw TM-R products. Those
users that do wish to work with TM-R should thoroughly investigate these
distortions and the effect they may have on their results.
Please contact us with your questions.
Daniel R. Steinwand (J
&(^
Charles E. Wivell
Attachments (5)
Copy to: R. Thompson
D. Binnie
J. Boyd
T. Holm
L. Oleson
K. Klenk
6. Johnson
R. Mckinney
R. Sunne
R. Feistner
B. Jones
C. Larson
J. Thonnodsgard
J. Sturdevant
D. Cameggie
6. Kelly
T. Loveland
J. Feuquay
J. Dwyer
J. Eidenshink
D. Gesch
-------
MRLC Consortium
Documentation Manual
January, 1994
-------
MRLC Consortium
Documentation Manual
January, 1994
«o
-------
KKLC Consortium
Documentation
-------
MRLC Consortium
Documentation Manual
January, 1994
I
u>
V)
I
a
Q.
2
•o
1
*e3
O
§
-------
MRLC Consortium
Documentation Manual
January, 1994
-------
MRLC Consortium
Relief 2 Documentation Documentation Manual
January, 1994
Corrects ground control points for relief displacement
ground control points for relief displacement errors due to terrain
for either MSS or TM data. Corrections are made relative to a user-entered
iatum.
Description /Algorithm;
?he tie point location file containing the control points to be adjusted
:or the effects of relief is opened and its header data read. The
jutput tie point location file is then opened, and the header record
:rom the input file is copied to the output file. The center latitude
-S calculated from the projection information. As each point record
-s read, the search image coordinates are adjusted for relief
lisplacement effects in the following manner:
'he NADIR pixel location is calculated by
Nadir Location • intercept + slope * Line Number
nee the Nadir Location is found, the ground-range distance (S)
see figure 3) which is related to the scan angle (d) is calculated
rom:
S - pixel_size * (sample - nadir location)
nd (d) is found using the equations below;
«• S / RE
» « sqrt(RE~2 + (RE + ALT)~2 - 2.0 * RE * (RE + ALT) * cos(s))
d - asin((RE / LOS) * sin(s))
dd - (RE + ALT) * h * d / (RE * (ALT - h))
is then used to calculate z;
2 - asin((RE + ALT)/(RE + H) * sin(d + dd) )
, zf s are used to calculate ds;
ds « z - s + -d + dd
DS - ds * RE
Change in pixel - DS / pixel_size
ich record is written to the output file and to a report.
er Notes:
the RELIEF2 module to correct for errors due to relief
isplacement in control points whenever elevations are
ilable for each point pair and when a pixel-by-pixel
terrain correction will not be performed. Relief will
correct the control points to a reference datum as speci-
fied by the DATUM parameter. By correcting the control
-------
MRLC Consortium
points, relief is not introduced into the geometricDa«jan«Ji*ation Manual
relief in the imagery is not corrected at all. January, 1994
2. The RELIEF2 module is normally used just prior to the con-
trol (tie) point modeling step, usually performed by the
POLYFIT module.
3. RELIEF2 differs from RELIEF by calculating the NADIR pixel
for each line instead of assuming that a straight line with
the center in the middle of the scene describes the NADIR
pixels. Some imagery is tilted so that the NADIR line is
sloped with the image.
diagram is available in the hardcopy user guide.
Figure 1
diagram is available in the hardcopy user guide.
Figure 2
diagram is available in the hardcopy user guide.
Figure 3
-------
MRLC Consortium
Documentation Notebook
April 1994
6.4 Spectral Clustering of Scenes
At the October 1993 MRLC Consortium meeting (see notes in Section 12 of this
notebook), the Consortium agreed that the standard product to be made available to the
participating programs will be a spectrally clustered scene with 240 classes. A performance-
optimized k-means algorithm developed at the Los Alamos National Laboratory was selected for
this purpose. Materials prepared by the staff of the NASA Ames Research Center in Mountain
View, CA, which describe the current implementation of this clustering algorithm in the
KHOROS image processing cantata, are included. An article authored by P.M. Kelly and J.M.
White of the Los Alamos National Laboratory describing the clustering algorithm
("Preprocessing remotely sensed data for efficient analysis and classification", Applications of
Artificial Intelligence 1993: Knowledge-Based Systems in Aerospace and Industry. Proc. SPIE
1993, pp. 24-30) is also included.
-------
MRLC Consortium
Documentation Manual
January, 1994
Preprocessing remotely-sensed data for efficient analysis and classification
Patrick M. Kelly, James M. White
Los Alamos National Laboratory, Computer Research Group
MS B-265, Los Alamos, NM 87545
ABSTRACT
Interpreting remotely-sensed data typically requires expensive, specialized computing machinery capable of stor-
ing and manipulating large amounts of data quickly. In this paper, we present a method for accurately analyzing
and categorizing remotely-sensed data on much smaller, less expensive platforms. Data size is reduced in such a way
as to retain the integrity of the original data, where the format of the resultant data set lends itself well to providing
an efficient, interactive method of data classification.
1. INTRODUCTION
A Landsat Thematic Mapper (TM) quarter scene consists of approximately 12 million pixels, each being repre-
sented by seven spectral reflectance values between 0 and 255. Each quarter scene, therefore, occupies 84 megabytes
of storage, and performing even simple data manipulations for analysis or display purposes requires a large number
of operations. By preprocessing the data by a technique known as vector quantization or clustering, computational
requirements necessary for image analysis and manipulation are greatly reduced.
The advantages to clustering large data sets are numerous. Many times when scientists work with multispectral
irnage data, they are interested in grouping together sets of similar data - something that clustering algorithms do
tomatically. Clustered data also has a number of properties that simplify data analysis and categorization. Data
impression is a very desirable by-product of the clustering process, reducing the computational resources necessary
to manipulate the data. Additionally, because pixels belonging to the same cluster are intrinsically associated with
one another, sets of pixels in an image which share common characteristics can be manipulated simultaneously.
Statistics for each cluster can easily be calculated during the clustering process, allowing many properties of the
original data to be retained. For many applications, we have found that once clustering has been performed, the
original data is no longer needed.
Each pixel in an image is commonly categorized according to its spectral signature. Many methods are used
for classifying multispectral data, including both supervised and unsupervised classification methods [1, 2]. When
using supervised methods for data classification, a user selects training areas representative of several types of
land cover, and a classifier is developed to discriminate between different classes. This classifier is then used to
categorize the remaining pixels in the scene. Numerous pattern recognition algorithms of this type exist, including
nearest neighbor algorithms, discriminant function techniques, artificial neural networks, and statistical methods. An
overview of these techniques can be found in standard pattern recognition textbooks [3, 4]. Statistical methods such
as maximum likelihood classifiers [3] have always been popular for this type of problem. In general, although these
techniques often work well, they are very time consuming both in computer time and operator effort. Additionally,
they do not tend to allow easy classifier adjustments (or "fine-tuning") for the system.
Unlike supervised methods of classification, which require a user to define training sets, unsupervised techniques
require no training sets at all. They instead attempt to automatically find the underlying structure of multi-
dimensional data, by "clustering" the data into groups sharing similar characteristics. Unsupervised classification
is an off-line process, requiring very little time of the system user. A user simply needs to specify a number of
clusters to find, and allow the classification program to do the rest. This technique assumes, however, that the
number of natural categories present in the data is known a priori, with data from different category clusters being
;l-separated.
-------
OLDSTER INDICES
MRLC Consortium
Documentation Manual
January, 1994
CODEBOOK
Figure 1: Clustered Representation of Multispectral Image Data
When using clustering methods for analyzing multispectral data, many people attempt to define a relatively
small number of clusters - between 5 and 100 clusters, for example. Our technique relies on the fact that many
clusters (between 256 and 4096) can be defined for the data. The method of data analysis and classification presented
in this paper first preprocesses the data using a fast clustering algorithm. We cluster the data using a relatively
large number of clusters (as compared to the number of categories we wish to define for the data), and then use
the clustered data for analysis and classification. For many applications, there is no need for the original data after
clustering is performed. Using the clustered data, we can efficiently manipulate computer displays as well as analyze
and categorize data. .
2. CLUSTERING METHODOLOGY
The basic principle of clustering (or vector quantization) is to take an original image (for our example, containing
around 12,000,000 pixels with each pixel being represented by a seven-dimensional vector), and represent the same
image using only a small number of unique pixel values. A codebook of N "best pixel values" to represent the image
must first be generated by some iterative method (the "construction" phase of the clustering algorithm). Once we
have generated these values, we step through the original image and assign each pixel to the cluster of the closest
match existing in our codebook (the "projection" phase of the clustering algorithm). Figure 1 shows the clustered
image representation, as compared to the original image representation.
In processing the data this way, two things have occured. First, we have reduced the volume of data needed to
represent the image by a factor of seven. This is reflected by the fact that we now need only a single band of image
data which contains indices into the codebook of reference vectors. Second, we have done a preliminary classification
of the data; similar pixels in the image are now intrinsically associated with one another.
Since we would like the clustered data to adequately represent the original data, the selection of the codebook
vectors is very important. By increasing the number of clusters, the accuracy of image representation can be
improved. Depending on the application, we use between 256 and 4096 clusters for a typical TM quarter scene. The
time required to cluster the image increases as the number of clusters increases. Aft-- clustering has provided a set
of clusters, the statistics for each cluster are computed and stored in the codebook ^.ong with the cluster reference
vectors. This is an important step because from these statistics, the combined statistics of the original data can
easily be computed.
As an extra step, the cluster indices are sorted according to values stored in the mean vectors. Before this step
is performed, the single two-dimensional band of cluster indices representing the data is meaningless unless it is
associated with its codebook. By sorting the clusters according to values in a single dimension, or by the sum of
multi-dimensional components in each one, a physical meaning is associated with each index. Bright pixels in the
-------
MRLC Consortium
Documentation Manual
January, 1994
iginal data set will be associated with larger cluster indices than the darker pixels. The result will be an image
which, when not associated with its codebook, can easily be displayed as a black and white image of the current
scene.
3. CLUSTERING ALGORITHM
Many types of clustering methods have been developed and analyzed for use with different types of data [3,5]. In
general, many of these algorithms attempt to find a partitioning of a given data set that minimizes a predetermined
cost function. The k-means clustering algorithm [4] attempts to minimize a squared error cost function by manipu-
lating a set of k cluster centers. la particular, this algorithm tries to partition the data into k clusters, denoted by
d, with the representative vector for each cluster (i«-) being defined as the within-cluster mean:
"' .,ec.
This algorithm iteratively moves vectors between clusters in such a way as to minimize the total squared error:
k
(2)
x,€C,
This algorithm, however, becomes painfully slow when using very large data sets. One basic problem is that a
tremendous number of vector distance calculations must be performed during both the "construction" and "projec-
tion" phases of the algorithm. Several methods have been developed to improve this situation [6, 7, 8]. Many of
these schemes work very well in lower-dimensional spaces, but still tend to have a difficult time as the dimension of
the problem and number of clusters increase.
TIMINGS VOX MOBCOV SCCNC
ll
.-l
TIMINGS rOK ALBUqUBKQtJE SCXNZ
3
tt.
Nubvr or Clu>l«r«
B-«-e Contract
T.t.l
n • P C«Hir«at
Total
Figure 2: CPU Timings for Moscow and Albuquerque Scenes
-------
MRLC Consortium
Documentation Manual
January, 1994
We use a. version of the nearest neighbor algorithm proposed in [9], where cluster positions are sorted along one
of the axes for the data. This algorithm, like many others, does not continue to work effectively as the problem
dimension increases. To combat this, we use the first principal component of the data as the axis on which to do the
sort. This axis gives the best possible separation of the data.
Another major hindrance with the k-means algorithm is that the "construction" phase can require many passes
through our tremendous data set to build tLe codebook. But this extra work is not necessary; the data has large
amounts of redundant information. We use a, monte carlo method for passing through the data, and only sample
about 10 percent of the actual data.
Our overall clustering technique yields the same results as the k-means algorithm, but converges much faster.
Clustering times for a TM quarter scene (seven-dimensional data, 3000 rows by 3500 columns) of the Moscow and
Albuquerque areas are shown in Figure 2. These were calculated on a desktop SUN SPARCstation IPX with 16
MB of RAM, and show CPU time required for clustering the data into 256, 512, 1024, 2048, and 4096 clusters. It
is important to note that the execution time grows linearly as the number of clusters is increased. This is not a
property of the algorithm in general, but it has seemed to hold true for the vast majority of real-world multispectral
data sets (as well as most others) that the authors have encountered.
4. DATA ANALYSIS AND CLASSIFICATION
Once our TM scene has been clustered, it requires only one-seventh of the storage originally required, and the
new clustered representation provides an opportunity to use common computer displays very efficiently. Since there
are only N unique "vectors" representing the image, it takes on the order of N operations to manipulate the data as
compared to 12 million operations before the clustering was performed. Calculating the vegetation vigor of pixels
in a TM scene shows an example of the savings incurred by clustering. One measure of vegetation vigor commonly
used by remote sensing specialists is (Band 4 - Band 3) / (Band 4 + Band 3). This transformation results in large
values (bright pixels) for pixels representing healthy vegetation, and requires three operations at each pixel, or 36
million operations for the entire scene. If we first cluster the data to 256 clusters, we can use 8-bit computer displays
effectively. Since the clustered image contains only 256 unique values, 768 operations are required for calculating the
vegetation vigor, and the results can be directly mapped into the computer display look-up-tables (LUTs). While
this is a simple type of operation, the same holds true for very complicated transformations such as the Tasseled Cap
transformation, Karhunen-Loeve transformation, principal component analysis, etc.
Using a display package called SPECTRUM, developed by Los Alamos National Laboratory and the University
of New Mexico, we are able to use any desktop workstation running Unix and Xwindows to analyze and categorize
clustered data. Figure 3 shows a clustered TM scene of Moscow as displayed in SPECTRUM. A user can design and
manipulate a legend that specifies categories of land cover, labels for each category, and pseudocolor representations
to be used when categorizing geographic areas in the clustered image. SPECTRUM can manipulate the color map
for the computer display using any transformation of the clustered data, and can display cluster positions on a
two-dimensional scatter plot. Using these features, users are able to analyze data in a variety of ways. Data can
be categorized by selecting areas with a known type of land cover, causing all associated pixels in the image to
be given the same pseudocolor representation. Using the TM data, for example, a user could locate a wheat field,
highlight the pixels in that field, and all other wheat fields in the entire image would be highlighted immediately.
After categorization, an image can be written out showing the different geographic areas for the scene.
Using the scatter plot, cluster positions can be displayed in a two-dimensional space with axes specified by
the user. Scientists can use this feature to interpret and categorize data by looking at different mathematical
transformations of the cluster positions, while results of the process are updated in the currently displayed clustered
image.
-------
MRLC Consortium
Documentation Manual
January, 1994
Figure 3: Manipulating Moscow Data with SPECTRUM
5. ERROR ANALYSIS
To examine the accuracy of the clustering relative to the number of clusters used, we will look at the average
error per pixel introduced by the clustering, the distribution of these errors, and a Chi Square goodness-of-fit measure
for different land cover training areas.
An 800 x 800 subsection was extracted from the original 3000 X 3500 original image of Moscow and the 3000 X
3500 clustered version of the image. An error image was created by averaging, for the 7 spectral bands, the absolute
difference between the original image and the clustered image data. In the clustered image, each pixel is represented
by the mean vector of the cluster to which it is assigned. It should be noted that errors for each of the individual
bands is similar in magnitude and distribution to the average between the 7 spectral bands. The first plot in Figure
4 shows a plot of the average error per band per pixel and this error ± one standard deviation. The average error
for 256 clusters is less than 2 digital numbers (DN) and drops to less than 1.25 DN average error for 4096 clusters.
The maximum error over the subsection was much larger. There were a few popcorn clouds in the subsection and
the error for the center pixel in the clouds ranged from about 70 DN for the 256 clusters image to about 30 for the
4096 clusters image but these outliers in the data set were few and it is an easy process to isolate them as outliers
during the clustering process. The second plot in Figure 4 shows a histogram of the per pixel errors. The histograms
show that even for the 256 clusters image almost all the pixels have an error within ± 3 DN.
Finally, we chose three training sites for each of 4 land cover types in the 3000 x 3500 Moscow image representing
, soil, water, and forest. The training sites were located in the center of large uniform land covers and chosen
-------
MRLC Consortium
Documentation Manual
January, 1994
Error
Error Hlmtofrmm
3
b
X
I I I I I I I 1 I I
umttmr «f Cl«>t«ra
i » 1 SU I I "I - I SU
•II
IM4
Figure 4: Per Pixel Errors for 800-by-800 Subsection of Moscow Scene
I I I I I I I I I I I I I I I
Crmti
, ror.!t I I I Soil M K X t.t.r
Figure 5: Chi-Squared Goodness of Fit for 7 DOF
-------
MRLC Consortium
Documentation Manual
January, 1994
if they were to be used in a traditional supervised classification. We then did a Chi Square goodness-of-fit test
to determine what our confidence was that the mean vectors representing the clustered data came from the same
process which generated the statistics from the training sets in the original data. The results are shown in Figure
5. A Chi Square test with 7 degrees of freedom has a value of less than 2.83 for greater than 90% confidence and a
value of less than 2.17 for a greater than 95% confidence. For an image with 4096 clusters all land covers had greater
than 95% confidence. For 256 clusters, the goodness-of-fit values were much worse for the water training sets than
for other land covers. The training sets for water were extremely uniform with a variance in each spectral band of
less than 1.5. This means that even small differences between mean vectors yield large Chi Square values.
The errors introduced in a fine grain clustering of the multi-spectral data were not large enough to affect a level
one land use classification. With 4096 clusters, the clustered image could be used to effectively represent the original
data. Each land cover type was identified as easily as with the original image data.
6. CONCLUSIONS
Using a clustering method to do a preliminary classification of multispectral data provides data sets that can
be rapidly categorized in an interactive fashion. A desktop workstation can be used to manipulate and analyze
the preprocessed data in real time. Unlike present uses of clustering, where scientists attempt to find relatively
small numbers of clusters in the data, our techniques define a large number of clusters to use. This data contains a
relatively small number of unique representative vectors that must be categorized, as compared to millions of pixels
in the raw data.
7. ACKNOWLEDGEMENTS
This work was performed under a U.S.,Government contract (W-7405-ENG-36) by the Los Alamos National
.boratory, which is operated by the University of California for the U.S. Department of Energy.
8. REFERENCES
[1] Paul M. Mather. Computer Processing of Remotely-Sensed Images. St. Edmundsbury Press Ltd., Bury St.
Edmunds, Suffolk, 1987.
|"2] Robert A. Schowengerdt. Techniques for Image Processing and Classification in Remote Sensing. Academic
Press. New York. New York, 1983.
[3] R.O. Duda and P.E. Hart. Pattern Classification and Scene Analysis. Wiley, New York, NY, 1973.
[4] J.T. Tou and R.C. Gonzalez. Pattern Recognition Principles. Addison-Wesley, Reading, MA, 1974.
[5] A.K. Jain and R.C. Dubes. Algorithms for Clustering Data. Prentice Hall, Englewood Cliffs, NJ, 1988.
[6] Jerome H. Friedman, Jon Louis Bentley, and Raphael Ari Finkel. An algorithm for finding best matches in
logarithmic expected time. ACM Transactions on Mathematical Software, 3(3):209-226, 1977.
[7] J.L. Bentley. B.W. Weide. and A.C. Yao. Optimal expected-time algorithms for closest point problems. ACM
Transactions on Mathematical Software, 6:563-580, 1980.
[8] M.E. Hodgson. Reducing the computational requirements of the minimum-distance classifier. Remote Sensing of
Environment. 25:117-128, 1988.
[9] Jerome H. Friedman, Forest Baskett, and Leonard J. Shustek. An algorithm for finding nearest neighbors. IEEE
Transactions on Computers, pages 1000-1006, October 1975.
-------
MRLC Consortium
Documentation Notebook
January, 1994
Khoros
"Software integration and development environment that emphasizes
information processing and data visualization."
X-windows image processing environment and system.
Contains programs to manipulate, enhance, and interpret images.
Maintains a programming environment to:
add new functionality
customize existing functions
proceduralize common tasks
store and retrieve records of complex processing sessions
System Size
363 Separate applications programs
Requires 220 Mbytes of disk storage for system
Written by John Rasure and students at University of New Mexico.
Copyright transferred to Khoral Research, Inc. in May 1993.
Open Software Package - Khoros can be used and modified only for
internal use in the organization obtaining it. The organization cannot
redistribute khoros unless the organization is a member of the Khoros
Consortium and has signed a redistribution license agreement.
Khoros Consortium - group of agencies and companies who fund
khoros development and maintenance. USGS has been a member.
Available through anonymous ftp over the Internet from site
ftp.eece.unm.edu (129.24.24.119).
Los Alamos programs available from this site (as the C3 Cluster
toolbox) or from Jim White at LANL Qwhite@lanl.gov)
Spectrum program available from this site (as the Classify toolbox)
-------
MRLC Consortium
Documentation Notebook
January, 1994
khoros 2.0
Major upgrade of khoros image processing environment
In alpha release now (to members of khoros Consortium)
Beta release expected in mic December
Public release scheduled for second quarter 1994
Active development on:
HP 9000/700 HP-UX 8.07
SGI Indigo OS 4.0.4
DEC Alpha OSF1.2
SUN SparcStations SUN OS 4.1.3 (Solaris 1.1)
SUN SparcStations SUN OS 5.1,5.3
(Solaris 2.1,2.3)
New Features
•Able to handle large images efficiently
•Removing reliance on Athena widget set - choose widget set at
compile time
•Image format more object oriented. Will recognize and deal with
non-Viff image formats
•Display programs will handle 16-bit images
•User can customize environment - select order within menus,
special help files, etc.
•Georeferencing information will be provided in the viff header
•Able to display irregular areas of interest (areas, points, polygons)
via the annotation layer. Eventually hope to have GIS file formats
directly supported by khoros (currently unfunded).
LANL will port construct, modify codebook, and project to khoros 2.0
UNM group estimates they could port khoros 2.0 to DG for about $20k
-------
MRLC Consortium
Documentation Notebook
January, 1994
Executing Khoros programs
Khoros programs can be run in several different ways
• cantata - visual programming environment
-programs are selected through pull down menus
-placed on workspace as "glyphs"
-linked to transfer output from one glyph to the next
-executed singly or as a unit
-workspace can be stored and retrieved
Requires Xwindows execution
• Batch mode
-command line specification of all program options
-programs may be executed sequentially, but output cannot
~ be "piped" between programs
• Command-line prompts
-user is prompted for program options, including defaulted
items
• Xv routines
-program name is entered by user
-program runs with pull down menus and options
-requires Xwindows execution
-can also be run through cantata
-------
MRLC Consortium
Documentation Notebook
January, 1994
Hver-Clusterin
•
Execution Time
Projection is the longest step to execute
Affected by system load, amount of uniform area (background
in input, data volume (# pixels, # bands)
•Run over the same geographic area with different band
combinations:
Construct Project Total Seconds/
# Pixels # Bands Seconds Seconds Seconds Mbyte
8,073,000 6 1013 2688. 3701 80.126
8,073,000 5 833 2160 2993 77.750
8,073,000 6 1281 3623 4904 106.170
•Limitations
-Memory
Sufficient memory (or swapspace) to store full multispectral
image
Ames limitation of 64 Mbytes
-Disk
Temporary files may eat up available free space
Input data needs to be in viff format and interleaved
Procedure is: transform each input band into viff, then
combine separate bands into one multispectral dataset
Requires three separate stores of images to disk.
-------
MRLC Consortium
Documentation Notebook
January, 1994
-------
MRLC Consortium
Documentation Notebook
January, 1994
si1
O'
CD
g
•**
a-
ci
I
O
•*+
*>
8
CO
v>
g
1
CO
(I)
CO
g
5
Q_
Q)
O
-------
MRLC Consortium
Documentation Notebook
January, 1994
fe
O)
I
I
I
§
8,
s f
O. -P
LU
Choose Selection
S
Regi
er Reduc
-------
LA-UR-93-301
Los Alamos National Laboratory is operated by the University of California for the United States Department of Energy under contract W-740S-ENG-36
TITLE: PREPROCESSING REMOTELY-SENSED DATA FOR
EFFICIENT ANALYSIS AND CLASSIFICATION
ALTTHOR(S):
Patrick M. Kelly, and James M. White
SUBMITTED TO:
International Society of Optical Engineering Conference
Orlando, Florida
April 12-16,1993
By acceptance of this article, the publisher recognizes that the U.S. Government retains a nonexclusive royalty-free license to publish or reproduce
the published form of this contribution or to allow others to do so, tor U.S. Government purposes.
Th« Ltw Atamoi National Lsbormtxy requests that trw puWsher terrify this «rtd«M work performed und»ir»«o«p«»« of m«U5.D«»rtrn«motEn«r5y.
Los Alamos National Laboratory
Los Alamos New Mexico 87545
-------
Preprocessing remotely-sensed data for efficient analysis and classification
Patrick M. Kelly, James M. White
Los Alamos National Laboratory, Computer Research Group
MS B-265, Los Alamos, NM 87545
ABSTRACT
Interpreting remotely-sensed data typically requires expensive, specialized computing machinery capable of stor-
ing and manipulating large amounts of data quickly. In this paper, we present a method for accurately analyzing
and categorizing remotely-sensed data on much smaller, less expensive platforms. Data size is reduced in such a way
as to retain the integrity of the original data, where the format of the resultant data set lends itself well to providing
an efficient, interactive method of data classification.
1. INTRODUCTION
A Landsat Thematic Mapper (TM) quarter scene consists of approximately 12 million pixels, each being repre-
sented by seven spectral reflectance values between 0 and 255. Each quarter scene*, therefore, occupies 84 megabytes
of storage, and performing even simple data manipulations for analysis or display purposes requires a large number
of operations. By preprocessing the data by a technique known as vector quantization or clustering, computational
requirements necessary for image analysis and manipulation are greatly reduced.
The advantages to clustering large data sets are numerous. Many times when scientists work with multispectral
image data, they are interested in grouping together sets of similar data - something that clustering algorithms do
automatically. Clustered data also has a number of properties that simplify data analysis and categorization. Data
compression is a very desirable by-product of the clustering process, reducing the computational resources necessary
to manipulate the data. Additionally, because pixels belonging to the same cluster are intrinsically associated with
one another, sets of pixels in an image which share common characteristics can be manipulated simultaneously.
Statistics for each cluster can easily be calculated during the clustering process, allowing many properties of the
original data to be retained. For many applications, we have found that once clustering has been performed, the
original data is no longer needed.
Each pixel in an image is commonly categorized according to its spectral signature. Many methods are used
for classifying multispectral data, including both supervised and unsupervised classification methods [1, 2]. When
using supervised methods for data classification, a user selects training areas representative of several types of
land cover, and a classifier is developed to discriminate between different classes. This classifier is then used to
categorize the remaining pixels in the scene. Numerous pattern recognition algorithms of this type exist, including
nearest neighbor algorithms, discriminant function techniques, artificial neural networks, and statistical methods. An
overview of these techniques can be found in standard pattern recognition textbooks [3, 4]. Statistical methods such
as maximum likelihood classifiers [3] have always been popular for this type of problem. In general, although these
techniques often work well, they are very time consuming both in computer time and operator effort. Additionally,
they do not tend to allow easy classifier adjustments (or "fine-tuning") for the system.
Unlike supervised methods of classification, which require a user to define training sets, unsupervised techniques
require no training sets at all. They instead attempt to automatically find the underlying structure of multi-
dimensional data, by "clustering" the data into groups sharing similar characteristics. Unsupervised classification
is an off-line process, requiring very little time of the system user. A user simply needs to specify a number of
clusters to find, and allow the classification program to do the rest. This technique assumes, however, that the
number of natural categories present in the data is known a priori, with data from different category clusters being
well-separated.
-------
CLUSTER INDICES COOKBOOK
Figure 1: Clustered Representation of Multispectral Image Data
UJ1
^^du
w*
When using clustering methods for analyzing multispectral data, many people attempt to define a relatively
small number of clusters - between 5 and 100 clusters, for'example. Our techaiqop relies on the fact that many
clusters (between 256 and 4096) can be defined for the data. The method of data analysis and classification presented
in this paper first preprocesses the data using a fast clustering algorithm. We cluster the data using a relatively
large number of clusters (as compared to the number of categories we wish to define for the data), and then use
the clustered data for analysis and classification. For many applications, there is no need for the original data after
iustering is performed. Using the clustered data, we can efficiently manipulate computer displays as well as analyze
d categorize data.
2. CLUSTERING METHODOLOGY
The basic principle of clustering (or vector quantization) is to take an original image (for our example, containing
around 12,000,000 pixels with each pixel being represented by a seven-dimensional vector), and represent the same
image using only a small number of unique pixel values. A codebook of N "best pixel values" to represent the image
must first be generated by some iterative method (the "construction" phase of the clustering algorithm). Once we
have generated these values, we step through the original image and assign each pixel to the cluster of the closest
match existing in our codebook (the "projection" phase of the clustering algorithm). Figure 1 shows the clustered
image representation, as compared to the original image representation.
In processing the data this way, two things have occured. First, we have reduced the volume of data needed to
represent the image by a factor of seven. This is reflected by the fact that we now need only a single band of image
data which contains indices into the codebook of reference vectors. Second, we have done a preliminary classification
of the data; similar pixels in the image are now intrinsically associated with one another.
Since we would like the clustered data to adequately represent the original data, the selection of the codebook
vectors is very important. By increasing the number of clusters, the accuracy of image representation can be
improved. Depending on the application, we use between 256 and 4096 clusters for a typical TM quarter scene. The
time required to cluster the image increases as the number of clusters increases. After clustering has provided a set
of clusters, the statistics for each cluster are computed and stored in the codebook along with the cluster reference
vectors. This is an important step because from these statistics, the combined statistics of the original data can
easily be computed.
As an extra step, the cluster indices are sorted according to values stored in the mean vectors. Before this step
1 performed, the single two-dimensional band of cluster indices representing the data is meaningless unless it is
associated with its codebook. By sorting the clusters according to values in a single dimension, or by the sum of
multi-dimensional components in each one, a physical meaning is associated with each index. Bright pixels in the
-------
original data set will be associated with larger cluster indices than the darker pixels. The result will be an image
which, when not associated with its codebook, can easily be displayed as a black and white image of the current
scene.
3. CLUSTERING ALGORITHM
Many types of clustering methods have been developed and analyzed for use with different types of data [3,5]. In
general, many of these algorithms attempt to find a partitioning of a given data set that minimizes a predetermined
cost function. The k-means clustering algorithm [4] attempts to minimize a squared error cost function by manipu-
lating a set of k cluster centers. In particular, this algorithm tries to partition the data into k clusters, denoted by
Cj, with the representative vector for each cluster (x,-) being defined as the within-cluster mean:
*J (1)
1 *>€C.
This algorithm iteratively moves vectors between clusters in such a way as to minimize the total squared error:
Error =
x,--i,-
(2)
tel
This algorithm, however, becomes painfully slow when using very large data sets. One basic problem is that a
tremendous number of vector distance calculations must be performed during both the "construction" and "projec-
tion" phases of the algorithm. Several methods have been developed to improve this situation [6, 7, 8]. Many of
these schemes work very well in lower-dimensional spaces, but still tend to have a difficult time as the dimension of
the problem and number of clusters increase.
TIMINGS TOU MOSCOW SCENE
TIMIKCB rOK ALBUQUERQUE SCENE
4)
s
"I
x-
I-
a
N«»»*r or Cluvtar*
of Clu*t*r>
B-B-C Cmiinet *-*-* Fr*J*ot I I I ToUl
Cmlnoi
Total
Figure 2: CPU Timings for Moscow and Albuquerque Scenes
-------
We use a version of the nearest neighbor algorithm proposed in [9], where cluster positions are sorted along one
of the axes for the data. This algorithm, like many others, does not continue to work effectively as the problem
dimension increases. To combat this, we use the first principal component of the data as the axis on which to do the
sort. This axis gives the best possible separation of the data.
Another major hindrance with the k-means algorithm is that the "construction" phase can require many passes
through our tremendous data set to build the codebook. But this extra work is not necessary; the data has large
amounts of redundant information. We use a monte carlo method for passing through the data, and only sample
about 10 percent of the actual data.
Our overall clustering technique yields the same results as the k-means algorithm, but converges much faster.
Clustering times for a TM quarter scene (seven-dimensional data, 3000 rows by 3500 columns) of the Moscow and
Albuquerque areas are shown in Figure 2. These were calculated on a desktop SUN SPARCstation IPX with 16
MB of RAM, and show CPU time required for clustering the data into 256, 512, 1024, 2048, and 4096 clusters. It
is important to note that the execution tune grows linearly as the number of clusters is increased. This is not a
property of the algorithm in general, but it has seemed to hold true for the vast majority of real-world multispectral
data sets (as well as most others) that the authors have encountered.
4. DATA ANALYSIS AND CLASSIFICATION
'"•"Once our TM scene has been clustered, it requires only one-seventh of the storage originally required, and the
new clustered representation provides an opportunity to use common computer displays very efficiently. Since there
are only N unique "vectors" representing the image, it takes on the order of N operations to manipulate the data as
compared to 12 million operations before the clustering was performed. Calculating the vegetation vigor of pixels
TM scene shows an example of the savings incurred by clustering. One measure of vegetation vigor commonly
d by remote sensing specialists is (Band 4 - Band 3) / (Band 4 -f Band 3). This transformation results in large
values (bright pixels) for pixels representing healthy vegetation, and requires three operations at each pixel, or 36
million operations for the entire scene. If we first cluster the data to 256 clusters, we can use 8-bit computer displays
effectively. Since the clustered image contains only 256 unique values, 768 operations are required for calculating the
vegetation vigor, and the results can be directly mapped into the computer display look-up-tables (LUTs). While
this is a simple type of operation, the same holds true for very complicated transformations such as the Tasseled Cap
transformation, Karhunen-Loeve transformation, principal component analysis, etc.
Using a display package called SPECTRUM, developed by Los Alamos National Laboratory and the University
of New Mexico, we are able to use any desktop workstation running Unix and Xwindows to analyze and categorize
clustered data. Figure 3 shows a clustered TM scene of Moscow as displayed in SPECTRUM. A user can design and
manipulate a legend that specifies categories of land cover, labels for each category, and pseudocolor representations
to be used when categorizing geographic areas hi the clustered image. SPECTRUM can manipulate the color map
for the computer display using any transformation of the clustered data, and can display cluster positions on a
two-dimensional scatter plot. Using these features, users are able to analyze data in a variety of ways. Data can
be categorized by selecting areas with a known type of land cover, causing all associated pixels in the image to
be given the same pseudocolor representation. Using the TM data, for example, a user could locate a wheat field,
highlight the pixels in that field, and all other wheat fields in the entire image would be highlighted immediately.
After categorization, an image can be written out showing the different geographic areas for the scene.
Using the scatter plot, cluster positions can be displayed in a two-dimensional space with axes specified by
the user. Scientists can use this feature to interpret and categorize data by looking at different mathematical
transformations of the cluster positions, while results of the process are updated in the currently displayed clustered
image.
-------
Figure 3: Manipulating Moscow Data with SPECTRUM
5. ERROR ANALYSIS
To examine the accuracy of the clustering relative to the number of clusters used, we will look at the average
error per pixel introduced b> the clustering, the distribution of these errors, and a Chi Square goodness-of-fit measure
for different land cover training areas.
An 800 x 800 subsection was extracted from the original 3000 X 3500 original image of Moscow and the 3000 X
3500 clustered version of the image. An error image was created by averaging, for the 7 spectral bands, the absolute
difference between the original image and the clustered image data. In the clustered image, each pixel is represented
by the mean vector of the cluster to which it is assigned. It should be noted that errors for each of the individual
bands is similar hi magnitude and distribution to the average between the 7 spectral bands. The first plot in Figure
4 shows a plot of the average error per band per pixel and this error ± one standard deviation. The average error
for 256 clusters is less than 2 digital numbers (DN) and drops to less than 1.25 DN average error for 4096 clusters.
The maximum error over the subsection was much larger. There were a few popcorn clouds hi the subsection and
the error for the center pixel in the clouds ranged from about 70 DN for the 256 clusters image to about 30 for the
4096 clusters image but these outliers in the data set were few and it is an easy process to isolate them as outliers
during the clustering process. The second plot in Figure 4 shows a histogram of the per pixel errors. The histograms
show that even for the 256 clusters image almost all the pixels have an error within ± 3 DN.
Finally, we chose three training sites for each of 4 land cover types in the 3000 x 3500 Moscow image representing
grass, soil, water, and forest. The training sites were located in the center of large uniform land covers and chosen
-------
vorafo Error
Error Bl«to«r«»
1 BU | | | - I BU
•It
Figure 4: Per Pixel Errors for 800-by-800 Subsection of Moscow Scene
Oood»o» of Pit •tatlctloi
I I I I I I I I I I I I I I I
Gwmit
L rvrait I I I loll M-M-M t«t«r
Figure 5: Chi-Squared Goodness of Fit for 7 DOF
-------
MRLC CooMrtium
Documentation Notebook
April 1994
SECTION?
SCENE LABELLING AND CLASSIFICATION
The MRLC Consortium is exploring ways in which the generation of landcover surfaces
by the participating programs can be accomplished in a coordinated fashion. The MRLC is
considering the use of a common software approach to the labelling of spectrally clustered data,
and pilot programs to explore the effectiveness and compatibility of joint labelling and
classification exercises. Information on the implementation of coordinated regional efforts is
included in Section 14 of this notebook. At the November MRLC Consortium meeting in
Mountain View, the participating programs decided to implement the SPECTRUM open software
system for labelling landcover classes. This section also includes landcover classification
requirements and strategies for each of the participating programs.
-------
-------
MRLC Consortium
Documentation Notebook
August 1994
7.1 SPECTRUM Software
The MRLC Consortium is considering using the SPECTRUM cluster labelling software
package as part of its joint efforts to provide landcover mapping across the United States. This
section contains information concerning this package. Section 7.1.1 contains information
prepared and provided by staff of the NASA Ames Research Center (Moffett Field, CA).
Section 7.2.1 contains a summary report discussing the software, generated following a
workshop on the use of SPECTRUM software for cluster labelling
-------
MRLC Consortium
Documentation Notebook
August 1994
7.1.1 NASA Ames notes on SPECTRUM
The attached handouts were prepared by staff of the NASA Ames Research Center
describing SPECTRUM in its currently available form. A users guide for the SPECTRUM
software is being developed by NASA Ames staff and will be included or referenced here in
future notebook updates.
-------
MRLC Consortium
Documentation Notebook
August 1994
7.1.2 Reston, VA SPECTRUM Pilot Workshop Summary
On June 28 to July 1, the MRLC Consortium conducted a workshop to explore and
evaluate the use of the SPECTRUM software system in a joint classification project. The
document contained within this section was written by the workshop participants, and
summarizes their findings on the functionality of the SPECTRUM software system, in its current
version.
-------
MRLC Consortium
Documentation Notebook
April, 1994
SPECTRUM - Satellite Image Interpretation With Automated Delineation:
A Workshop-Based Assessment of SPECTRUM Software
Prepared June 1994
by workshop participants
Wayne Myers, Perm State University
Gail Thelin, USGS-WRD NAWQA
Susan Benjamin, NMD NASA-AMES Research Center
Ann Raspberry, Maryland, DNR
Joy Hood, EROS Data Center
Paul Etzler, EMSL, Las Vegas, NV
Jim Majure, Iowa State University
John Brakebill, USGS-WRD Potomac NAWQA
Pat Green, EPA-EMAP Forest, RTF, NC
John Findley, USGS-NMD, Reston, VA
Abstract
A workshop was conducted June 28-30, 1994 at the USGS National Center in Reston, VA by
representatives of the MRLC (Multi-Resolution Land Characteristics) consortium for the purpose
of learning and evaluating SPECTRUM image analysis software relative to joint goals of
consortium programs. The software is reasonably user-friendly, and permits satellite image data
(notably Thematic Mapper) to be approached in an interpretive mode for land-use/land-cover
mapping without the necessity of painstaking feature delineations. Suggestions were developed
for mapping strategy, a few inconveniences were noted, and recommendations made for possible
future enhancements.
Introduction
SPECTRUM implements an unsupervised classification approach to multi-spectral image data.
Unsupervised classification involves first "clustering" the image data to capture the major image
information, and then assigning clusters to categories of interest for mapping. The SPECTRUM
version of the unsupervised approach was developed by Patrick M. Kelly and James M. White
-------
MRLC Consortium
Documentation Notebook
August 1994
in the Los Alamos National Laboratory, Computer Research Group. The original context of
development was defense intelligence. The clustering mechanism uses a nearest-neighbor
algorithm giving results similar to the k-means program in the SAS statistical package, but
utilizes several innovative strategies to improve speed and accommodate large data sets. A
simple user's perspective for MRLC is that SPECTRUM provides a computer-assisted mode of
"photointerpreting" satellite image data that is rapid, highly interactive, and does not require
extensive prior experience in remote sensing. As is typical of more conventional
photointerpretation, however, the quality of the final map improves with the analyst's knowledge
of the landscape being mapped and with amount of ancillary information available.
A particular advantage of the system relative to clustering is that many more clusters are
generated than typical for other versions of unsupervised analysis, thus capturing more of the
scene information. This multiplicity of clusters is called "hyper-clustering," and enables
reasonable reproduction of the scene from just the cluster information alone. Therefore,
hyper-clustering also constitutes a method of image data compression. Another substantial
advantage for MRLC users is that EROS Data Center will precluster the scene and provide this
information in the manner of an additional image band. Thus MRLC users need not be bothered
with the clustering phase at all, and can get right to the business of assigning clusters to desired
map categories with the SPECTRUM software.
Mapping Scenario
One begins by loading the cluster image and associated cluster information into memory of a
UNIX workstation computer. The next order of business is to select three "image bands" for
display on the screen. In fact, the resulting display is an approximation of the original image
as rendered through the spectral band means for the several clusters. Analysts with
photointerpretation experience will probably choose either a band combination that gives a
"color-infrared" view or a "conventional color" view. Each has advantages for interpreting
particular types of landscape features. Various "indexes" such as greenness, brightness,
wetness, and so on can also be displayed if the analyst is familiar their formulation as ratios or
linear combinations of spectral bands.
The desired map legend is next entered as a set of category labels for landscape features of
interest (e.g., land-cover classes). Along with specifying a category label, one chooses a color
to appear on the screen for "pixels" which will be placed in that category. The actual process
of assigning clusters to map categories then begins. A "zoom" window is opened, and a
representative sector of the image is moved into the zoom window with the mouse-driven cursor.
As the cursor is moved around in the zoom window, the number of the cluster in that pixel
location is displayed. One chooses a pixel location for which the map category is known from
ancillary information, "ground truth," or general "lay of the land" as seen in the image display.
Double clicking the location brings up a window for assigning the particular cluster number to
a map category. All other pixels belonging to the same cluster then appear in the designated
category color throughout the rest of the image. Clusters can be transferred from one map
-------
MRLC Consortium
Documentation Notebook
August 1994
category to another if desired. For those with digital image analysis experience, this latter
process is very much like "training set" selection in supervised analysis.
If one is interested only in a very general categorization (perhaps water, forest, agriculture, and
other), the assignment can probably be accomplished without recourse to ancillary information
according to the appearance of the landscape in the image. If one is interested in a more
detailed categorization (perhaps vegetation community types), it becomes necessary to adopt the
traditional photointerpreter's approach to convergence of evidence using ancillary information
(topographic maps, soils maps, airphotos, etc.). This involves a special "highlight" category in
which each cluster is temporarily placed by itself so that the distribution of its member pixels
over the landscape can be viewed readily. The cluster can then be examined in terms of
elevation, aspect, soils, and so on in order to determine its characteristics relative to criteria for
map categories. Although more time-consuming, it may be appropriate to run a text editor as
a separate process in a window so that the characterization for each cluster can be documented
in the course of interpretation. A bit of counsel based on photointerpretation experience is that
careful assignment is generally more than repaid by avoidance of frustration in correcting errors
later.
We would advise that you carry a typical quarter-scene (TM) through the entire process,
including verification, before proceeding with the rest of your imagery. This will alert you to
the likely pitfalls for the remainder of work, give you a good sense of expected accuracy, and
perhaps reveal some category confusion that simply cannot be resolved in this particular mode
of mapping. In the latter case, you should plan on refining your draft map by subsequent
exploitation of other sources of information.
Multi-temporal Mapping
Phenology is very important in separating land-use/land-cover and vegetation classes on the basis
of spectral information. The scene with which we experimented in the workshop was clustered
as a composite of two images, one from early summer (June) and the other from fall (late in
October). This is a particularly advantageous combination relative to phenology, and the
composite clustering is much better than having the same two scenes clustered separately.
The composite gives rise to a large number of clusters, several of which are likely to represent
the same map category. It is much easier, however, to assign several clusters to the same map
category than to face the prospect of lack of separability between categories. A given forest type
may be in different stages of fall color change as a result of elevation differences, giving several
clusters for the same category. However, such changes also permit detecting conifers in mixture
with hardwoods and induce crop differences associated with senescence or harvest. More
ancillary information may be needed to account for phenological distinctions between clusters,
but the distinctions at least become possible. Dual dates also allow working under clouds, as
long as the clouds do not coincide in both images.
-------
MRLC Consortium
Documentation Notebook
August 1994
Working with a multi-date composite will require the interpreter to alternate views of the image.
It will be necessary to switch back and forth between early-season infrared and late-season
infrared, perhaps along with conventional color for one or both dates. Multiple dates also
increase the importance of learning expected spectral signatures, which are levels of differing
reflectance between bands and dates for particular types of features. SPECTRUM makes
available a signature profile (plot of band means) when an instance of a cluster number is
pending category assignment.
Multi-date composites will complicate the prospect of preclustering at the EROS data center.
EROS may find it logistically impractical to precluster in different combinations of years and
dates. This will serve as motivation for user sites to undertake their own clustering.
Provision for Refinement
It would be unrealistic to expect that the foregoing SPECTRUM scenarios will adequately
address all map categories for all thematic contexts. Thus it is only prudent to anticipate possible
need for further refinement after you have done your best in SPECTRUM. SPECTRUM itself
does not currently embody substantial capabilities for on-screen map editing outside the cluster
environment. There are several paths by which the results of SPECTRUM work can be carried
into other software systems that are better geared to editing operations. Unfortunately, the
transport utilities are also not currently part of SPECTRUM per se. You are referred to remote
sensing personnel at EROS Data Center for determining the most expedient import/export
capability relative to your favorite GIS.
Making SPECTRUM More Commodious for Interpreters
SPECTRUM developers have apparently done little in the way of multi- temporal interpretation
themselves, else they would have made it unnecessary to keep repeating some of the interpretive
operations. The most obvious instance involves switching of image views. It is presently
necessary to associate a spectral band with each color plane of the computer display each time
you want a different view. When you have once set up a view in this manner, it should be
possible to "save" the view under some name so that it can be reselected easily when it is needed
again. We strongly urge that such a capability be added to SPECTRUM in its next version.
Equally annoying is the need to specify a numeric level of color for each plane in assigning a
color to a category. Susan Benjamin currently has a sheet of paper that associates color levels
with color names. We wholeheartedly encourage the incorporation of name-based color selection
as an option in SPECTRUM. However, the capability to specify colors by numeric level should
also be retained.
We also view as practical necessity the ability to "quick save" and retrieve the status of category
assignments along with cluster means by cluster and band number to/from an ASCII file. This
-------
MRLC Consortium
Documentation Notebook
August 1994
would not only allow interruption/resumption of work sessions and going-back to prior stages,
but would also allow local programming of bridgework to statistical packages.
Procurement and Platforms
SPECTRUM was developed to run in the Khoros software environment on UNIX workstation
computers. It is possible to obtain Khoros with SPECTRUM by anonymous FTP through the
Internet. If interest lies solely in SPECTRUM, however, one should seek a stand-alone version
from EROS Data Center.
It must also be noted that all UNIX workstations are not created equal relative to SPECTRUM.
SPECTRUM saw its first intensive use on Data General platforms at the workshop. While
individually and collectively instructive, the workshop was not thematically productive due to
frequent lock-up of the DCs during SPECTRUM sessions. Such problems have not occurred
on Sun workstations. Version 2.0 of SPECTRUM is due for release in September, and will
have been tested on DGs.
Wish List for Sophisticated Analysts
We would like to:
a) Have current cluster enter scatter plot last so that color/position is not obscured by
plotting of other clusters;
b) Have optional scatter plots on principal component axes;
c) Examine the spectral heterogeneity of individual clusters (standard deviations to go with
means);
d) Retain the seed for a cluster and examine its relation to the ultimate cluster mean;
e) Examine the spectral heterogeneity of clusters assigned to a thematic category;
f) Explore the prospective addition of clusters to a thematic class on the basis of spectral
similarity;
g) Create supercategories of categories for spectral comparison;
h) Explore the intercluster spectral structure though higher-dimensional displays and/or
collapsing dendrogram;
i) Create spatial partitions of a spectral cluster for separate labeling by polygonal enclosure
-------
MRLC Consortium
Documentation Notebook
August 1994 4
with cursor;
j) Have capability for explicit seeding of clusters, including cluster means from other scenes
that may not actually exist as a pixel in present scene;
k) Restrict Monte Carlo sampling with an exclusionary binary mask, ie. cluster for multiple
strata;
1) Display multiple spectral reflectance curves, ie. display curves for deciduous forest types
to compare 'characteristic' spectral signatures;
m) Save a library of spectral reflectance curves;
n) build a menu of 'standard' indices or formulas, ie. greeness, wetness, brightness, etc.
so the user doesn't have to type them in.
-------
„
in
DC
Q.
LLI
f—
2 0 DC
cn Q o
LU _| LJU
D_
CO
Q. CO
CO
I-
LJJ
MKLC Consortium
Documentation Notebook
January, 1994
-------
MRLC Consortium
Documentation Notebook
January, 1994
SPECTRUM
An interactive program to visually interpret land use / land cover from
classified multispectral images
Input: "Clustered" file written by project
Output: "Legend" file describing land cover units
"Image" file with header information to assign clusters to
the legend units
"Colormap" file of RGB values for land cover units
Interpretation is a visual process
Image is displayed
interpreter outlines polygons of contiguous land cover
Clusters within that polygon can be:
-assigned to a new or existing unit
-ignored
-transferred from a current assigned unit to a new or different
one
Use of codebook statistics on cluster mean values (stored in the
image header) lets the program treat the classified image as through
it was still a multispectral image.
Can display different band combinations, functions of bands, or
transforms of bands.
Hardware requirements are simple:
Unix and Xwindows-compatible platform
8-bit color display
mouse and keyboard
-------
MRLC consortium
Documentation Notebook
January, 1994
Spectrum Processing Files
Raw Image Single band per file. No header information.
VIFF-format Khoros Image format. Band interleaved with header
Codebook Binary file of per-band cluster means. Created by
the program "Construct" and modified by the
program "Modify Codebook" to include class 0.
Cluster Image Image file created by the program "Project"
Single-band, each pixel has a value from 0 to
maximum number of clusters. Header contains a
copy of the codebook file, modified to reflect the
pixels assigned to each class.
Input to Spectrum for land cover interpretation.
Clusout Image Image file created by Spectrum. Header contains
a "count column" indicating # of pixels in
each class.
If the image has been interpreted, the header
contains a "class column" indicating which classes
are assigned to each land cover unit.
Legend File File of land cover unit names and colors, created
by Spectrum.
Colormap File File of Red-Green-Blue color values used to display
each land cover unit. Created by Spectrum. Ascii text.
Image with Viff-format image file created by Spectrum. Header
Colormap contains a color map with the color assignments made
during interpretation. Pixel values range from 0 to
# of classes.
-------
MRLC Consortium
Documentation Notebook
January, 1994
Spectrum Output
To transfer interpretation back to khoros (or to another system) the
interpreted image is written out with its colormap as a standard viff format
file with colormap stored in the header
This file can be converted to a "raw" format fiie (no header) for transfer to
another system.
The color map can be written to an ascii fiie for transfer to another system.
Within khoros, the colormap from one viff image can be applied to another.
This transfers interpretation of one section of an image to another section.
-------
MRLC Consortium
Documentation Notebook
January/ 1994
-------
MRLC Consortium
Documentation Notebook
January, 1994
Input/Output
Khoros Image Files (VIFF)
Input File :
Output File:
HOP | {Close
Inputs the "clustered" rnage from project
Legend Tiles descrbe Land Cover Unts,
Legend files assigned to each, and colors assigned t<
Input File :
Output File:
Colormap Output Saves the interpreted Land Cover Map
Output Image & Colormap: <^
Output Colormap Only:
lasses
jach.
The Spectrum Input/Output Window
-------
MKLC Consortium
Documentation Notebook
January, 1994
Control Display of Image
Type Of Normalize
When To Normalize
How To Normalize
Local)
Contrast Stretch Control
Uhen Necessary
C 0 < norm < HaxColors
Change Hap Columns Currently Displayed as Red, Green, Blue:
HO
Ml
, Controls for rnage band combinati
I "2 function display.
Can be changed at any stage of
Define Red, Green & Blue as Functions of map columns:
relation.
BLUE
File to View [^ Shows a text file (wth function parameters^
-------
MftLC Consortium
Documentation Notebook
January, 1994
-------
MRLC Consortium
Documentation Notebook
January, 1994
Legend
Colorspace models
RGB
Control for class and Land C<
[]HSV [] HL?Ol° GREY
I Clear Polygons from Image | j flbort Polygon Creation |
| Delete Categories) |
I Empty Categonj(s) |
I Catch-fill Category I
j Shoui Selected Category(s) j
I Hide Selected Categoru.
-------
MKLC Consortium
Documentation Notebook
January, 1994
Examine
clusters in
spectral
space.
Add or
delete
clusters from
land cover
unis
92*219406 x 99.235680
-------
MRLC Consortium
Documentation Notebook
January, 1994
Tools for Interpretation
•Standard Legends
•Function and Transform Files
•image Stratification Using ARC/INFO
•Image Stratification Using khoros Thresholding
•Use of Ancillary Data with Classifications
-------
MRLC Consortium
Documentation Notebook
January/ 1994
Standard Legend Files
Have written "empty" legend files for established/proposed
classification systems:
LUDA
new USGS
UNESCO/GAP
C-CAP
NALC
Legend File = list of land cover units and standard colors to be used
with them.
Using Standard Legend File:
Input File = Spectrum-created image file with "count" column
Input Legend = standardized legend file
Legend initially appears colorless
As clusters are assigned to each unit, standard colors appear
Function/Transform Files
Spectrum "Display Form" allows scrolling display of ascii file while
interpreting land cover classes
Function and transform equations can be input to
Band Display (red, green, or blue color guns)
Scatterplot
TM and MSS-specific equation files have been written
Equations are in "map column" form
-------
MKLC Consortium
Documentation Notebook
January, 1994
Standard Legends
Anderson Level 1 Legend NALC Level 1 Legend
250 0 0 'Urban or built-up land1 1 220 0 220 '1.0 Developed Land1 1
200 150 0 'Agricultural land' 2 200 150 0 '2.0 Cultivated Land1 2
255 200 0 'Rangeland1 3 250 200 0 '3.0 Grassland
(herbaceous)' 3
0 200 88 'Forest land1 4 0 200 0 '4.0 Woody1 4
0 0 250 'Water1 5 200 200 200 '5.0 Exposed Land1 6
0 150 200 'Wetland' 6 255 255 255 '6.0 Snow and Ice' 5
200 200 200 'Barren land' 7 0 200 255 7.0 Wetland1 7
200 225 200 Tundra' 8 00 200 '8.0 Water and submerged
land1 8
255 255 255 'Perennial snow or ice1 9
USGS_new Level 1 Legend
220 0 220 'Developed Land' 1
200 150 0 'Cultivated Land' 2
250 200 0 'Grassland' 3
0 200 0 'Woody Land' 4
0 0 200 'Water1 5
0 200 255 'Wetland' 6
200 200 200 'Exposed Land' 7
200 225 200 Tundra' 8
255 255 255 'Snow and Ice' 9
-------
MRLC Consortium
Documentation Notebook
January, 1994
Function / Transform Files
Thematic Mapper Equations for Spectrum Display
NDVI : ((M3-M2) / (M3+M2))
TVI : (SORT (((M3-M2) / (M3+M2)) +0.5) )
WATER-BODIES: ((M4 - Ml) / (M4 + Ml))
KAUTH-THOMAS (Tasseled Cap) TRANSFORM
Brightness : ((MO*.3037)+(Ml*.2793)+(M2*.4743)+(M3*.5585)+(M4*.5082)+(M5*.1863))
Greenness : ((MO*(-.2848))+(Ml*(-.2435))+(M2*(-.5436))+(M3*.7243)+(M4*.0840)+(M5*(-.1800)))
Wetness : ((MO*. 1509) + (M1*.1973) + (M2*.3279) + (M3*.3406) + (M4* (-.7112) ) + (M5* (-..4572)))
NormStress : ((((((M3**2)/M4)-M2)/(((M3**2)/M4)+M2))*127)+127)
NornDiff : ( ((M3-M2)/(M3+M2)))
PRINCIPAL COMPONENTS (Moscow TM)
Eigenvalues: 3610.533203 686.259888 255.051575 65.266129 31.927553 19.888693 4.04020
PCO- ((MO*(0.599741))+(Ml*
PCI- ((MO*(-0.257315)) +(Ml
PC2- ((MO* (0.527533)) +(Ml*
PC3- ((MO*(-0.299325)) +(Ml
PC4- ((MO*(0.436562)) +(Ml*
PCS- ((MO*(-0.102381))+(Ml
PCS- ((MO* (0.071733)) +(Ml*
HYDROTHERMAL ALTERATION
Red gun: (MO / Ml)
Grn gun: (Ml / M2)
Blu gun: (M2 / M3)
(0.343662) ) + (M2*(0.456674) ) + (M3*(0.232064) ) + (M4*(0.398616) ) + (M5*(0
*(-0.044480)) + (M2*(0.210288)) + (M3*(-0.635556)) + (M4*(0.305435) ) +(MS
(0.096978)) +(M2*(0.116357)) + (M3*(-0.593837)) + (M4*(-0.568068)) +(MS*
*(0.323399)) +(M2*(0.445601) ) + (M3*(0.342477)) + (M4*(-0.564085)) +(MS*
(-0.264745)) + (M2*(-0.471570)) + (M3*(0.242320)) + (M4*(-0.121913)) + (M5
*(-0.033738))+(M2*(-0.202029))+(M3*(0.010212))+(M4*(-0.290847))+(K
(-0.833486))+(M2*(0.521473))+(M3*(0.116008))+(M4*(-0.086412))+
MSS Equations for Spectrum Display
NDVI : ((M3-M1) / (M3+M1))
TVI : (SQRT (((M3-M1) / (M3+M1)) +0.5) )
DVI : (2.4 * M3) - Ml
AVI : (2.0 * M3) - Ml
KAUTH-THOMAS
Brightness :
Greenness :
Yellowness :
Non-Such :
(Tasseled Cap) TRANSFORM
(MO* (0.332)) + (Ml* (0.603)) + (M2*(0.675)) + (M3* (0.262))
(MOM-0.283)) - (Ml*(0.660)) + (M2*(0.577)) + (M3*(0.388))
(MO*(-0.899)) + (Ml* (0.428)) + (M2* (0.076)) - (M3* (0.041))
(MOM-0.016)) + (M1*(0.13D) - (M2* (0.452)) + (M3*(0.882))
HYDROTHERMAL ALTERATION
Red gun: (MO / Ml)
Gm gun: (Ml / M2)
Blu gun: (M2 / M3)
-------
KRLC Consortium
Documentation Notebook
January, 1994
Imade Stratification Using ARC/INFO
Spatial stratification prior to clustering
One band of raw image data selected for strata delineation
Convert image to ARC "Image" format by:
Storing in ARC workspace with file extension .BIL
Create a .HDR file of # of rows, # of columns, pixel resolution,
and georeferencing information
Stratification Procedure:
Display image file in ARCEdit
Draw strata boundaries as arcs
Convert arcs to polygon coverage
Run POLYGRID
Run GRIDIMAGE and output file as BIL to form a strata image
Convert strata image to viff format
Turn into bit masks - within strata and outside strata
Apply each bit mask to each input band of multispectral data
•multispectral within-strata image
•multispectral outside-strata image
Construct and Project can then be run separately on stratified images.
A similar procedure can be run for post-interpretation stratification of
a classification. The khoros colortable is converted to a .CLR file for
ARC display of the classification as an image.
-------
MRLC Consortium
Documentation Notebook
January, 1994
nage Stratification Using khoros
Spectral stratification
One band selected to best discriminate between desired strata
A strata reflectance DN "threshold" chosen
Bit masks created: within-strata and outside-strata
Each bit mask is applied to each band of multispectral input image
•multispectral within-strata image
•multispectral outside-strata image
Construct and Project can the n be run separately on stratified images.
Choose Selection
Subsample
1 Threshold
•••••••••••^•••••••••H
Extract Sub Image
Dilation
Erosion
Median Filter
Invert Image
Print Image
Sun2VIff
Warp Image
Simple Idarp
HELP
QUIT
File Based Image thresholding utility* j
Inout Imaoe
Output Image
| Threshold Level |l28. W&BB&&SBB&K
K Output data type,
| Byte
Q Bitmap
| Invert Q False) •
Execute Help
-------
HRLC Consortium
Documentation Notebook
January, 1994
Use of Ancillary Data with Classifications (ARC/INFO)
Existing ARC/INFO datasets can be combined with images in the
hyper-clustering and interpretation process
•Before clustering, for stratification
•Before clustering, as an information band in the multispectral
image
•After interpretation, for clarification, plotting, selection by
feature
Classified Image to ARC/INFO
•Input as an ARC "Image" file
build .HDR file
convert khoros colormap to .CLR file
Allows image display
-^* Allows vector overlay
•Convert to ARC "GRID" format using IMAGEGRID
Allows image display
Allows vector overlay
Allows value query and selection from GRID
-------
MRLC Consortium
Documentation Notebook
January, 1993
7.2 MRLC Labelling Pilot Programs
This section is reserved for the future discussion and results
of the pilot programs for cooperative labelling and landcover
classification.
-------
MRLC Consortium
Documentation Notebook
April 1993
7.3 Landcover Classification References for MRLC Agencies
The following paragraphs provide information on references describing the landcover
classification requirements and efforts of the MRLC participating agencies.
GAP
o Jennings, MJ. 1993. Natural Terrestrial Cover Classification: Assumptions and
Definitions. GAP Analysis Technical Bulletin. U.S. Department of the Interior.
C-CAP
o Dobson, J., E. Bright and others. 1994. NOAA CoastWatch Chanpe Analysis
Project - Cruidspce for Regional Implementation.
o Klemas, V.V., J.E. Dobson, R.L. Ferguson, and K.D. Haddad. 1993. A coastal
land cover classification system for the NOAA CoastWatch Change Analysis
Program. Journal of Coastal Research. 9(3): 862-872.
o NODC Environmental Information Bulletin No. 92-3 describing the C-CAP
Chesapeake Bay landcover change analysis project (see Section 11.4 of this
notebook for a copy of this bulletin).
-------
MRLC Consortium
Documentation Notebook
April 1994
SECTIONS
MRLC METADATA STANDARDS
The MRLC Consortium is seeking to develop a unified approach to metadata which is
consistent with the metadata requirements of each of the participating programs, and conforms
to the metadata standards being developed by the Federal Geographic Data Committee. GAP
has developed a metadata content standard that will serve as the basis for reporting MRLC
metadata. This document entitled "Metadata Standards for Gap Analysis", is included in this
Section.
-------
-------
Metadata Standards For Gap Analysis
Department of The Interior
U.S. Fish and Wildlife Service
"and
National Biological Survey
Christopher B. Cogan, Idaho Cooperative Fish and Wildlife Research Unit, University of Idaho,
Moscow, ID 83844-1136
Thomas C. Edwards, Jr., Utah Cooperative Fish and Wildlife Research Unit, Utah State
University, Logan, UT 84322-5210
23 February 1994
-------
Introduction
Proper documentation of all information sources used to assemble Gap Analysis data
layers is central to the scientific defensibility of the Gap Analysis program. The information
used to describe Gap Analysis data is called metadata. Metadata are information about data.
Metadata contain information about the source(s), lineage, content, structure, and availability of
a data set. Metadata also provide descriptions of the intent and potential uses of data. Several
descriptions of metadata functions have recently been published by the Federal Geographic Data
Committee (FGDC) (1992,1993,1994) and by the National Research Council (1993). In this
paper, we outline metadata standards to be used by Gap Analysis cooperators. These standards
are mandatory for all work produced with Gap Analysis funding. These standards draw heavily
on existing metadata proposed by the FGDC (FGDC 1994), and are designed to meet rapidly
evolving national standards. Because the national standards are in flux, these standards can also
• *.
be expected to change. Subsequent editions of this document will be maintained on the Gap
Analysis Bulletin Board in Utah.
Need For Standardized Metadata
Historically, geographic data have been collected, analyzed, and documented by
individuals and agencies which needed little interaction and cooperation outside their immediate
discipline or organizational unit The rapid changes in data products, and increased accessibility
of spatial data has resulted in an increased demand for metadata standards. Today, vast amounts
of shareable research products are being generated without adequate documentation. Without
this documentation, there is a loss of information content and the data becomes less valuable.
As Gap Analysis has evolved, state, federal, and private agencies are increasingly
requesting and using Gap data to achieve their missions. Although Gap cooperators are familiar
with the basis of their particular data sets, outside users of the information are not.
Consequently, there is a need to provide users from a wide range of disciplines and organizations
with Gap information which is thoroughly documented.
With hundreds of researchers across the country contributing to Gap Analysis, another
role for metadata becomes critical: metadata provide a means of selective access to the data. For
example, there may be a need to search the Gap records for information on land ownership in a
specific geographic location. One type of information available in the metadata is a description
-------
of the spatial extent of a data base. A query based on latitude and longitude coordinates can
produce a list of all map products in the region described.
Another argument for requiring high quality, standardized metadata arises from our need
to reach scientifically defensible decisions which are based upon our data sets. A regional or
national survey of biodiversity and gap analysis may never be perfectly repeatable, for the
necessary human interpretation and ecological assumptions are extremely complex. We can
however, thoroughly document our sources, methods, and assumptions thus creating - to as great
an extent as practical - a scientifically defensible product.
There are currently several types of software which can query metadata. One promising
access system is the network of wide area information servers (WAIS) being tested now by the
Department of Interior (Nebert 1993). Another software tool called PGBIO has been developed
for cataloging and browsing of spatial data (Davis et aL in press). Other tools including Archie,
Gopher, and Mosaic are also gaining popularity (Krol 1992).
Demands for metadata will increase as electronic networks expand across the national •
and international scene and more requests are made for distribution of information. As the
number of users and the diversity of disciplines and programs sharing the data expand, the
information carried by metadata will become increasingly important One of the goals in
defining today's metadata standards is to anticipate these future needs.
Although these standards are continuing to evolve, it is critical to produce metadata
documentation as our current data sets are being created. Attempting to document work after it
has been completed is often costly, inefficient, and of poor quality. The most recognized and
current publication of metadata standards is the "Content Standards for Spatial Metadata" (1994)
available from the FGDC. These standards are being developed for compliance with data
transfer systems (i.e., WAIS), and data exchange format standards. The FGDC metadata
standard is being considered as a Federal Information Processing Standard (FTPS) while the
Spatial Data Transfer Standard (STDS) has already been adopted as FIPS 173.
Gap Analysis Metadata
The Gap Analysis metadata standards have been closely matched to the FGDC standards
to ensure current and future compatibility. As the FGDC standards evolve beyond the current
draft publication, we anticipate corresponding refinements in Gap documentation. The format of
-------
the Gap metadata consists of nine major documentation sections containing one or more
metadata elements. Each element is named (e.g. Map Projection Name), and the "Type" of entry
(text, integer, date, time) and "Domain" of the entry (i.e. x > 0) are also defined. Appendix A
describes the content Standards for the Gap Analysis Project Metadata. This list of elements
represents the mining™ metadata documentation required of Gap principal investigators. Where
additional documentation is available, we strongly encourage data set developers to maintain
conformity with current FGDC elements. At the time of this writing, there are an additional 209
elements defined by the FGDC which are not described in this document. To clarify the items
described in Appendix A, an example using the Gap Analysis land ownership map of Utah is
shown in Appendix B.
The metadata standards outlined in these appendices do not describe how to assemble a
detailed data dictionary. The dictionary is a specific subset of metadata which contains
• *.
definitions for attribute codes and lists the contents of tables which make up the data base. For
illustration, a sample data dictionary page from the Utah vegetation coverage is presented in '
Appendix C. In this example, a polygon labeled 1 Ala3 is described as a specific type of
mountain fir vegetation.
The full data set, including metadata and data dictionary, must be designed for digital
transfer which will allow the metadata to be distributed separately, however the actual database
must always contain the metadata. Regardless of the software used to query the metadata, a
digital ASCII version similar to Appendix B should be maintained.
-------
Acknowledgments
We wish to thank Thaddeus Bara, Blair Csuti, Frank Davis, Frank D'Erchia, Al Fisher,
Maurry Nyquist, Ann Rasberry and David Stoms for reviewing the first draft of the manuscript,
their comments were greatly appreciated.
Bibliography
Banks, Richard CM Roy W.McDiannid, and Alfred L.Gardner. 1987. Checklist of vertebrates
of the United States, the US. Territories and Canada. U.S. Fish and Wildlife Service,
Resource Publication 166 79pp.
Bourgeron, P. and L. Engelking. 1993. A preliminary vegetation classification of the western
United States. Draft report prepared for The Nature Conservancy by the Western Heritage
—Task Force, Boulder, CO. 400+pp.
Congalton, R.G. 1991. A review of assessing the accuracy of classifications of remotely sensed
data. Remote Sensing of Environment 37:35-46.
• •&.
Davis, Frank W., Steve MUey, David M. Stoms, Michael J. Bueno, and Allan D. Hollander. A
cataloguing and browsing tool for geographic data. Submitted to: The Professional
Geographer.
Federal Geographic Data Committee (FGDC). 1994. Draft, Content Standards for Spatial
Metadata. 25 January J994. 54pp.
.Federal Geographic Data Committee (FGDC). 1992. Draft, Content Standards for Spatial
Metadata. 3 November 1992.
Krol, Ed. 1992. The Whole Internet: User's Guide & Catalog. O'Reilly & Associates, Inc.,
Sebastopol, CA. 400 pp.
National Institute of Standards and Technology. 1992. Federal Information Processing
Standard Publication 173: Spatial Data Transfer Standard. U.S. Department of Commerce.
National Research Council. 1993. Toward a Coordinated Spatial Data Infrastructure for the
Nation. National Academy Press, Washington, DC 171 pp.
Nebert, Douglas D. 1993. Implementation of a wide area information server (WAIS) software
to disseminate spatial data on the internet, In Proceedings of the Thirteenth Annual ESRI
User Conference vol. L Environmental Systems Research Inst., Redlands, CA pp. 575-584.
The Nature Conservancy, Science Division. In progress. Element codes for vertebrate species
names from the central databases of the national heritage programs and conservation data
centers. The Nature Conservancy, Arlington, VA
U.S.D.A. Soil Conservation Service. 1992. Plants database access guide. U.S.DA., Soil
Conservation Service. Beltsville, MD 104pp.
-------
Appendix A.
Content Standards For Gap Analysis Project Metadata
Key to Metadata Element Entries
The elements described here are organized into nine general categories (see list
below). Each element is described, and characterize in the following format:
Data Element — An item used to describe a particular aspect of a data set. Examples
include Access Restrictions, Map Projection Name, and Source Date. If
additional elements are needed, refer to the FGDC (FGDC 1994) guidelines
before inventing new formats. Elements marked with an asterisk (*) in Appendix
A are not found in the FGDC version.
Type — A one word description of how theiof ormation is enoeded for each particular
data element. Options are: text, integer, real, date, time, and compound. For
example, in section n for the element "Distance Resolution", a single real number
is appropriate (e.g. 50.0), thus the Type is listed as "real".
Domain - The restrictions on the values that can be assigned to the element Options
include: free text, specific numeric bounds (i.e. x>0), or a list of keywords. This
is not a comprehensive list of all options, it is intended as a guideline to help
standardize the values. By using standardized domain values, automated
processing of metadata will be greatly facilitated.
Metadata Data Element Categories
I Identification Information VI Entity/Attribute Information
II Spatial Reference VH Distribution Information
IQ Status Information VHI Metadata Ref. Information
IV Source Information DC Contact Information
V Processing History Information
-------
I Identification Information
Data Set Identity — the name or tide by which the data set is known.
Type: text
Domain: free text
Identification Code - unique item or stock code by which the item can be
ordered, or the full path name to the file.
Type: text
Domain: free text "n/a" "Unknown"
Data Set Description — a description of the spatial data set, including its intended
use and limitations.
Type: text
Domain: free text
Theme Keyword - common-use word or phrase used to describe the thematic
content of a data set . » -*. .
Type: text
Domain: free text (see Appendix D for the default domain of keywords)
Thematic Accuracy - an estimate of the certainty of the identification of the
entities and assignments of values in the data set, expressed as a percentage. The
techniques used to determine this value will vary depending on the data structure
in use.
Type: integer
Domain: 0 <= Thematic Accuracy <= 100 "Unknown" "n/a"
Thematic Accuracy Explanation - a definition of the thematic accuracy
measure, and a description of how the estimate was derived, (e.g. deductive
estimate, misclassification matrix, random point sample). If the method relies on
specific source materials, they should be appropriately referenced.
Type: text
Domain: free text
Logical Consistency - an explanation of the fidelity of relationships encoded in
the data. Include a description of each test performed, (e.g. tests of valid values).
Type: text
Domain: free text "n/a" "Unknown"
Completeness — information about omissions, selection criteria, generalization,
definitions used, and other rules used to derive the data set. Include description of
minimum mapping unit.
Type: text
Domain: free text "n/a" "Unknown"
-------
Browse Graphic File Name - name of a related graphic file that provides an
illustration of the data set, including a legend for interpreting the graphic, and the
location of the graphic file.
Type: text
Domain: free text
Browse Graphic File Description - a text description of the illustration.
Type: text
Domain: free text
Browse Graphic File Type - graphic file type of a related graphic file.
Type: text
Domain:
Domain Value Definition
"COM" Computer Graphics Metafile
"EPS" Encapsulated Postscript
"GIF" Graphic Interchange Format
"JPEG" Joint Photographic Experts'Group format
"PBM" Portable Bit Map
"PS" Postscript
"TIFF' Tagged Image File Format
"XWD" X-WindowDump
Data Set Citation - the recommended reference to be used for the data set
Type: text
Domain: free text
Form Candidates:
Patrias, Karen. National library of medicine recommended formats for
bibliographic citation. Bethesda, MD : U.S. Dept. of Health and
Human Services; 1991.190p.
dark, Suzanne M., and Mary L. Larsgaard. Cartographic citations: a
style guide. Chicago, IL: Map and Geographic Roundtable, American
Library Association. 1992.23p.
Native Data Set Environment - a description of the data set in the producer's
processing environment. For digital data, include items such as the name of the
software (including version), the computer operating system, and file name
(including host and path names). Include a discussion of the medium and
appropriate scales.
Type: text
Domain: free text
8
-------
•Raster File Format - key words describing the software type and file structure.
Type: text
Domain:
Value Definition
"ADRG"
"ADRT
"BIL"
"BIP"
"BSQ"
"DEM"
"DGSTA"
-ERD73"
"ERD74"
"ERM"
"GRASS"
"GRAS4"
TDL"
"IGDS"
"IGES"
"IPW"
"MOSS"
"NTIF"
"SDTSR"
"SEP"
"SLF"
"SPECT"
"TIGRP"
"TIGRC"
ARC Digitized Raster Graphic
ARC Digitized Raster Imagery
Imagery, band interleaved by line
Imagery, band interleaved by pixel
Imagery, band sequential
U.S. Geological Survey Digital Elevation Model format
Digital Geographic Information Exchange Standard (DIGEST)
Annex A -ISO 8211 form
ERDAS image files, version 73
ERDAS image files, version 7.4
Earth Resources Mapper image file.
Geographic Resources Analysis Support System, version 3
Geographic Resources Analysis Support System, version 4
Interactive Data Language image file
Interactive-Graphic Design System fomatd(Intergraph Corporation)
initial Graphics Exchange Standard
Image Processing Workbench image file
Multiple Overlay Statistical System export file
National Imagery Transfer Format
Spatial Data Transfer Standard raster profile
Standard TnyarJiangp. Format (DOD Project 2851)
Spectrum image file (based on Khoros Visual Image File Format)
Tagged Image File Format
Topologically Integrated Geographic Encoding f"d Referencing
System, pre-census version
Topologically Integrated Geographic P-nm^jng and Referencing
System, census version
*Raster File Sensor - Key words describing the sensor type used for data
collection.
Type: text
Domain: landsat tm, landsat mss, noaaavhrr, spot, aircraft scanner,
aircraft video, scanned air photos, scanned hard copy, radar, other
-------
* Vector File Format - key words describing the software type and file structure.
Type: text
Domain:
Value Definition
ARC/INFO Export format, version 5
ARC/INFO Export format, version 6
ARC/INFO Export format, version 7
ASCII file, formatted for text attributes, declared format
User-created coordinate file, declared format
Digital Feature Analysis Data
Digital Geographic Information Exchange Standard (DIGEST)
Annex A - ISO 821 1 form
U.S. Geological Survey Digital Line Graph-Optional format
U.S. Geological Survey Digital Line Graph-Standard format
AutoCAD Drawing Exchange Format, version 9
AutoCAD Drawing Exchange Format, version 10
AutoCAD Drawing Exchange Format, version 1 1
Interactive Graphic Design SystemJbniat (Intergraph
Corporation)
Spatial Dam Transfer Standard topological vector profile
Standard Interchange Format (DOD Project 2851)
Standard Linear Format
Topologically Integrated Geographic Encoding and
Referencing System, pre-census version
Topologically Integrated Geographic Encoding and
mo System*
"ARCE5"
"ARCE6"
"ARCET
"ASCII"
"COORD"
"DFAD"
"DGSTA"
"DLGO"
"DLGS"
"DXF9"
"DXF10"
"DXF1 1"
"IGDS"
"SDTSV
"SIF"
"SLF"
"TIGRP"
"TIGRC"
"VET" Vector Product Format (MIL-STD-600006) (also known as
Digital Geographic Information Exchange Standard
(DIGEST) Annex C - Vector Relational form, and Vector
Relational Format)
*Nonspatial File Format ~ key words describing the software type and file
structure.
Type: text
Domain: arc/info graphic (.gra), arc/info plot file (.pit), pbm, wordperfect,
microsoft word, ascii, applix, postscript, eps, splus, tiff, gif, other
Use Restrictions - terms, including copyright, governing the use of the data set
after access has been provided.
Type: text
Domain: free text
Access Restrictions - restrictions imposed on access or distribution of the data
set.
Type: text
Domain: free text
10
-------
II Spatial Reference
Native Data Structure — the mechanism used to represent the data (i.e. spatial,
Nonspatial, raster, vector, quadtree, tin)
Type: text
Domain: free text
Map Projection Name - name of the map projection.
Type: text
Domain:
Albers Conical Equal Area
Azimuthal Equidistant
Equidistant Conic
Equirectangular
General Vertical Near-sided
Geographic' (not projected)
Gnomonic ... • «*. •
Lambert Azimuthal Equal Area
Lambert Conformal Conic
Mercator
Miller Cylindrical
Modified Stereographic for Alaska
Oblique Mercator
Orthographic
Perspective
Polar Stereographic
Polyconic
Robinson
Sinusoidal
Space Oblique Mercator
Stereographic
Transverse Mercator
Universal Transverse Mercator
Van der Grinten
Unknown
*Map Projection Description - projection parameters used to project the map.
Description should include items such as units, spheroid, standard parallels,
central meridian, false easting, and false northing. For arc/info coverages, this
information can usually be copied from the "describe" listing.
Type: text
Domain: free text
11
-------
Distance Resolution — the minimum distance measurable between two points,
expressed in (ground) meter: With maps of high quality, such as the USGS
1:24,000 scale quadrangles, the distance resolution in map units is approximately
1/2 mm. To convert this measure to ground units, divide the denominator of the
map scale by 2000 (i.e., 1:24,000 scale yields a distance resolution of 12 meters).
In other cases, the pixel cell size may also be used.
Type: real
Domain: x > 0.0
Point/Vector Vertical Resolution - the minimum distance between two adjacent
elevation values, expressed in meters.
Type: real
Domain: z > 0.0
Raster File Row (Line) Count — the maximum number of raster objects along
the ordinate (y) axis. The number of rows or lines in the image.
Type: Integer
Domain: RowCount">0 " • *r
Raster File Column (Sample) Count - the maximum number of raster objects'
along the abscissa (x) axis. The number of columns or samples in the image.
Type: Integer
Domain: Column Count > 0
Raster File Depth (Band) Count - the maximum number of raster objects along
the depth (z) axis. For use with rectangular volumetric raster objects (voxels).
Type: Integer
Domain: Depth Count >0
•Raster File Number of Bytes per Pixel - A descriptor for the range of values
possible for each image pixel.
Type: Integer
Domain: 1,2,4
Horizontal Positional Accuracy — an estimate of the locational certainty of the
horizontal coordinate measurement in the data set (i.e. longitude) expressed in
meters. This measure should not be confused with Distance Resolution.
Type: real, text
Domain: x >= 0, unknown
Horizontal Positional Accuracy Explanation - a definition of the horizontal
positional accuracy measure and how the estimate was derived.
Type: text
Domain: free text
12
-------
Vertical Positional Accuracy - an estimate of the locational certainty of the
vertical coordinate measurement in the data set (i.e. latitude) expressed in meters.
This measure should not be confused with Distance Resolution.
Type: real, text
Domain: x >= 0, unknown
Vertical Positional Accuracy Explanation -- a definition of the vertical
positional accuracy measure and how the estimate was derived.
Type: text
Domain: free text
West Bounding Coordinate -western-most longitude coordinate of the coverage.
Type: real
Domain: -360
-------
IV Source Information
Source Identity - list of sources (e.g., landsat TM, existing vegetation maps
etc.) and a short discussion of the information contributed by each.
Type: text
Domain: free text
*Source Date — time of original data collection for source materials.
Type: text
Domain: free text
•Source Distance Resolution - the minimum distance measurable between two
points, expressed in (ground) meters. With maps of high quality, such as the
USGS 1:24,000 scale quadrangles, the distance resolution in map units is
approximately 1/2 mm. To convert this measure to ground units, divide the
denominator of the map scale by 2000 (i.e., 1:24,000 scale yields a distance
resolution of 12 meters). In other cases, the pixel celljiizc may also be used.
Type: real
Domain: x > 0.0
V Processing History Information
Process Description - description of the processing steps and tolerances used to
construct the data or the location of a text file or report describing same. Include
information about the responsible parties.
Type: text
Domain: free text
VI Entity/Attribute Information
Attribute Label — the code name and description of the primary attributes
associated with the data set. Vertebrate data should use the species codeset
developed by The Nature Conservancy (TNC, in progress). Plant species data
should use the U.S.D.A Plants Database Access Guide (U.S.D.A. 1992).
Vegetation community types should follow TNC Natural Heritage Network
(Bourgeron and Engelking, 1993).
Type: text
Domain: free text
14
-------
Attribute Definition Source - the authority of the primary database attributes.
Provide a full citation reference. For thematic classification data, identify the type
of system used (e.g. Anderson, Cowardin, Driscoll, Holland, TNC, UNESCO).
Type: text
Domain: free text
*Related Documents — Reference citations and location information on
documents directly relating to the data set. Examples include path name to the
data dictionary, reports and publications. Where possible, all gap-generated
reports should be available as Postscript files for electronic distribution.
Type: text
Domain: free text
VII Distribution Information
• 4*.
Distribution Contact - the party from whom the data set may be obtained. This
may be the local system administrator or a national distribution contact.
Type: compound
Domain: free text
Distribution Liability - statement of the liability assumed by the distributor.
Type: text
Domain: free text
Transfer Options — the ways in which the data set may be obtained or received,
and related instructions and fee information.
Type: text
Domain: free text
File Compression Technique - information on algorithms or processes applied
to the data set in its transfer format to reduce the size of the file.
Type: text
Domain: free text, "None"
Transfer Size - the size in megabytes of the data set
Type: real
Domain: Transfer Size > 0.0
15
-------
VIII Metadata Reference Information
Metadata Date ~ the date that the metadata were created or last updated.
Type: date
Domain: free date
Metadata Review Date — the date of the latest review of the metadata entry.
Type: date
Domain: free date; Metadata Review Date later than Metadata Date
Metai a Contact - the party responsible for the metadata information.
I'ype: compound
IX Contact Information
• -u
Contact Person — the name of an individual who can provide further information
on the dfltfl set
Type: text
Domain: free text
Contact Mail Address - the address of the organization or individual to which
the contact type applies.
Type: text
Domain: free text
Contact Voice Telephone - the telephone number of the organization or
individual to which the contact type applies.
Type: text
Domain: free text
Contact Facsimile Telephone - the telephone number of a facsimile machine of
the organization or individual to which the contact type applies.
Type: text
Domain: free text
Contact Electronic Mail Address - the address of the electronic mailbox of the
organization or individual to which the contact type applies.
Type: text
Domain: free text
16
-------
Appendix B.
Example metadata construction using land ownership from the Utah Gap Analysis Project.
I IDENTIFICATION INFORMATION
Data Set Identity
Utahland93
Identification Code
Utahland93
Data Set Description
Utah land ownership for Gap analysis, based on 1:100,000 scale source map
resolution and current as of 9/93.
Theme Keyword
ownership . „ ^
Thematic Accuracy
95%
Thematic Accuracy Explanation
Accuracy estimate based on misclassification matrix of completed coverage
against all main BLM source maps, as well as other independent BIA, Utah,
DWR, and USFS maps used for adding additional information to coverage.
Thematic accuracy assessment of the BLM source maps is unknown. For further
explanation of the matrix methods, see Congalton, R.G. (1991) and Richards,
J.A (1986).
Logical Consistency
Each attribute label was confirmed to be a member of the set of valid values, and
each land area has to have one and only one label.
Completeness
Base land ownership derived from BLM 1:100,000 quads. Additional land
ownership categories such as wilderness areas, wildlife reserves and Indian
reservations derived from USFS, Utah DWR and BIA sources. Land ownership
changes made since BLM quad map dates updated from USFS region 6
information. Land areas containing less than the 100 ha minimum mapping unit
(mmu) were dissolved.
Browse Graphic File Name
n/a
Browse Graphic File Description
n/a
17
-------
Browse Graphic File Type
n/a
Data Set Citation
Edwards, Thomas C. 1993. Utah land ownership map: based on 1:100,000 scale
BLMmaps. Digital GIS file format. UTCFWRU Utah State University, Logan
UT.
Native Data Set Environment
Arc/Info software version 6.1.2 running on a Sun4 with Solaris 22 OS.
Raster File Format
n/a
Raster File Sensor
n/a
Vector File Format " • *r
ARCE6
Nonspatial File Format
n/a
Use Restrictions
Copyright 1993. This information is not accurate for legal boundary, or
navigation purposes. This information is intended for use by the Gap Analysis
Project, any other use constitutes copyright infringement
Access Restrictions
No portion of this database may be stored, reproduced, or redistributed in any
manner without prior permission from: THOMAS C. EDWARDS, JR.,
UTCFWRU, UTAH STATE UNIVERSITY, LOGAN UT 84322-5210,
801-750-2529 (PH) tce@rsgisjir.usu.edu (INTERNET).
18
-------
II SPATIAL REFERENCE
Native Data Structure
spatial vector
Map Projection Name
Lambert Conformal Conic
Map Projection Description
Projection LAMBERT
Units METERS
Spheroid CLARKE 1866
Parameters:
1st standard parallel 3901 0.000
2nd standard parallel 4039 0.000
central meridian -111300.00
latitude of projection's origin 38 20 0.000
false easting (meters) 0.00000
false northing (meters) 0.00000
Distance Resolution (meters)
50
Point/Vector Vertical Resolution (meters)
n/a
Raster File Row (Line) Count
n/a
Raster File Column (Sample) Count
n/a
Raster File Depth (Band) Count
n/a
Raster File Number of Bytes per Pixel
n/a
Horizontal Positional Accuracy (max. displacement from true coordinates, in meters)
unknown
Horizontal Positional Accuracy Explanation
in progress
19
-------
Vertical Positional Accuracy
Not to exceed 125 meters
Vertical Positional Accuracy Explanation
unknown
West Bounding Coordinate (degrees longitude)
-114.125
East Bounding Coordinate (degrees longitude)
-108.997
North Bounding Coordinate (degrees latitude)
42.0
South Bounding Coordinate (degrees latitude)
37.0
ffl STATUS INFORMATION
Data Set Status
in use
Release Date
24Novl993
IV SOURCE INFORMATION
Source Identity
Land ownership information from: Bureau of Land Management Land Status
Maps, Utah Division of Wildlife Resources, U.S. Forest Service Bureau of Indian
Affairs, and U.S. Geological Survey. All maps printed at 1:100,000 scale.
Source Date
Map source dates vary from 1981-1993
Source Distance Resolution (meters)
50
20
-------
V PROCESSING HISTORY INFORMATION
Process Description
Land ownership boundaries were manually digitized form 1:100,000 scale paper
maps. Processing details are further described in the text file: Utahland93.txt,
included in the data set.
VI ENTITY/ATTRIBUTE INFORMATION
Attribute Label
USER: The land owner.
CD: The manager of the land.
Attribute Definition Source
n/a
Related Documents
in progress
VH DISTRIBUTION INFORMATION
Distribution Contact
Thomas C. Edwards, Jr., National Biological Survey, Utah Cooperative Fish and
Wildlife Research Unit, Utah State University, Logan, UT 84322-5210
phone: 801-750-2529,
fax: 801-750-4025,
internet: biod@rsgisjir.usu.edu
Distribution Liability
The National Biological Survey assumes no responsibility for application of the
data beyond their original intent
Transfer Options
Data is available through anonymous FTP, tape, CD, and other forms.
File Compression Technique
Arc/Info export file (ARCE6) with full compression option.
Transfer Size (megabytes)
17.2
21
-------
VIII METADATA REFERENCE INFORMATION
Metadata Date
24Novl993
Metadata Review Date
N/A
Metadata Contact
Thomas C. Edwards, Jr., National Biological Survey, Utah Cooperative Fish and
Wildlife Research Unit, Utah State University, Logan, UT 84322-5210
phone: 801-750-2529,
fax: 801-750-4025,
internet: biod@rsgisjir.usu.edu
IX CONTACT INFORMATION
Contact Person
Thomas C. Edwards, Jr.
Contact Mail Address
National Biological Survey, Utah Cooperative Fish and Wildlife Research Unit,
Utah State University, Logan, UT 84322-5210
Contact Voice Telephone
801-750-2529
Contact Facsimile Telephone
801-750-4025
Contact Electronic Mail Address
biod@rsgis.nr.usu.edu
22
-------
Appendix C.
Example of a data dictionary entry adopted from Utah vegetation map.
1. MOUNTAIN FIR
Douglas-fir, Pseudosuga menziesii
White fir, Abies concolor
1A Closed Forests.
1 Al. Mainly evergreen forests.
lAla. Temperate and subpolar evergreen coniferous forests
1 Ala3 Evergreen (non-giant) conifer forest with conical
crowns.
2. MOUNTAIN FIR / MOUNTAIN SHRUB
Douglas-fir, Pseudosuga menziesii (dominant) * ~~
White-fir, Abies concolor
Serviceberry, Amelanchier spp.
Alder leaf mountain mahogany, Cercocarpus montanus
Manzanita, Arctostaphylos patula
Bitterbrush, Purshia tridentata
Snowberry, Symphoricarpos spp.
Oregon Grape, Berbaris repens
Oak, Quercus gambelii
Maple, Acer glabrum
2A Closed Forests.
2A1. Mainly evergreen forests.
2Ala. Temperate and subpolar evergreen coniferous forests
2Alal Evergreen (non-giant) conifer forest with conical
crowns.
2B Scrubs (shrublands or thickets).
2B1. Mainly deciduous scrub.
2Blc. Cold deciduous scrubs.
2Blc4. Temperate deciduous scrub.
23
-------
Appendix D.
Default Theme Keyword Thesaurus*
ANTHROPOGENIC*
Administrative Units * Ownership ** State Boundaries **
Cadastral * Pipelines * Structures *
Census Units * Political Units * Transmission I-inf-s *
fVinm^inipafinfi j.in^c * Populated Places * Transportation *
County Boundaries ** Population* Waterways*
Geographic Grid ** Public Land Survey System * Zoning **
Landuse** Railroads*
Named Places* Roads*
ATMOSPHERIC COMPOSITION
AeiDSOlS Pnntmnrn«nt< Oxygen
Air Quality Humidity Ozone
Ash Methane Trace
Carbon Dioxide Nitric Acid Trace Gases
Chlorofluorocarbons Nitrogen Tracers
Clouds Nitrogen Dioxide Water Vapor
ATMOSPHERIC DYNAMICS
Air Quality ** Evapotranspiration Pressure
Altitude Geopotential Height Solar Radiation
Atmospheric Temperature HeatFmx Storms
Climate* Humidity Visibility
Cloud Types Paleodimate Indices Winds
Evaporation PnM*,ipi»qripp
BIOLOGICAL ENTITIES
Birds Habitat** Ocean Wildlife
Domesticated Animals Land Wildlife Protected Areas **
Domesticated Plants Microorganisms Surface Vegetation
Ecoregions ** Minor Species Vertebrates **
Endangered Species Ocean Vegetation
- Adapted from Directory Interchange Format (DIP) Manual, April 1993, version 4.1. section 2.11, 'Parameter
measured." Entries marked with an asterisk (*) are extensions to the DIP Manual. (**) are added for Gap analysis.
24
-------
Addiction
Bacterial
Cardiovascular
Chronic
Communicable
Dermatologic
Digestive System
Endocrine
Eye
Fungal
DISEASES
Inununologic
Injury
Musculoskeletal
Neonatal
Neoplasms
Nervous System
Nutritional pud Metabolic
Occupational
Ophthalmic
Otorhinolaryngologic
Parasitic
Poisoning
Pregnancy Complications
Respiratory
Skin
Stomatogastric
Urologic
Virus
Albedo
Brightness Temperature
Heat Bin
EARTH RADIATIVE PROCESSES
Tnyilarinn **
Irradiance
Radiance
Solar Activity
Temperature
EroswD
Geodesy
Geothennal
GEODYN AMIC FEATURES
Gravity Fields
Magnetic Fields
Polar Motion
Seismic
Structures
Tectonophysics
Terrain Elevation
Volcanoes
Albedo
Aspect**
Cultural Features
Elevation
Fixes
Glaciers
Hydrology **
GEOGRAPHY AND LAND COVER
Ice
Lakes
Rivers
Slope**
Snow
Soils
Surface Vegetation
Surf ace Water
Topographic Data
Watersheds **
Wetlands
GEOLOGICAL PARAMETERS
Age Determinations
Aquifer*
Coal
Economic Minerals
limology
Mineralogy and Crystallography
Paleontology
Petroleum
Petrology
Sedimentary rocks
Soils
Stratigraphy
Surficial Geology*
Care
HEALTHCARE
Community Care
Institutional Care
25
-------
MRLC Consortium
Documentation Notebook
January, 1994
SECTION 9
MRLC DATABASE MANAGEMENT
This section contains information on the database management
system being developed at the EROS Data Center to manage the TM
scenes and the MRLCMS database. A draft version of the Land
Science Data Archive Data Base Design Review, prepared and provided
by the EROS Data Center is included in this section. A formal
database design review for the MRLC project is under preparation at
the EDC, and will be included once it is available. The target
date for this report is currently February 1994.
-------
LAND SCIENCE
DATA ARCHIVE
DATA BASE
DESIGN
REVIEW
MRLC Consortium
Documentation Notebook
January, 1994
-------
MRLC Consortium
Documentation Notebook
January, 1994
LAND SCIENCE DATA ARCHIVE
DESIGN REVIEW AGENDA
1. Introduce Team Members:
D. Larsen, D. Knell, T. Smith, L. Hansen, R. Sunne, C. Larson
2. Land Science Data Archive Objectives.
3. Land Science Data Archive Logic.
4. Land Science Data Archive General Requirements.
5. Specific Project Data Sets.
A. Background
B. Individual Schematics
a. NALC
b. HTFI
C. NSDGT
d. EOS Test Sites
e. MRLC
f. S6C
6. RDBMS to be used and DSB priority development timeline.
-------
MRLC Consortium
Documentation Notebook
January, 1994
06/30/93
EDC Land Science Data Archive Data Base Design Project
Objectives
1. Accomodate existing and future land science data archive projects with a
functional data base design to provide metadata access and production
inventory support . Define DB data sets/fields, relational data set
interface and project requirements.
2. Consequently, this DB must be able to interface readily with'DORRAN for
ordering purposes.
3. This DB must be able to meet AMS project tracking/maintenance requirements
and DORRAN ordering requirements and provide periodic management reports
(statistical). This DB must be able to feed the Version 0 IMS system
and/or GLIS.
Specifically: A) Provide the ability to query the availability of project
data for each path/row and support geographic queries.
B) Provide the ability to search metadata (per path/row
and/or study region) which relates to each level of
derivative products generated (i.e. triplicates,
geo-registered, etc).
C) Provide relational constructs to ensure that higher level
datasets (i.e. cluster data sets, land cover thematic
maps, etc.) derived from each project can be indexed back
to the original source scenes resident in either this DB
or in the National Satellite Land Remote Sensing Data
Archive. This includes providing a tracking system for
contributor derivative data forwarded to EDC for
indexing, archiving and distribution.
4. Accessibility to outside users will be handled through the Version 0 IMS
and/or GLIS.
5. This DB must have the flexibility to add fields and additional data sets
when deemed necessary for new projects (will need software support).
-------
MRLC Consortium
Documentation Notebook
January, 1994
06/30/93
EDC Land Science Data Archive Data Base Design Project
Land Science Data Archive DB Logic
1. A maximum of three primary data sets apply to each project, they are the
project's:
BASIC DATA SET • this data set contains the original project data.
COMPLEX DATA SET * this data set contains the EDC product data generated
from the project's basic data.
DERIVATIVE DATA SET » this data set contains the derivative data '
generated from the EDC complex project data which
is returned to EDC from contributors for
archive/distribution purposes.
2. Each project will be put into Its own series of primary and secondary
data sets. The fields will be project specific in. both types of data sets.
3. All projects do not require all three primary data sets because those
specific projects COMPLEX data is not retained and/or DERIVATIVE data has
not been considered at this time (i.e. HTFI, SGC, NSDGT, etc.). Each
project will be handled on a case-by-case basis.
4. Secondary data sets can be added, as needed, to the primary data sets to
accomodate specific project metadata. These secondary data sets need to be
relationally compatible with the .primary data set it is connected to.
5. The BASIC DS and the COMPLEX OS needs to be relationally compatible via
links. The COMPLEX DS and the DERIVATIVE DS needs to be relationally com-
patible via links. An SQL or script is needed to relate the DERIVATIVE DS
record back to the BASIC DS via the COMPLEX DS to provide management
reporting.
-------
MRLC Consortium
Documentation Notebook
January, 1994
CO
UJ
CO
-
< or
or o
< H- O
o LU or
o H-
CO CO
O LU
z or
CS -J —1
O <
< CO CQ
h- UJ O
< or _i
O I
o *-
or z
H II
O —
—J U_
< H-
o.
Z S CO
II II II
H- O
O -J O
o or c9
CO Z CO
-------
MRLC Consortium
Notebook
January, 1994
Land Science Data Archive Data Base General Requirements
1. Each link is to be automatically incremented by 1 when a record is added.
2. Each date entered field is to be automatically populated with the date the
record is added.
3. Each date updated field is to be automatically populated when the record is
modified.
4. Software needs to be in place to calculate corner points from the BASIC OS
and capture this coordinate metadata to an ascii file for plotting purposes.
Software consideration is needed to provide standard footprint coordinate
calculation routines for Landsat, SPOT, Aircraft and AVHRR data...these
routines must be accessible from any data base and meet the needs of CAM
plotting format.
5. AMS will require all incoming derivative data to be converted to 3480
archive media. OOPS needs to assure all COMPLEX data is output to 3480
archive media.
6. When AMS enter the path, row and acquisition date into the DERIVATIVE DS
software needs to take this metadata and search the COMPLEX DS to find the
complex link. This complex link then needs to be automatically stuffed into
the derivative record(s) being entered.
7. At this time cross-inventory searching is not required, this should be ad-
dressed in Phase II Land Science Data Archive Data Base development on GLIS.
8.' Require software to automatically transfer pertinent metadata from Landsat/AVHRR
/Aircraft/SLAR/SPOT data base inventories to the appropriate landsat science
data archive data base project inventories.
9. All ordering id's first two characters serve to define appropriate data set
DORRAN requirement.
10. Automatically cross reference tape library to all data sets for updating of media
locations, seqa, and seqb and filejibr fields.
11. A Derivative data set product with more than one media location will require a
Derivative media relational data set to be added.
-------
MRLC Consortium
Documentation Notebook
January, 1994
22X (NALC) NORTH AMERICAN LANDSCAPE CHARACTERIZATION DATA
Background -
==========
Under an interagency agreement between the USGS and the Environmental Protection
Agency (EPA) to support the NALC project, the EDC is producing co-registered
Landsat data images for 803 Landat WRS Path/Row locations which cover most of
Alaska, Hawaii and the Conterminous U.S., Mexico, Latin America and the
Cam'bean. The data bases consist of from three to five geo-coded/co-registered
Landsat MSS scenes from three time periods (70's. 80's and 90's) and
co-registered digital terrain data; the end product is called a triplicate.
EDC Project Contact * John Dwyer
-------
NALC
NORTHAMERICAN LANDSCAPE CHARACTERIZATION
DATA SETS
MRLC Consortium
Documentation Notebook
January, 1994
(Data Origination
Source • NLRSOA)
(Data Type •
Landsat MSS Data)
NALC BASIC OS
NALC COMPLEX OS
NALC MOSAIC/
COMPOSITE OS
NALC DERIVATIVE OS
NALC
DERIVATIVE
CONTRIBUTOR DS
-------
PROPOSED LAND SCIENCE DATA ARCHIVE DATA BASE - NALC DATA SETS
MKLC Consortium
Documentation Notebook
January, 1994
06/30/93
^•NALC BASIC os —- — — - -
VFc link ~ <
ordering id i
orig pat" i
orig~row
sensor
orig_aca_date i
orig'cloud cover i
band~c.ua lily
projection I
sun_azimuth I
sun'elevation i
scene_ctr_lat I
scene'ctr'long i
usage'restrict 1
data Tormat 1
media location 1
tripllcate_path i
triplicate'row
triplicate'decade
date_entered
date'updated
S 6)
S 13
N 3
N 3
S 6
0 10
N 1
S 4
S 3
N 3
N 2
N 7
N 6
s *
S 6
S 6
N 3'
N 3
N 3
0 10
0 10
—• NALC COMPLEX OS —• — -
'basic link
'complex link
ordering id
path
row
decade
acquisition date
resampling used
base reg source
band'comEi nation
usage restriction
data Tormat
media location i
date entered I
date'updated 1
^^^^~
S 6
S 6
S 13
N 3
N 3
N 2
0 10
S 1
S 13
N 6
S 2
S 6
S 6
0 10
D 10
^•ALC DERIVATIVE OS"-*—'
^•rex link (
derTv ITnk
contrlb code
ordering id
path
row
acquisition date
S 6)
S 6
S 2
S 14
N 3'
N 3
0 10
description' (S 150
projection
resampling
band combination
usage restriction
data Tormat
media location
date entered
date'updated
S 3
S 2
S 10
S 2
S 6
S 6
0 10
0 10
"~-mC MOSAIC/COMPOSITE OS- *~
complex_link
sequence_nbr
scene id'
acqdale
ctlpts
rmserr
date entered
date'updated
S 6
N 2
S 13
0 10
N 4
N 4
0 10
0 10
-—TIALC DERIVATIVE CONTRIBUTOR
•contrlb code
contrib'name
contrib'email
contrib'phone
contrib'company
contrib'address
contrib'state •
contrib'zip
date entered
date'updated
S 2!
S 40'
S 24
S 14
S 40
S 40
S 2
S 10
0 10
0 10
-------
MRLC Consortium
Documentation Notebook
January, 1994
(HTFI) HUMID TROPICAL FOREST INVENTORY DATA
Background -
Under a NASA/USGS agreement to support the HTFIP, the EDC is producing digital
and photographic products for nearly 2700 Landsat MSS and TM (UTM projected)
scenes covering three time periods (70's, 80's and 90's) and three geographic
locations (Brazil, Central Africa and Southeast Asia). Only the original data
is retained by EDC.
EDC Project Contact * Dave Carneggie
-------
MRLC Consortium
Documentation Notebook
January, 1994
HTFI
HUMID TROPICAL FOREST INVENTORY
DATA SETS
(Data Oribination Source •
MLRSDA, Thailand. NSOGT/EOSAT)
(Data Type • Landsat HSS/TM
Data)
HTFI BASIC DS
HTFI DERIVATIVE DS
HTFI
DERIVATIVE
CONTRIBUTOR DS
10
-------
06/30/93
PROPOSED LAND SCIENCE DATA ARCHIVE DATA BASE -
MRLC Consortium
Documentation Notebook
HTFI DATA nuary' 1994
HTFI BASIC DS
*basicjink
orderTng_id
path
row
offset
sensor
acquisition_date
cloud cover"
band quality
projection
sun azimuth
sun'elevation
scene_ctr_l ati tilde
scene~ctr"longi tude
usage~reslriction
data_Tormat
media location
date_?ntered
date'updated
(S 6)
(S 16)
(N 3)
(N 3)
(N 2)
(S 6)
(D 10
N 1
S 8
S 3)
N 3)
N 2
N 7
N 8
S 2
S 6
S 6
D 10
D 10)
NO HTFI-COMPLEX DATA RETATNED. THUS NO DATA SET.
HTFI DERIVATIVE
'basic link
*deriv~link
*contrTb_code
ordering id
path
row
offset
sensor
acquisition date
description"
projection
resampling
band combination
usage restriction
data Tormat
mediallocation
date entered
date~updated
V*<" •• •
(S
S
S
S
N
N
N
S
D
S
(S
(S
(S
s
S
s
D
D
• w « i
6
6
2
14
3
3
2
w
3)
10)
150)
3)
2
10
2
6
6)
10)
10)
~"~HTFI DERIVATIVE CONTRIBUTOR DS
*cohtri b_code
contri b~name
contnb~email
contriib'phone
contri b~company
contrib~address
contrib~state
contrib~zip
date_entered
date'updated
(S 2
11
-------
HRLC Consortium
Documentation Notebook
January, 1994
(NSDGT) NASA DATA GRANT DATA
Background -
Under a NASA agreement with EOSAT a total of 400 TM scenes are to be purchased
and archived as source data for the NASA DATA GRANT program allowing other
land science data archive projects to use this data such as HTFIP.
EDC Project Contact = Bill Draeger
12
-------
(Data Origination Source,
NLRSDA, ESA, EOSAT)
(Data Type •
Landsat TH)
MRLC Consortium
Documentation Notebook
January, 1994
NSDGT
NASA DATA GRANT
DATA SET
NSDGT
BASIC DS
NO COMPLEX DS
NO DERIVATIVE DS
13
-------
MHLC Consortium
Documentation Notebook
January, 1994
06/30/93
PROPOSED LAND SCIENCE DATA ARCHIVE DATA BASE - NSDGT DATA SET
NSDGT BASIC OS ~
*basic link
ordering id
path
row
sensor
acqui s i t i on__date
cloud cover"
band quality
projection
sun_azimuth
sun~elevation
scene_ctr latitude
scene~ctr~longi tude
usage~restriction
data Tonnat
medi allocation
date entered
^^
(S 6)
(S 16)
(N 3)
(N 3)
(S 6)
(D 10
N 1
S 8
S 3
N 3
(N 2
N 7)
N 8)
S 2)
S 6)
S 6)
D 10)
D 10)
NO NSDGT COMPLEX DATA RETATNED, THUS NO DATA SET.
NO NSDGT DERIVATIVE DATA RETATNED, THUS NO DATA SET.
14
-------
MRLC Consortium
Documentation Notebook
January, 1994
(EOS) EARTH OBSERVATION SYSTEM TEST SITE DATA
Background -
SZ88SSSSZS
The concept of EOS test sites involve the maintenance of remote sensing data,
metadata and ancillary data for 60 to 70 test sites around the world to support
the development and testing of algorithms for EOS products. A strawman proposal
is currently being reviewed by the LAND DAAC Advisory panel. The complex and
derivative product types are TBD.
The test sites would involve the NLRSDA's AVHRR and Landsat MSS/TM data, as well
as other types of data for prototype exercises. The first site may involve
10 MSS scenes, 6 TM scenes and 540 AVHRR passes.
The basic data types include:
1) Landsat MSS (historical)
2) Landsat TM (satellites 4, 5 and 6)
3) AVHRR 1KM
4) DEM
5) Land Cover Maps
6) Aircraft data sets
7) Soils Maps
8) Synthetic Aperature Radar (SAR) data
9) Insitu data = field notes, data points
EDC Project Contacts « Paul Seevers, Bryan Bailey, Lyn Oleson, Dave Carneggie
15
-------
EOS TEST SITES
EARTH OBSERVATION SYSTEM TESTS SITES
DATA SETS
MRLC Consortium
Documentation Notebook
January, 1994
(Data Sources •
NLRSOA, EOSAT, ETC>)
(Data Type • Landsat
MSS/TH. AVHRR, Aircraft
Data Land Cover Naps,
Soils Naps, DEN, SAR)
£05 BASIC DS
EOS COMPLEX DS
EOS
DERIVATIVE DS
EOS
DERIVATIVE
CONTRIBUTOR DS
16
-------
PROPOSED LAND SCIENCE DATA ARCHIVE DATA BASE - EOS TEST SITES DATA SETS
TMOSAT DATA SET——
— ' - EOS BASIC OS
•basic link
(3'
-------
MRLC Consortium
Documentation Notebook
January, 1994
(MRLC) MULTI-RESOLUTION LAND CHARACTERISTICS DATA
Background -
SSSSSS88SS
The Multi-Resolution Land Characteristics monitoring system has the goal to
provide a current baseline of global multi-scale environmental characteristics
and mechanisms for monitoring, targeting and assessing environmental changes.
The MRLC objectives include:
1) Development of a global 1-KM land characteristics data base.
2) Development of a prototype multi resolution regional data base.
3) Development of a multi-resolution environmental monitoring system that:
a) Provides a framework and methods for quantifying change over time.
b) Monitors synoptic environment processes and targets significant areas of
change.
MRLC will involve basic data from the NLRSDA's Landsat MSS/TM and AVHRR
archives. The MRLC participants include:
1) EPA: EMAP « Environmental Monitoring and Assessment Program
2) WRD: NAWQA « National Water-Quality Assessment Program
3) NOAA: C-CAP * CoastWatch Change Analysis Program
4) USFWS: GAP = GAP Analysis Program
EDC Project Contacts * Jeff Eidenshink, Chuck Larson, Ron Feistner, Tom Holm,
Jim Sturdevant
18
-------
KRLC Consortium
Documentation Notebook
January, 1994
MRLC
MULTI RESOLUTION LAND CHARACTERISTICS
DATA SETS
(Data Origination Sources
KLRSOA, EOSAT. ETC>)
(Data Types • Landsat
NSS/TM, AVHRR)
MRLC BASIC DS
LANDSAT DS
AVHRR DS
A
I
MRLC
COMPLEX DS
MRLC
DERIVATIVE DS
MRLC
DERIVATIVE
CONTRIBUTOR OS
19
-------
MRLC Consortium
Documentation Notebook
PROPOSED LAND SCIENCE DATA ARCHIVE DATA BASE - MRLC TEST SITES DATA SETS Qftftfiiary, 1994
8/18/93
~ MRLC BASIC OS
•basicjink
orderTng_i d
sensor ~
path
row
acquisition_date
cloud_cover~
usage~restn'ction
data_7ormat
medi allocation
date_entered
datejjpdated
"~~ MRLC COMPLEX OS
*basic link
•complex link
ordering id
path
row
acquisition_date
resamplingjlsed
base_reg source
band"coinEi nat i on
usage restriction
data format
mediajocation
date entered
datejjpdated
MRLC DERIVATIVE
•complex link
•deriv link
*coritrTb_code
ordering" Id
path
row
acquisltionjjate
description"
projection
resampling
band_combination
usage restriction
data format
media location
date entered
date'updated
S 6
(S 16)
(S 6)
(N 3)
(N 3)
(D 10)
(N 1)
(S 2)
(S 6
(S 6)
I
1
IS ' '
D 10)
D 10)
S 6)
S 6)
S 16)
N 3)
(N 3)
fE or
iO 10)
S 2)
S 13)
N 10)
S 2
S 6
S 6
0 10
D 10;
S 6)
S 6
S 2
S 16
:N 3
X 3
D 10
(S ISO
[S 3
Is 2
S 10
:s 2
S 6
S 6
D 10
(D 10
~LANDSAT DATA SET
ordering id
sensor i?
sun_azTmuth
sun'elevation
scene ctr_lat
scene~ctr~lon
offset
projection
date entered
datejjpdated
AVHRR DATA SLI
ordering id
source Link
scan Line cnt
start time
stop Time
scene ctrj.at
scene~ctr~Lon
date entered
datejjpdated
(S 16)
(S 6)
(N 3)
(N 2)
(N 7)
(N 8)
(N 2)
(S 3)
(D 10)
(0 10)
(S 16
(N 6
(N 6
(N 6)
(N 6
(N 5
(N 6
(D 10
(D 10
1
~~*Rl.C DERIVATIVE CONTRIBUTOR DS~"~
*contr1b code
centribjiame
contrib'email
contribj>hone
contrib'company
contriheaddress
contrib'state
contrib~z1p
date_en?ered
date'updated
20
-------
(SGC) SPOT GLOBAL CHANGE DATA
Background -
MRLC Consortium
Documentation Notebook
January, 1994
To date (06/30/92) EDC has received 6 SPOT scenes.
EDC Project Contacts * Tom Holm, Wayne Rhode, Ron Meyer
See synopsis of Purchase of 800 SPOT scenes dated 12/31 '?. below:
U/ll/t*
it* Miimrr •**•« IM IIM.M* >• MIH* r»n«iM* w wor w
itiir
IIM Mil *MM(
«M* nl*M
•*•*!«•• «»»>•« • lt«t •! «t»«
•»«•»»» MM tttmf yn*t»MlT
yn*t»MlT *»****••» •» tta •.».
( 1141. M • IIM.M*
U »»•« •« •*•*•• MIOTtM »f «*• »T
kcnla •!!! M t»nnM« w u* ••«. »l* fwa*. It u
MMt «M *M«»» *kll M
•f rant»t •! W«M Mt M Uwr tMa Mtm II, It*}.
It u «M MMM (Ml UM nMtal*« 4M IIIMI Mil M
•MM* M Uwr 1MB MMMM Jl. IM1.
Mnr«a««t «lt* MOT Ml«cttwi
«• M U» *f*IKI»MJ* IM UH Mt
»«tkti«*MiMMii raui CM-** nr r
ri*» M* MMIIU Mr»MI MM*t MMIM C. Mil
CIM
U M« »T •*•». Ml. MM. •
tM Men' •! tM MI*CI»tU. M14. Ml. Mat.
M« tft >1>I MM ••«••• t« IM HOT ai*MI Cil«>l
*>«»!«• <•*•! tk« •••• *•«• ••• CMMittMM •• Mlt^M
OrMt.
c*rai
1.
I.
•t ia.*M.M
IM l»r Uv*l
*• • *M«mc SNT
SNT
D*tt
111*.
»00
1. Ml MMt M rM**M« *t *MT ky Cl*M *f kM*lM(* II MC*M*f
IM*.
«. fcir**** M4 *|M. tU UU Tam*ri.
ItMllf*. r*r«*t Mrvu* lOMck Bull).
1. ri*> ttmtnt MMMt «Ul M MlMtM >T tM mHCHoU •Ithin 10
My* •{ rM*i*t •< Ml. UMUM* *C*M* «11J M CM*» »7 10
JMJ.
«. ik* wet eiM«i ekMM MM Mt CM M MC**M« ky IB*
•M ky MM. UM B*MrtM«t •( tte Ut*cl*r |Wt>. tM H.s.
MMCUMM •< *|CU«lMt* (MM). M4 tM UvifMMOUl Fr*l«CttOO
. MX. ISM. *n« (fA
th« **••
7. •A/lilUtM' fMf** •« tM
•ill M»« MC**t «• tM MVT
ten* M4 ••Miti*** •• Mfla*4 kcratm.
I. ill MM 41*trtMt«4 Mt •< tM SMT ei«Ml Ck«M* «rchl«« «ill b.
•H»j«ct t* tM CMMral f*c«* MM) CaiUltl** •((•••••t ti ««(in«« in
tk« MctiM IM1« •! tM net/srOT IM|* C«r>«r*itwi
CMtfMt U»«M*tMl«UM«> *M*»t (*r UM f«llM.tik«i
•••Mr •! c»pi«* t* M »*M •( • (!••• MtM (*r «l»trlkuit0n i»
tM MU«rlM« •••!• M4 tMlt (ffUUtoi *M 2) tk« *|tnci«i k«««
tM rlffet t* MM41* MkMtl *f tM SMT Mt* lat* e**t*tt*t or CI-
IM *t tMir Mm *SMM*.
21
-------
KRLC Consortium
Documentation Notebook
January, 1994
SGC
SPOT GLOBAL CHANGE
DATA SETS
(Data Origination
Source: 000)
(Data Types: SPOT)
SGS BASIC DS
NO COMPLEX DS
NO DERIVATIVE DS
22
-------
MRLC Consortium
Documentation Notebook
April 1994
SECTION 10
MRLC ACCURACY ASSESSMENT
This section contains information on accuracy assessment issues and initiatives relating
to the efforts of the MRLC Consortium programs.
o In February 1994 the GAP sponsored an accuracy assessment meeting in Santa
Barbara, California. Meeting notes are included in Section 10.1. Conference
participants will also produce a technical report on GAP accuracy assessment
procedures which will be included in future updates.
o C-CAP has funded research through 1993 on change analysis accuracy
assessment Final reports, once available, will be included in this section.
o The EPA, through the Environmental Monitoring Systems Laboratory in Las
Vegas, NV, is participating in cooperative efforts to pursue research on methods
development for assessing and improving the thematic accuracy of large area
datasets derived from digital remotely sensed imagery. Relevant results and
reports will be included in this section as they become available.
-------
-------
MRLC Consortium
Documentation Notebook
April 1994
10.1 GAP Accuracy Assessment Workshop
In February 1994 the GAP sponsored an accuracy assessment meeting in Santa Barbara,
California. Conference participants will produce a technical report detailing GAP accuracy
assessment procedures. Currently included in this section is a summary of the workshop
prepared by the University of California at Santa Barbara. A copy of the final report will be
included in a future update.
-------
-------
MRLC Consortium
Documentation Notebook
April, 1994
From man Wed Feb 9 14:26 EST 1994
Received: from pollux.geog.ucsb.edu by ardgsv.rtpnc.epa.gov (5.4.2/200.1.1.4)
id AA02402; Wed, 9 Feb 1994 14:25:53 -0500
Received: from sage.geog.ucsb.edu by pollux.GEOG.UCSB.EDU id aa07872;
9Feb9411:26PST
To: als@rsgis.nr.usu.EDU, cogan@ursus.wildlife.uidaho.EDU,
reiners@corral.uwyo.EDU, jennings@uidaho.EDU, tce@rsgis.nr.usu.EDU,
reed@edcsnw20.cr.usgs.GOV, kelly@stein3.u.washington.EDU,
fd@crseo.UCSB.EDU, JED@ORNLSTC.bitnet, SHAW.DENICE@epamail.epa.GOV,
AMDLDW@vegasl.las.epa.GOV, tolsen@heart.cor.epa.GOV,
good@goodrs.GEOG.UCSB.EDU, stevste@suvm.acs.syr.EDU,
gpthelin@sl01dcascr.wr.usgs.GOV,
Denny_Grossrnan+aNA_HERITAGE+aTNCHQ%Nature@mcimail.COM,
tbara@ardgsv.rQmc.epa.gov, cconvis@esri.COM,
moisen@edumam.math.usu.EDU, tmuir@quarsa.usgs.GOV
Subject: accuracy assessment (LONG) report outline & mail list
Date: Wed, 09 Feb 1994 11:25:34 -0800
From: David Stoms
Message-Id: < 9402091126.aa07872@poUux.GEOG.UCSB.EDU >
Content-Type: text
Content-Length: 14341
X-Lines: 439
Status: RO
thanks to everyone for attending the workshop last week and for
the lively discussion and especially for the strong direction
the workshop generated for the GAP program.
attached is an ascii version of the revised report outline from
the workshop, including the updated list of all participants.
please review the outline to make sure it captures the key points
and decisions, if you are assigned one of the sections, please e-mail
me your draft text by the end of february. i will then incorporate it
into the text for a round of review by the group before we send it
to mikes scott/jennings by the end of march, for those of you from other
federal programs, please note that there is a small writing request
for you too in the appendix for a brief (1-2 paragraphs) summary of
the potential relationship of your program to GAP accuracy assessment.
thanks for your help, i think we are well on the way to a useful
document.
David Stoms stoms@sage.geog.ucsb.edu phone:(805)893-7044
U.C. Santa Barbara/Biogeography Lab fax:(805)893-3146
-------
MRLC Consortium
Documentation Notebook
April, 1994
Assessing Vegetation Map Accuracy for Gap Analysis
Draft 2/07/94
1. Preface by Mike Jennings
-brief introduction explaining compilation of vegetation maps for GAP, need
for accuracy assessment guidelines, peer review report recommendations,
interagency coordination objective for accuracy assessment
A workshop was held in Santa Barbara, California on February 3-4, 1994, with
representatives for both the GAP program, other federal mapping programs, and
The Nature Conservancy (see appendix for list of participants). The group
covered of a broad spectrum of disciplinary interests, from vegetation mappers
and remote sensing specialists, to botanists, hydrologists, and statisticians.
The purpose of the workshop was to outline the major issues of accuracy
assessment and to develop a recommended protocol to be used for state GAP
projects. This report presents the results of that discussion and is intended
as guidance for state GAP principal investigators. It will also inform
reviewers of GAP and potential users of GAP vegetation maps of the methods to
be used and the data to be compiled for accuracy assessment to allow them to
determine if they are suitable for their own purposes.
2. Uses of Gap Analysis Vegetation Maps by David Stoms
GAP vegetation maps are primarily compiled to answer the fundamental question
in gap analysis: how much of each vegetation type is there and how well
protected are they? A second product derived from the vegetation maps is for
predicting the distribution of vertebrate species and from these data to
determine how well protected each species is. As a minimum, the primary users
of GAP vegetation maps will need to have an estimate of overall map accuracy.
Besides giving a measure of reliability of the vegetation map for gap analysis,
the assessment will also need to identify which classes or which portions of
the map do not meet the accuracy standards for the GAP program. Thus the
assessment will identify where additional effort will be required when the map
is updated. Note: A map that does not meet the standard for all classes will
not be rejected, nor will gap analysis using such a map be delayed. Rather,
the results of the accuracy assessment will be reported with the analysis
stating that the map is the best available information).
We fully realize, however, that GAP is creating the first interme, :& scale
vegetation maps of the United States, and as such, these data will be of
-------
MRLC Consortium
Documentation Notebook
April, 1994
interest to a large number of other potential users. The maps should be useful
for biogeographic analyses and as input to mesoscale models of climate and
ecosystem processes. They will be used to evaluate smaller scale land cover
maps to assess their accuracy. Clearly, this type of user will need to know
the error characteristics of the GAP data in making their evaluation. GAP maps
will be used in coordinated national interagency land cover assessments. And
they are already being used in local and regional conservation planning
efforts. Many of these secondary users will need to know the per class
accuracy of the maps. Some users would also benefit from having some measure
of accuracy by polygon or geographic area, such as where the map is most
reliable. We expect that in general, secondary users will need more details
about the accuracy assessment to make appropriate uses of the map than primary
users who are more familiar with the data will.
3. Purposes of the Map Accuracy Assessment by David Stoms
3.1. Types of error and their impacts on map utility
Locational accuracy:
Thematic accuracy:
Measurement accuracy:
Temporal accuracy:
3.2. Scientific and programmatic criteria for the assessment
3.3. Uses of GAP accuracy assessment data
4. Approaches and Measures by Chris Cogan
—literature review
what measures have been used; what units of measure: pixel, polygon, or maplet?
4.1. Locational accuracy
4.2. Thematic accuracy
5. General Constraints by David Stoms
5.1. Technological constraints
5.2. Logistical constraints
5.3. Financial constraints
6. Sampling and Measurement Strategies by Tom Edwards and Frank Davis
6.1. Specific constraints
-------
MRLC Consortium
Documentation Notebook
April, 1994
6.2. Options
—tony olsen strongly suggested specifying the range of alternatives considered
and why the others were not recommended; specify which accuracy assessment
questions can or cannot be answered by each alternative
high = full access to all locations, level 5 classification
medium = preferred alternative, reasonable access, level 4 with some level 5
low = aerial photos to level 4 only, no field visits
6.3. Proposed sampling design
-sampling units will be regular shapes, not mapped polygons, and must be large
enough to capture heterogeneity representative of the mapped polygons = 1
square kilometer units
—shape of sampling unit not decided at meeting; problem with squares not
fitting inside linear polygons
—stratification: by region within state (state's option for regionalization)
to ensure samples throughout the geographic area
—also stratify by boundary vs interior of polygons, buffer polygon boundaries
by 500 m. if randomly selected center point falls within the buffer zone,
replace it with a new point.
—randomly select center points of 1 square kilometer and accumulate by class
until you have an adequate number in each class in each region (define
adequate).
-rare types: no decision reached on whether to require rare types to be
sampled or not
6.4. Proposed measurement strategy
—segment sampling unit on HAPP aerial photos with 1 ha MMU to guide selection
of sample points or plots within the sampling unit
-compile the set of Level 5? classes in field for each sampling unit
—some classes like agriculture could be measured from aerial photos instead
-------
MRLC Consortium
Documentation Notebook
April, 1994
—do not a priori assume a site is inaccessible by being far from roads; make a
good faith effort to reach it; if it is definitely inaccessible, then drop or
replace it.
-field crew can bring topo maps and aerial photos (with overlay of segmented
types) along to help find sampling unit but must not bring the GAP map which
might bias interpretation
7. Projected Costs by Tom Edwards, Allan Falconer, and Bill Reiners
at high, medium, and low cost options
-frank to provide table of number of samples needed to achieve specified
confidence levels, etc. from cdf report
8. Analysis and Reporting of Assessment by Frank Davis and David Stoms
—for location accuracy, just report RMSE of image registration
—what percent of samples were inaccessible? what percent of center points
fell in boundary buffer and where not sampled as estimate of sampling probabil-
ity?
-compare set of map polygon attributes (classes) with the list for each
sampling unit in the field, build contingency table, Kappa statistic ??? (refer
to EROS validation effort with AVHRR data).
-metadata documentation, see FGDC and GAP standards
—document which classes did not meet standard and why
9. Management of Accuracy Assessment Data by David Stoms
10. Research Priorities in Vegetation Map Accuracy Assessment by Kelly
Cassidy
how to quantify the contribution of uncertainty from observational error
to the apparent map error
the potential of videography as a sampling tool
determine how much area of each class is inaccessible from roads (this
could even be done as part of funding proposal for each state)
-------
MRLC Consortium
Documentation Notebook
April, 1994
develop new statistics for using maplets in accuracy assessment
11. References
Chrisman, N. R., 1991. The error component in spatial data, in: Geographical
Information Systems - volume 1: Principles, edited by D. J. Maguire, M. F.
Goodchild, and D. W. Rhind. Longman Scientific & Technical, pp. 165-174.
Congalton, R. G. 1991. A review of assessing the accuracy of classifications
of remotely sensed data. Remote Sensing of Environment, 37: 35-46.
Fenstermaker, L. K., 1991. A proposed approach for national to global scale
error assessments, in Proceedings of GIS/LJS'91, Atlanta, Georgia, October
28-November 1, 1991, pp. 293-300.
Goodchild, M. F., F. W. Davis, M. Painho, and D. M. Stoms, 1991. The Use of
Vegetation Maps and Geographic Information Systems for Assessing Conifer
Lands in California, NCGIA Technical Report 91-23, NCGIA, Santa Barbara,
California, 75 p.
*** goodchild et al for cdf recommendations
Goodchild, M. F., and S. Gopal, editors, 1989. The Accuracy of Spatial
Databases. Taylor & Francis, London.
story and congalton
Jennings, M. D., 1993. Natural Terrestrial Cover Classification: Assumptions
and Definitions. Gap Analysis Technical Bulletin 2, U. S. Fish and
Wildlife Service, Moscow, ID.
Merchant, J., and others, 1993. Validation of continental-scale land cover data
bases developed from AVHRR data, in Proceedings of Pecora 12, Sioux Falls,
SD, August 24-26, 1993.
Moisen, G. G., T. C. Edwards, Jr., and D. R. Cutler, 1994. Spatial sampling to
assess classification accuracy of remotely sensed data, in Proceedings of
the Symposium on Environmental Information Management and Analysis:
Ecosystem to Global Scales, Albuquerque, NM, May 20-22, 1993. National
Science Foundation, Washington, D. C., in press.
-------
MRLC Consortium
Documentation Notebook
April, 1994
UNESCO, 1973. International Classification and Mapping of Vegetation. United
Nations Educational, Scientific and Cultural Organization, Paris, 35 pp.
12. Appendices
12.1. Summary of relationships with other agencies and organizations
Note: each summary should be 1-2 paragraphs, and cover the same basic points:
purpose of mapping program, classification scheme, accuracy assessment method
and data collected, how GAP accuracy assessment field data would be useful to
your program or vice versa
C-CAP by Jerry Dobson, ORNL
NALC by Dorsey Worthy, EPA
Conterminous US Land Cover Database, Brad Reed, USOS
NAWQA by Gail Thelin, USGS
EMAP by Denice Shaw
MRLC by Denice Shaw
National Park Service Land Cover Mapping by Charles Convis, ESRI
TNC by Denny Grossman
12.2. Summary of classification scheme definitions and levels
-TNC and GAP are revising the 1993 technical report by Jennings
12.3. List of participants at the workshop, February 3-4, 1994, Santa
Barbara, CA
Thaddeus Bara
ManTech Environmental Technology, Inc.
2 Triangle Drive
Research Triangle Park, NC 27709
(919) 541-2755
tbara@ardgsv.rtpnc.epa.g6v
-------
MRLC Consortium
Documentation Notebook
April, 1994
Kelly Cassidy
Wash. Coop. Fish and Wildlife Unit
University of Washington
Seattle, WA 98195
(208) 685-4195
kelly@stein3.u.washington.edu
Christopher Cogan
Idaho Coop. Fish and Wildlife Unit
College of Forestry
University of Idaho
Moscow, ID 83844-1136
(208) 885-5788
cogan@ursus.wildlife.uidaho.edu
Charles Convis
ESRI Conservation Program
380 New York Street
Redlands, CA 92373
(909) 793-2853 x 1529
cconvis@esri.com
Frank Davis
Department of Geography
U.C. Santa Barbara
Santa Barbara, CA 93106-4060
(805) 893-3438
fd@geog.geog.ucsb.edu
Jerry Dobson
Oak Ridge National Lab
Box 2008, MS 6237
Oak Ridge, TN 37831
(615) 574-5937
JED@ORNLSTC.bitnet
Thomas Edwards
UTCFWRU
Utah State Univ.
Logan, UT 84322-5210
(801) 750-2529
tce@rsgis.nr.usu.edu
-------
MRLC Consortium
Documentation Notebook
April, 1994
Allan Falconer
Department of Geography
Utah State Univ.
Logan, UT 84322-5210
(801) 750-1334/750-1790
als@rsgis.nr.usu.edu
Michael Goodchild
Department of Geography
U.C. Santa Barbara
Santa Barbara, CA 93106-4060
(805) 893-8049
good@geog.geog.ucsb.edu
Violet Gray
Department of Geography
U.C. Santa Barbara
Santa Barbara, CA 93106-4060
(805) 893-7044
gray@geog.ucsb.edu
Dennis Grossman
The Nature Conservancy
1815 N. Lynn Street
Arlington, VA 22209
(703) 841-5300
Denny_Grossman+aNA_HERTTAGE+aTNCHQ %Nature@mcimail.COM
Mike Jennings
Idaho Coop. Fish and Wildlife Unit
College of Forestry
University of Idaho
Moscow, ID 83844-1136
(208) 885-6336
jennings@uidaho.edu
Gretchen Moisen
USDA Forest Service
FIA-INT Research Station
507 25th Street
Ogden, UT 84401
(801) 625-5384
moisen@edumath.usu.edu
-------
MRLC Consortium
Documentation Notebook
April, 1994
Tom Muir
National Biological Survey
413 National Goiter
Reston, VA 22092
(703)648-5114
tmuir@quarsa.usgs.gov
Tony Olsen
USEPA
200 SW 35th Street
Corvallis, OR 97331
(503) 754-4790
tolsen@heart.cor.epa. GOV
Brad Reed
USGS, EROS Data Center
Sioux Falls, SD 57198
(605) 594-6012
reed@edcsnw20.cr.usgs.gov
Bill Reiners
Department of Botany
Box 3165 University Station
Laramie, WY 82071
(307) 766-2235
reiners@uwyo.edu
Denice Shaw
EMAP Center
USEPA
Research Triangle Pk, NC 27711
(919) 541-2698
SHAW.DENICE@epamail.epa.GOV
Mirjam Stadelmann
ESRI Applications Division
380 New York Street
Redlands, CA 92373
(909) 793-2853 x 1796
mstadelmann@esri.com
-------
MRLC Consortium
Documentation Notebook
April, 1994
Steve Stehman
SUNY-ESF
320 Bray Hall
Syracuse, NY 13210
(315) 470-6692
stevste@suvm.acs.syr.edu
David Stoms
Department of Geography
U.C. Santa Barbara
Santa Barbara, CA 93106-4060
(805) 893-7044
stoms@sage.geog.ucsb.edu
GailThelin
USGS-WRD
2800 Cottage Way, RM W2234
Sacramento, CA 95825
(916) 978-4645
gpthelin@sl01dcascr.wr.usgs.gov
Dorsey Worthy
USEPA/EMSL-LV/AMS
944 £ Harmon Ave.
Las Vegas, NV 89119
(702) 798-2200
AMDLDW@vegasl .las.epa.gov
-------
-------
MRLC Consortium
Documentation Notebook
April 1994
SECTION 11
INFORMATION ON PARTICIPATING AGENCIES
This section describes information specific to each of the participating programs
regarding their involvement in the MRLC Consortium. The section is intended to include
MRLC-related reports and documents, as well as updates on agency requirements and/or usage
of MRLC data and resources. Actual documents are included when practical. Other documents,
for which their size is too large to be included in this notebook are referenced in the appropriate
section and are being held in the MRLC Consortium central file system.
-------
-------
MRLC Consortium
Documentation Notebook
April 1994
11.1 EMAP
The following documents describe the activities of EMAP, and are being held in the
MRLC Consortium files.
o EMAP Program Guide (6/93)
o EMAP Project Descriptions (9/93)
-------
-------
MRLC Consortium
Documentation Notebook
April 1994
11.2 GAP
The following documents describe the activities of GAP, and are being held in the MRLC
Consortium files.
o GAP Analysis Technical Bulletin 2: Natural Terrestrial Cover
Classification: Assumptions and Definitions (2/93)
o Gap Analysis: Geographic Information for Conserving Biodiversity
(Michael Scott and others, Wildlife Monographs, in print)
o GAP Analysis Bulletin No. 3, Winter/Spring 1993 (a semi-regular bulletin
describing ongoing GAP activities; Bulletin No. 4 is expected to be
released in the near future)
-------
-------
MRLC Consortium
Documentation Notebook
February 1995
11.3 NAWQA
This section contains the following NAWQA documents related to the MRLC
Consortium:
• Implementation Plan for the National Water-Quality Assessment Program
(U.S. Geological Survey Open-File Report 90-174, 1990)
The following document describes the activities of NAWQA, and is being held in the
MRLC Consortium files:
• Concepts for a National Water-Quality Assessment Program (U.S.
Geological Survey Circular 1021, 1988)
The U.S. Geological Survey maintains a repository of information on NAWQA which
can be accessed through the following World Wide Web home page:
• http://wwwrvares.er.usgs.gov/nawqa/nawqa_home.html
-------
JSKLC Consortium
Documentation Notebook
January, 1994
IMPLEMENTATION PLAN FOR
THE NATIONAL WATER-QUALITY
ASSESSMENT PROGRAM
U.S. GEOLOGICAL SURVEY
Open-File Report 90-174
-------
MRLC Consortium
Documentation Notebook
January, 1994
IMPLEMENTATION PLAN FOR THE NATIONAL WATER-QUALITY
ASSESSMENT PROGRAM
By P.P. Unity, J.S. Rosenslwin, and D.S. Knopman
U.S. GEOLOGICAL SURVEY
Open-File Report 90-174
Reston, Virginia
1990
-------
MRLC Consortium
Documentation Notebook
January, 1994
DEPARTMENT OF THE INTERIOR
MANUEL LUJAN, JIL, Secretary
US. GEOLOGICAL SURVEY
Dallas L. Peck, Director
For additional information
write to:
Assistant Chief Hydrologist, Program
Coordination and Technical Support
U.S. Geological Survey
414 National Center
12201 Sunrise Valley Drive
Reston, Virginia 22092
Copies of this report can
be purchased from:
U.S. Geological Survey
Books and Open-File Reports
Federal Center, Building 810
Box 25425
Denver, Colorado 80225
-------
CONTENTS
MRLC Consortium
Documentation Notebook
January, 1994
Page
Abstract 1
Introduction 1
Plan for implementation of program 1
Study-unit investigations 2
Investigation phases 2
Scope of activities . 2
Study-unit reports • 2
Regional and national synthesis of study-unit results 7
Regional and national water-quality concerns 7
Regional and national reports - 8
Coordination 10
References 10
FIGURES
Page
Figure
1. Map showing location of proposed study units for the National Water-Quality Assessment
Program '. 3
2. Graph showing schedule of first cycle of study-unit investigations, by dominant activity,
for the National Water-Quality Assessment Program, fiscal years 1991-2002 6
3. Graph showing information provided at different scales by the National Water-Quality
Assessment Program 9
TABLES
Page
Table
1. Proposed study units for the National Water-Quality Assessment Program 4
2. Examples of water-quality concerns of national and regional interest to be addressed by
the National Water-Quality Assessment Program and examples of policy questions
supported by this information 8
m
-------
MRLC Consortium
Documentation Notebook
January, 1994
IMPLEMENTATION PLAN FOR THE NATIONAL WATER-QUALITY
ASSESSMENT PROGRAM
By PJ*. Leahy, J.S. Rosenshein, and D.S. Knopman
ABSTRACT
The National Water-Quality Assessment (NAWQA)
Program is designed to describe the status and trends in
the quality of the Nation's ground- and surface-water
resources and to provide a sound understanding of the
natural and human factors that affect the quality of
these resources. To meet its goals, the program will
integrate information about water quality at different
spatial scales—local, study unit, and regional and
national—and will focus on water-quality conditions
that affect large areas or are recurrent on the local scale.
As part of the program, study-unit investigations will
be conducted in 60 anas throughout the Nation to
provide a framework for national and regional
water-quality assessments. The study-unit investigations
will consist of intensive assessment activity of 4 to
5 years duration followed by 5 years of less intensive
activity. Twenty study units will be in an intensive data-
collection and analysis phase during each fiscal year
(FY), and the first cycle of intensive investigations
covering the 60 study units win be completed in
FY2002. . . ' '"•
National and regional assessments of ground-and
surface-water quality will be provided from
issue-oriented findings of nationally consistent
information from the study units. By including study
units (60) that cover both a large part of the United
States and diverse hydrologic systems that differ in their
response to natural and human factors, the NAWQA
Program ensures that many critical water-resources and
water-quality concerns or issues can be addressed by
comparative studies that are national and regional in
scale.
INTRODUCTION
The Nation's water resources are composed of
many interrelated ground- and surface-water systems.
The response of each of these systems to natural and
human factors manifests itself b a corresponding set
of hydrologic, chemical, and biological characteristics
that reflect the water-quality effects of these factors.
Many national water-quality concerns arise from the
recognition of recurring local and regional problems
related to managing and protecting water quality. In
order to address these complex concerns and related
issues, the U.S. Geological Survey (USGS) proposed
a National Water-Quality Assessment (NAWQA)
Program in 1985 to:
(1) provide a nationally consistent description of
current water-quality conditions for a large part of the
Nation's water resources;
(2) define long-term trends (or lack of trends) in
water quality; and
(3) identify, describe, and explain, to the extent
possible, the major natural and human factors that
affect observed water-quality conditions and trends.
In 1986, a pilot NAWQA program was begun, the
purpose of which was to develop, test, and refine
methods useful for a full-scale national water-quality
assessment program (Hirsch and others, 1988). In
1987, the USGS requested the National Academy of
Science's (NAS) Water Science and Technology
Board to review the NAWQA pilot program. In
September 1989, the NAS review committee
submitted an interim report, which stated that (1) the
implementation of a national water-quality
assessment is in the best interest of the Nation, and
(2) the USGS is well qualified to csfrHfcb and imple-
ment a NAWQA Program. In late 1989, the
Administration determined that the USGS should
proceed with implementation of the NAWQA
Program in FY 1991 and requested that Congress
appropriate $18 million to begin the full program,
which in 4 years is planned to increase to about
$60 million annually. Background information on the
objectives, design, and plan of implementation for the
program is provided in this report
PLAN FOR IMPLEMENTATION OF PROGRAM
The NAWQA Program consists of two major
elements—study-unit investigations and regional and
national syntheses of study-unit investigation results.
Study-unit investigations, the basic building blocks of
the NAWQA Program, are designed to address study
unit and local water-quality issues and to provide the
framework upon which regional and national water-
quality assessments can be made. Findings from these
comparative studies will provide an improved under-
standing of key national, regional, and local
water-quality concerns.
-------
Study-Unit Investigations
The major activities of the NAWQA Program are
organized to take place within a set of hydrologic
systems referred to as study units. Sixty study units
(fig. 1, table 1), in which both ground- and surface-
water quality will be studied, have been identified.
Collectively, the study units encompass about 45 per-
cent of the land area of the conterminous United
States, an area in which withdrawals account for 60 to
70 percent of the Nation's water use as measured by
total withdrawal and population served by public
water supply. The water resource to be emphasized in
each study-unit investigation will depend on water use
in the study unit and the nature and importance of the
ground- or surface-water-quality concerns. Coordi-
nating activities among the USGS and representatives
of Federal, State, and local interests will aid in identi-
fication of water-quality concerns. In FY1991, plan-
ning and some limited water-quality sampling will
begin in 20 study units. Selection of these 20 units will
be based on the following criteria: (1) coverage of
major hydrologic regions, (2) coverage of agricultural
areas in keeping with the President's Water-Quality
Initiative, (3) consideration of water-quality concerns
-and programs of other Federal and State agencies,
and (4) water-quality concerns of the USGS. ... .
: Investigation Phases
The assessment activities beach of the study units
will include 4 to 5 years of continuous and intensive
data collection and analysis, immediately followed by
5 years of less intensive assessment activities (chiefly
intermittent monitoring of water quality). The study-
unit investigations win be conducted so that one-third
will be in intensive assessment activities at a given
time. In 12 years (FY 1991-2002), an intensive activity
period will have been completed for all 60 study units.
The schedule of investigations by principal activity for
the NAWQA Program for FY 1991-2000 is shown in
figure 2. During the less-intensive low-level activity
period of each study-unit investigation, a project chief
and one or two support project members will be
needed to continue assessment activities. During the
intensive period, as many as 10 project members hav-
ing expertise in a wide range of scientific disciplines,
including ground- and surface-water hydrology, water
quality, geochemistry, ecology, geomorphology, and
statistics will be involved in a study-unit investigation.
Scope of Activities
Major activities to be performed as part of the
study-unit investigations include the compilation of
MRLC Consortium
Documentation Notebook
.... .. . , . January, 1994
available water-quality information, sampling and
analysis of water quality for a wide array of physical,
chemical, and biological properties, and the
interpretation and reporting of results. Although the
NAWQA Program is designed as an operational pro-
gram, the approaches to be used will be "state-of-thc-
science" techniques and methodologies. Throughout
the program, improved methods will be developed
and adapted to meet the objectives of the program.
Priority will be given to the development of
(1) improved analytical methods for quantifying the
concentrations of trace elements and trace-organic
compounds in water, sediment, and tissue; (2) biolog-
ical assessment techniques; (3) methods for evaluat-
ing ground- and surface-water quality; and (4)
stafiyfjgyl and deterministic techniques of data analy-
ses and interpretation on a regional and national
scale.
Water-quality data available from water-resource
agencies at all governmental levels will be assembled,
screened, and evaluated to the extent possible. These
data will be stored in the computerized USGS data
base for the study-unit investigations. Additional
water-quality data collected specifically for the study
units including quality-assurance and ancillary infor-
mation, such as local land use, wOl be stored in the
computerized data bases and made readily accessible.
The intent of this effort is to ensure that the data can
be used effectively and efficiently for the study-unit
investigations and for regional and national synthesis
of study-unit investigations results by USGS and other
Federal, State and local agencies, academia, and the
private sector.
Study-Unit Reports
Results of each study-unit investigation will be
presented in several reports during each period of
intensive activity. Early in each investigation, the
project team will prepare a wcrk plan. This plan will
present refined boundaries of tne study unit, describe
the hydrogeologic setting of the study unit, identify
major water-quality concerns, define specific objec-
tives of the assessment, and describe approaches that
will be used. Briefing materials on planned water-
quality assessments will be prepared and released to
the public to aid in coordinating and ensuring that
local interests are addressed by the program, and to
keep the public informed about activities in each
study-unit investigation. Informal reporting on
activities through participation in public meetings
addressing local water-quality concerns will be an
important component of the study-unit investigations.
$
-------
MRLC Consortium
Documentation Notebook
January, 1994
-------
MRLC Consortitim
Documentation Notebook
Table 1.—Proposedstudy units for the National Water-Quality Assessment ftogonaary, 1994
Map identification
(Fig. 1)
Study-unit name
Statc(s)
NORTHEASTERN REGION
1. New Hampshire-Southern Maine Basins
2. Southeastern Nr*v pngl^nH
3. Connecticut Valley Drainage
4. Hudson Basin
5. Long Island and New Jersey Coastal Plain
6. Delaware Basin
7. LoWCT Sng/|tielianna Ratin
8. Dehnarva Peninsula
9. Potomac Basin
10. Allegheny and Monffigabftla Pflsins
1L Kanawha Basin
12. Lake Erie-Lake Saint Claire Drainage
13. Great and Little Miami River Basins
14. White River Basin ' ,.-
15. Upper Illinois River Basin
16. Lower Illinois River Basin ""'"..
17. Western Lake Michigan Drainage
18. Minneapolis-St Paul Basin
19. Red River of the North
SOUTHEASTERN REGION
20. Albemarle-Pamlico Drainage
21. Upper Tennessee River Basin
22. Santee Basin and Coastal Drainage
23. Apalachicola-Chattahoochee Basin
24. Georgia-Florida Coastal Plain
25. Southern Florida
26. Kentucky River Basb
27. Mobile River and Tributaries
28. Mississippi Embayment
29. Chicot-Evangclinc
30. Lower Tennessee River Basin
ME,NH,MA
MA.RI
NH,VT,MA,CT
NY,VT,MA,CI;NJ
NY.NJ
NY,NJ,PA,DE
PA.MD
DE,MD,VA
WV.MD.VA
NY.PA.WV
WV.VA.NC
MI, OH, IN
OH
IN
IL,IN,WI
IL
WI.MI
MN
MN.ND
NC.VA
TN.NC.VA
SC.NC.GA
GA.FL.AL
FL.GA
FL
KY
AL.MS
MS, LA, AR, TN, KY, MO
LA
TN,AL,KY
-------
MRLC Consortium
Documentation Notebook
January, 1994
Table I.—Proposed study units for the National Water-Quality Assessment Program—Continued
Map identification
(Fig. 1)
Study-unit name
Statc(s)
CENTRAL REGION
31. Eastern Iowa Basins
32. Ozark Plateau
33. Central Oklahoma
34. Trinity River Basin
35. Balcones Fault Zone
36. Central Nebraska Basin
37. Kansas River Basin
38. Upper Arkansas River Basin
39. Central High Plains
40. Southern High Plains
41. South Platte Basin
42. North Platte Basin
43. Cheyenne and Belle Fourche Basins
44. Yellowstone Basin
45. Upper Colorado Basin
46. Rio Grande Valley
47. Great Salt Lake Basins
48. Northern Rockies Intermontane Basins
WESTERN REGION
49. Upper Snake River Basin
50. Southern Arizona
51. Mid-Columbia Basin
52. Yakima River Basin
53. Puget Sound Drainages
54. Willamette Basin
55. Sacramento Basin
56. Western Great Basin
57. San Joaquin-Tulare
58. Santa Ana Basin
59. Oahu
60. Cook Inlet Basin
IA.MN.IL
MO,AR,OK,KS
OK
TX
TX
NE
KS,NE,CO
CO
KS,TX,OK,CO
TX.NM
CO,WY,NE
WY,CO,NE
SD.WY
MT;WY,ND
. ;OO,UT
NM.CO
UTJID.WY
MT;ID,WA
ID.WY.NV
AZ
WA
WA
WA
OR
CA.OR
NV.CA
CA
CA
HI
AK
-------
MKLC Consortium
Documentation Notebook
January, 1994
03-0
003
O>
cn o tn
to
3 i
*o o
o» S
I 8
CO —
"x S
o a
* «
^ ^
2? c
i i
a
o
O
•
o>
"5.
CO
"8
o
o
V)
-------
The results of each of the study-unit investigations
will be described as appropriate in technical journals,
reports for the general public, and USGS formal and
informal series reports. A series of interpretative
reports presenting results of the investigation will be
prepared at the completion of each period of inten-
sive activity. The first chapter is reserved for a sum-
mary of key findings. Subsequent chapters may
include an analysis of available water-quality informa-
tion and more detailed discussion of pertinent find-
ings from the intensive activity period and previous
less intensive activity periods. ~.
Regional and National Synthesis of
Study-Unit Results -
Regional and national synthesis of information
from selected study units will be the foundation for
comprehensive assessments of the Nation's water
quality. The synthesis activities will consist of compar-
ative studies of specific water-quality issues using
nationally consistent information and will focus on
differences and similarities in observed water-quality
conditions, trends, and causes of these conditions and
trends among the 60 study units. To permit m«Mi«ig-
ful comparisons a major part of the synthesis activities
will be the characterization of each study unit in terms
of nationally consistent information on water quality
and factors such as land use, geology, climate, agricul-
tural practices, and hydrology. Some of the synthesis
activities wQl focus on water-quality issues that affect
large contiguous hydrologic regions. Other synthesis
activities wfll focus on large noncontiguous areas that
are affected by similar specific water-quality issues or
concerns.
An example of a specific water-quality issue is the
presence of atrazine, one of the most heavily applied
herbicides in the United States. Most of the usage of
atrazine is concentrated in agricultural areas in the
Midwest, along the Mid-Atlantic coast, and in specific
regions of many other States. Thus, a "regional" anal-
ysis of the presence of atrazine to natural and human
factors would focus on several large noncontiguous
geographical areas of the Nation. Therefore, the
NAWQA approach to synthesis of study-unit investi-
gation results provides a unique opportunity to exam-
ine the presence of this herbicide in ground and
surface water in different parts of the country that are
characterized by distinct differences or similarities in
climate, hydrology, and agricultural practices.
Regional and National Water-Quality Concerns
Some of the national water-quality concerns to be
addressed in the first cycle of NAWQA studies along
MRLC consortium
Documentation Notebook
January, 1994
with regional and national water-quality policy
questions are given in table 2. These water-quality
concerns are comprehensive and represent a wide
range of difficulty and scope. The regional and
national synthesis of information from study-unit
investigations wfll significantly contribute to answer-
ing fundamental water-quality questions facing the
Nation. For example, a concern that will likely be
addressed during the early years of the program is the
relation of the presence of pesticides in ground and
surface water to application rates, cropping practices,
and climatic, geologic, and soil factors. Information
on the factors affecting ground- and surface-water
contamination by pesticides will be useful to
water-resource policymakers and managers for
(1) developing effective water-resource manage-
ment approaches regarding pesticide contamination,
(2) determining the appropriate pesticide
standards for particular geographic regions and
hydrologic settings rather than using rigid nationwide
standards that may overprotect the resource in some
areas and underprotect it in others, and
(3) developing effective and efficient ways to
monitor water-quality.
By including a large number of study units (60) and
a large part of the United States, the NAWQA Pro-
gram ensures that many critical water-quality con-
cerns in diverse hydrologic and land-use settings can
be evaluated. Water-quality concerns to be covered by
the regional and national synthesis will be reviewed
periodically and refined on the basis of findings from
study-unit investigations and other programs, and
advice from USGS coordinating and technical
advisory committees.
The relation among national, regional, study-unit,
and local scales of study are shown in figure 3. This
figure also summarizes the type of information that
will be reported for each scale of study. Because of
the interdependence between the study-unit investi-
gations and the regional and national synthesis of
study-unit investigations results, elements of the pro-
gram are being concurrently planned. Planning activ-
ities are being coordinated at both the regional and
national levels with appropriate Federal, State, and
local interests. Detailed planning of the regional and
national synthesis activities will begin in FY 1990.
These plans will affect ancillary data needs, the
emphasis of the local scale investigations in selected
study units, and to some extent, the staging of
study-unit investigations.
-------
MRLC Consortium
Documentation Notebook
January, 1994
Table 2.—Examples of water-quality concerns of national and regional interest to be addressed by the
National Water-Quality Assessment Program and examples of policy questions supported by this information
Water-Quality Concerns
• Occurrence and concentration of pesticides in ground and surface water and their relation to human and
aquatic health criteria,
• Relation of the presence of pesticide in ground and surface water to application rates, cropping practices
and climatic, geologic, and soil factors,
• Relative magnitude of various point- and nonpoint-source contributions to different types of ground- and
surface-water contamination,
• Effects of agricultural best management practices on ground- and surface-water quality,
• Regional occurrence and concentration of trace elements and industrial organic compounds in ground and
surface water, and -
• Effects of changes in municipal wastewater-treatment practices on water quality and ecosystem health.
Policy Questions
• AT^ "^'""al '"atf r-
-------
MRLC Consortium
Documentation Notebook
January, 1994
JI
u
CO
J
O
H
jj-j O
s-f
?1
CO
O
8.
2 i<£
5|
£ g
.8*
JJ1
REGIONAL SCALE
(Multistudy unit)
•ies for large hydrologic reg
•ies for geographic regions t
by (he specific issues under
c c*°
E E!
3 3*5
COCO'S
t
STUDY-UNIT SCALE
(Stream-aquifer systems ;
il summaries
iographic descriptions
area identification
.0 M E"
••sis
S 2 P
1= M
? 1 -8
C Q> •*
« /" •?
LOCAL SCALE
parts of stream-aquifer syst
1 geographic descriptions ol
uality conditions
jment of cause-and-effect re
1 = ^~
3 " « >
vi j?
• •
.—
*co
•O «
C ^
* c
cj
CO
•£ §
3 S
CO
>
-------
As the program progresses, the regional and national
synthesis reports will address more complex
water-quality concerns in greater detail.
COORDINATION
External coordination at all levels is an integral
component of this program. Information exchange
and coordination through study-unit liaison commit-
tees in the pilot program was highly successful and
this coordinating mechanism will be used extensively
to ensure local input to the 60 study units. liaison
committees will help ensure that the water-quality
information produced by the program is relevant to
regional and local interests. The liaison committees
will be comprised of non-USGS members who repre-
sent a balance of technical and management interests.
Represented organizations win include, as appropri-
ate, Federal, State, interstate, and local agencies,
Indian Nations, and universities. Specific activities of
each I'fl'son committee will include (1)
information about water-quality issues of regional
and local interest, (2) identifying sources of data and
information, (3) fK«fii««ng adjustments to program
design, (4) assisting in the design of project products,
and (5) reviewing and commenting on planning
documents and project reports. ,-.
A Federal/non-Federal advisory subcommittee
specifically designated for the NAWQ A Program wDl
be formed to ensure that both Federal and
Don-Federal interests and needs at the regional and
national level are met The USGS Office of Water
MRLC Consortium
Documentation Notebook
January, 1994
Data Coordination provides staff assistance to
Geological Survey advisory committees for water
resources and will provide support to the NAWQA
committees.
Finally, in addition to these activities and
committees, communication and coordination of
NAWQA and other USGS Programs with other Fed-
eral agencies will continue through several inter-
agency committees and Memorandums of Agreement
specifically developed to meet the need of the
NAWQA Program. Appropriate interagency commit-
tees include, for example, the U.S. Environmental
Protection Agency/U-S. Geological Survey Inter-
agency Committee for Program Coordination, and
interagency committees with the National Oceanic
and Atmospheric Administration, Office of Surface
Mining. U.S. Bureau of Reclamation, U.S. Forest
Service, and US. Soil Conservation Service.
The USGS is exploring a number of approaches to
ensure that national, regional, and local concerns are
effectively taken into consideration in the program
and that Federal, State, and local agencies have
opportunities to participate in and influence the pro-
gram; they wiD be kept apprised of data availability
and finding? that result from the program.
REFERENCES
Hirsch, R.M., Alley, W.M., and Wilber, W.G., 1988,
Concepts for a National Water-Quality Assess-
ment Program: US. Geological Survey Circular
1021,42 p.
«OS.GOVERNMENT nUXTINC OFFICE. 1»1« -*S1- 0»?«OOZ2
10
-------
MRLC Consortium
Documentation Notebook
February 1995
11.4C-CAP
The attached NODC Environmental Information Bulletin No. 92-3 describes the C-Cap
Chesapeake Bay landcover change project which has been recently completed. Additional C-
CAP Related documents being held in the MRLC Consortium files include:
• NOAA CoastWatch Change Analysis Project - Guidance for Regional
Implementation (J. Dobson, E. Bright and others, 1994)
• Klemas, V.V., I.E. Dobson, R.L. Ferguson, and K.D. Haddad. 1993. A
coastal land cover classification system for the NOAA CoastWatch
Change Analysis Program. Journal of Coastal Research. 9(3):862-872.
NOAA's Coastal Ocean Program maintains a repository of information on its programs
which can be accessed through the World Wide Web. A WWW page describing projects
undertaken by C-CAP may be found at the following location:
• http://hpcc.noaa.gov/cop/ccap.html
-------
NODC Environmental Information Bulletin No. 92-3
COASTWATCH CHANGE ANALYSIS PROJECT (C-CAP)
Chesapeake Bay Land Cover Classification Data, 1984 and 1988-89
The National Oceanographic Data Center
(NODC) announces the availability of a data set
showing changes in land cover for the Chesapeake
Bay area over me 5-year interval from 1984 to 198&S9.
This data set is the result of a multiyear effort that has
focused on the Chesapeake Bay as a prototype for fee
CoastWatch Change Analysis Project (C-CAP) of the
NOAA Coastal Ocean Program's Estuarine Habitat
The goal of C-CAP is to develop a comprehensive,
nationally standardized information system for
monitoring land cover and ha^**at' change in the
coastal regions of me United States. Its purpose is to
improve understanding of coastal uplands, wetlands,
and sea grass beds and their linkages with the distri-
bution, abundance, and health of living marine
.resources. The coastal region of the US wfll be
monitored every one to five years depending on me
rate and magnitude of change in each region.
The effort emphasizes a geographic approach
including the use of geographic information systems,
the Landsat Thematic Mapper (Figure 1), other
satellite sensors, and aerial photographs. C-CAP has
served as a catalyst for cooperative development
among many separate government agencies involved
in wetlands management and land cover analysis,
These include the National Wetlands Inventory
(NWD of the U. S. Fish and Wfldlife Service, the
Environmental Monitoring and Assessment
(EMAP) of the Environmental Protection Agency
(EPA), and the National Mapping Division of the U. S.
Geological Survey (USGS).
The Project Basconducted nine workshops,
involving more than 250 specialists, and is currently
funding five protocol development projects; Designed
to address some of the most vexing problems in large
pffffl fanj COVET chaqgf anplysiSt C-CAP may
serve as a model for rftniiy efforts in global
environmental monitoring.
The Chesapeake Bay data sets were devel-
oped with additional funding from the Chesa-
peake Bay Program and collaboration with the
Maryland Department of Natural Resources,
Salisbury State University, the Chesapeake
Research Consortium, the Virginia Institute of
Marine Sciences, the University of Maryland,
and the Federal agencies listed above,
The Chesapeake Bay data set constitutes one
of the largest change detection efforts ever
attempted, covering an area of approximately
30,000 square mfles with a source data resolu-
tion of 30 meters by 30 meters. Its greatest value
is in its synoptic coverage and consistent classifi-
cation over such a large area.
Figure 1. Gray-scale rendering of color Coast-
Watch image for the entire Chesapeake Bay area.
This image was based on Thematic Mapper data
for August 27 and September 21,1984.
UA DEPARTMENT OF COMMERCE
National Environmental Satellite, Data, and Information Service
National Oceanographic Data Center
December 1992
-------
Product Description
The Chesapeake Bay Land Classification Data Set
is based on an analysis nf landsat Thpfnatir Mapper
(IM) scenes of the Chesapeake Bay area (Table 1).
TaMel. LsndsatTMi
PATH ROW DATES
14 33 O9-21-1984 11-03-1988
14 34 09-21-1984 11-03-1988
15 33 08-27-1984 10-12-1989
15 34 08-27-1984 10-12-1989
The ^ata set consists of three magnetic tapes:
Tape 1. Analysis nf fnitr T nnrisat
scenes from 1984,
Tape 2. Analysis of four Landsat Thematic Mapper
scenes from 1988-89 for me same area, and
TapeS. Analysisofthe resulting change between '
1984 and 1988*9.
The land raver analyses distinguish 14 land cover
classes: (1) developed-high intensify, (2) developed-
low intensity, (3) cropland, (4) grassland, (5) decidu-
ous forest, (6) evergreen forest, (7) mixed forest, (B)
mixed shrub/scrub, (9) pahistrine forest, (10) estua-
rine emergent wetland, (11) pahistnne emergent
wetland, (13) tidf^ *!**«, (13) ^«]»Htf^1 land (bare soil
+ sand), (14) water. The change analysis resulted in
81 classes.
Each data set has several ASCII header records
• Product Name
• The UTM zone number
• The computer and operating systems used to
create the tape
• A fltntcmgnt that tb** *flp* retains p binary fite
• A description of me header file
• A description of the data record format
• The maximum value per pixel for each type of
product
• Summaries of the number of pixels in each
class on each product*
• A statement of lineage
• A statement of data quality
The data are in a binary unblocked format A
record equals one row of pixels, and the first data
record value is in the upper left-hand corner of the
image.
(•Note that this is In number of Fuels in contrast to most
maps which show the class summaries In acnes.)
Positional Accuracy and Precision
The positional accuracy and precision of this
data set are based on the Landsat Thematic Mapper
database. Nominally the source data are 30 meter by
30 meter cells with a positional accuracy of 0.5 cell
(15 m) in each direction. Additional uncertainties,
however, reduce the spatial precision of this data set
to about 1.5 cells (45 m) hi each direction. This, in
turn, yields a minfrmim detection unit of 3 cells by 3
cells (90 m by 90 m) or about 2.5 acres.
Attribute Accuracy and Precision
Tests indicate that overall confidence hi the
satellite derived maps is warranted as high as for
aerial photograph/field maps provided the spatial
resolution is 2J> acres or greater.
Tests for logical consistency indicate that all row
and mliimn positions in the selected latitude/longi-
tude window contain data,. Conversion and integration
with vector files falcate &at all positions are consis-
tent with earth coordinates covering the same area.
The classification scheme comprehensively
includes all anticipated land covers, and aH pixels
DataAvauabinty
The data sets are available on three 94rack, 6250
bpi magnetic tapes. The cost of the tapes is as follows:
.$232.00*
.$168.00*
.$104.00*
Any two tapes (please specify):
Single tape (please specify):
Orders must be prepaid by check, money order
On US. dollars drawn on a US. bank, and made
payable to "Dent of Commerce/NOAA/NODCT), or
by credit card (VISA and MasterCard onry). Purchase
orders can be accepted from non-Federal customers
only with prior authorization from the NODC. Orders
should be sent to:
National Ocaanographic Data Center
User Services Branch
NOAA/NESDIS E/OC21
1825 Connecticut Avenue, NW
Washington, DC 20235
Telephone: 202-6064549
Fax: 202-6064586
Omnet NODC.WDCA
Internet: servlcesOnodc2.nodc.noaa.gov
('Prices an lor Fiscal Year 1993; they are In effect until
September 30, 1993. For prices after that date, please
contact the NODC.)
-------
MRLC Consortium
Documentation Notebook
February 1995
11.5 NALC
The following pages provide an introduction to the EPA/USGS NALC project.
The U.S. Geological Survey maintains a repository of additional information on NALC
which can be accessed through the following World Wide Web home page:
• http://sunl.cr.usgs.gov/gUs/hyper/guide/nalc
-------
United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
P.O. Box 93478
Las Vegas NV 89193-3478
May 1993
EPA/600/S-93/0005
North American Landscape
Characterization (NALC)
Research Brief
INTRODUCTION AND OBJECTIVES
The NALC project has
been developed to take
advantage of historical and
current Landsat satellite
remote sensor measure-
ments for evaluation of
global processes. These
efforts involve characteriz-
ing land cover types or
landscape features, and
evaluating their change
using satellite sensors.
Land cover (LC) is the
characteristic elements of
the earth's surface includ-
ing vegetation, soil, topog-
'aphy and human features.
Typically, changes in land
:over occur when agricul-
:ure/pasture is converted
o urban, or forest to
agriculture/pasture. The
esul^^s to type of land
:r ^Bd change in land
x ^^/ill be valuable as
nput to U.S. Global
Change Research Pro-
jram (GCRP) measure-
nent and modeling efforts.
NALC products will have
an important role in evalu-
ations of land processes
and characteristics. Pro-
cesses refers to actions of
the atmosphere, water, *
and soils that are influen-
tial on the earth. These
could include changes in
trace gas fluxes, and
changes in biodiversity.
The goals of the NALC-
Pathfinder project are to
produce standardized data
sets for the majority of the
North American Continent.
The project will develop
standard data analysis
methods to perform inven-
tories of land cover.^utn-
tify land cover change
analyses, and produce
digital data base products
in support of the U.S. and
international global change
research programs.
The NALC project is a
component of the National
Aeronautics and Space
Administration (NASA)
Landsat Pathfinder Pro-
gram. Pathfinder efforts
are focused on evaluation
of global change using
available remote sensor
technologies. The results
and methodologies from
NALC will help address
current problems, and
establish the "path" to
more advanced Earth
Observation System
(EOS) technologies.
, CUIU L«3?«K«-^*W3Eflr>Si5'-*si
Research) Products
coverjproducts.for.priority
-------
LANDSAT MSS TRIPLICATE
Landsat MSS scenes for
part of the State of
Chiapas in Mexico.
Figure 1a is from the
1970's period, and is a
mosaic of two different
scenes (12/05/75 on the
right and 1/17/73 on the
left). This is due to the
shift in image scene or
path/row locations over
time, and is a result from
changes in Landsat
satellite orbital configura-
tions, Figure 1b is from
3/11/86, and Figure 1c is
an image product called
a Reduced Cloud Cover
Composite (RCCC) from
3/03/92 and 4/04/92. In
Figure 1a the arrow
ooints to a coastal man-
grove area and the pink
iolor of healthy, vigor-
usly growing vegetation
this False Color Com-
isite (FCC). In Figures.
b and 1c the pink color
s mostly absent and is
ndicative of a sustained,
decrease in plant vigor
dentified by local scien-
!sts (arrow on Fig. 1c).
ne presence of forest
're bum scars (top arrow)
ind a new reservoir
bottom arrow) are
narked in Figure 1b.
Figure 1c.
-------
GROUND
obal change results from
teration of natural atmo-
spheric, oceanic and
terrestrial processes.
Changes in the quantity
and variety of biotic and
abiotic components of
ecosystems are important
global change indicators.
Understanding change in
natural processes and the
;nfluence of human contri-
Jtions is important to
addressing the impact of
global climatic effects on
ecosystems.
To address these pro-
cesses and supply infor-
mation to decision makers,
a program of measure-
ments and modeling of
landcover conditions and
th- ^Aange will be re-
qu^^To meet these
requirements the U.S.
Environmental Protectibn
Agency has initiated the
NALC project to provide
land cover (LC) determina-
tions and change over
time. Study of land cover
change along with earth
systems processes will
allow causative factors
and feedback effects to be
identified and quantified.
Quantifying meaningful
measures of landscape
characteristics, monitoring
of natural processes, and
evaluating human influ-
ences pose difficult scien-
tific challenges. Global
and regional scale moni-
toring of atmospheric,
terrestrial, and aquatic
processes, and under-
standing the linkages of
these processes, are
required. In particular,
issues of carbon cycling
(inventory or pool, carbon
release, and sequestra-
tion) need to be evaluated
at the regional and global
scale. Changes in land
cover over time are impor-
tant spatial data to assist
in understanding the flux
of atmospheric trace gases
such as methane and
nitrous oxide.
To supply information for
these evaluations, mea-
surements of variables
must be made over large
areas of the earth's sur-
face and at suitable incre-
ments in time. Satellite
remote sensor data are
very appropriate as«4h«y
supply repetitive, consis-
tent, and global measure-
ments for process-related
research and modeling.
The spectral reflectance
characteristics of earth
surface materials can be
used to quantify the spatial
distribution of land cover
(LC). The quantity, vari-
ety, and spatial distribution
of land cover types are
important data inputs for
the inventory and model-
ing of terrestrial carbon
stored in geographic
regions of interest.
The NALC project has a
number of linkages to
Global Change Research
Programs in EPA, as well
as to other Agency domes-
tic efforts and to interna-
tional programs of global
research and inventory.
Examples of collaborative
efforts include contribu-
tions to be made to pro-
grams on Deforestation,
Biomass Burn Monitoring,
Emission Modeling, EPA's
Environmental Monitoring
and Assessment Program
(EMAP), and the
Intergovernmental Panel
on Climate Change
(IPCC).
MANAGEMENT
he NALC global change
^search project at the Las
sgas Environmental
Monitoring Systems
Laboratory (EMSL) is a
component of the Office of
Research and
Development's (ORD)
national program on global
:hange research. The
effort is being conducted
as part of the National
Aeronautics and Space
\dmMration (NASA)
.5 ^^Pathfinder Pro-
jra... of pilot studies. The
joals of these studies are
to evaluate existing satel-
lite data for use in current
and future satellite sensor
programs in support of
U.S. and international
GCRP efforts. The
Landsat Pathfinder Pro-
gram will also develop
some of the methods to
archive, process and
distribute the future high
volume Earth Observation
System (EOS) data. Work
is being conducted by EPA
and other government
scientists, university
cooperators, and contrac-
tor scientists.
In particular, work is being
performed in collaboration
with several groups. The
U.S. Geological Survey
EROS Data Center (EDC)
is providing support in the
areas of data acquisitions,
pilot studies of data pre-
processing techniques,
MSS triplicate data archive
and management, and
ultimately in the production
and dissemination of data
sets. This collaboration
creates great efficiencies
in assembling requisite
technical expertise, and
allows NALC goals to be
achieved with the available
resources of EPA.
The Canada Centre for
Remote Sensing (CCRS)
is also participating in the
NALC project. The CCRS
efforts will initially focus on
the development of meth-
ods for the creation of
large area image mosaics
from NALC MSS Tripli-
cates. Work will initially be
concentrated in both the
Canadian and U.S. por-
tions of the Great Lakes
Watershed.
-------
SCIENTIFIC APPROACH
To conduct change detec-
tion and other analyses
over time and space it is
best to utilize historical and
current data from the same
or similar instrument. The
Landsat Multispectral
Scanner System (MSS)
Sensor has acquired data
from July 1972 through
September 1992. These
data have been archived
in digital form and can be
used for quantitative
analyses. No other exist-
ing sensor system has a
digital archive with a long
term record of acquisitions
over a major portion of the
earth. Hence, these data
have been selected for
use in the initial NALC
retrospective change
detection effort.
Research and develop-
ment activities will focus
?n the data products to be
generated and organized
nto data sets for use in
3CRP activities. The
.pecific research and
levelopment tasks in-
iude: a) acquiring
andsat MSS images with
jss than 30% cloud cover
uring 1992, b) assem-
ling the individual scenes
om 1973,1986, and
991, plus or minus one
Bar, to be used for gener-
:ing coregistered Iripli-
ate" scenes (Figure 1), c)
creating triplicate scenes
georeferenced to a 60 x 60
meter (m) Universal
Transverse Mercator
(UTM) ground coordinate
grid (Figure 2), d) creating
Reduced Cloud Cover
Composites (RCCC's) for
scenes when necessary
(Figure 3), e) generating
derivative products from
the georeferenced image
data, such as land cover
categorizations, f) develop-
ing capabilities to facilitate
archive/management, and
distribution of the image
data and attendant de-
scriptions of the data or -
"meta" database, g) dis-
seminating products to
global change researchers
via EDC, and h) conduct-
ing research on important
issues such as image
categorization and change
detection using the NALC
data sets.
The georegistered image
products will be made
available through coopera-
tive research agreements
with EPA-EMSL-LV, and
at the cost of duplication
from USGS-EDC. The
MSS database products
will be available in whole
scenes corresponding to
the Landsat World Refer-
ence System Two (WRS2)
(Figure 1). Procedir&s
have been developeo to
create high quality
georegistered images in
which systematic cr >
tions for radiometry (vari-
ability in detector re-
sponse) and geometry
(earth rotational skew,
picture element or pixel
oversampling).
NALC images will be
geometrically rectified,
georeferenced, and placed
into a UTM map projection.
Pixels will be resampled
into a 60m x 60m size
format. The 60m x 60m
pixel resolution was se-
lected for compatibility with
the 30m x 30m Landsat
Thematic Mapper (TM)
data resolution. • *r
Some efforts will be de-
voted to developing com-
posites of multiple date
Landsat images of the
same area. This is neces-
sary as some image
scenes will be collected
with cloud cover in excess
of 30%. These Reduced
Cloud Cover Composites
or RCCC images (Figure
3) will be made of cloud
free portions of images
from different dates.
These Reduced Cloud
Cover Composites will
exhibit some scene vari-
ability resulting from
changes in the sun's
position, atmospheric
conditions, vegetation
growth patterns or phenol-
ogy, and other temporal
influences. These sources
of variability may cause
similar materials or land
cover types to exhibit
dissimilar spectral re-
sponses. Portions of this
systematic variability may
be reduced to facilitate
data processing and LC
categorizations.
Coregistered, derivative
products will be developed-
and made available along
with the original data.
These derivative products
would include pixel identity
images to index pixels of
mosaics or Reduced Cloud
Cover Composite images
(RCCC) to the original
input scene (Figure 2).
Additional images would
include multi-spectral
categorization images, and
land cover change images.
-------
ARCHIVING, MANAGEMENT, AND DISSEMINATION
Data will be distributed to
the non-commercial
research community by
EDC at the nominal cost
of reproduction. It is
anticipated that products
will be available on media
or in formats such as nine
track magnetic tape, 8mm
magnetic tape, and/or
3480 tape cartridges.
Later, data may be or-
dered and delivered
through communication
networks. In addition,
characteristics of these
scenes will be incorpo-
rated in USGS's "meta"
database of satellite image
scenes. A UNIX based
information management
system (IMS) and the
Global Land Information
System (GLIS) will be
available to query data
sets that are involved in
the Landsat Pathfinder
Project. This will allow
inventory and archiving of
NALC products, as well as
facilitate browsing of NALC
image scenes, and identifi-
cation and procurement of
suitable products.
ANALYSIS EFFORTS
Efforts are under way to
develop standard proce-
dures for generation and
analysis of NALC prod-
ucts. These standard
meVfjs for Landsat data
ar ^B are of major
importance to the NALC
project. The standardiza-
tion of analytical methods
will provide consistent land
cover and land cover
change products over the
Morth American continent.
These standard ap-
Droaches also address
mportant Agency issues
elated to data Quality
Assurance/Quality Control
QA/QC), such as data
'alidation.
5ilot studies will test and
letermine the standard
and cover (LC) categori-
ation procedures for the
reject. Methods develop-
tent projects will evaluate
pproaches using study
reas in forested, agricul-
jre/Qi£ture, and cloud-
;/^tttL
prone tropical forested
areas. An additional,
important activity is the
formulation and testing of
standard change detection
procedures using NALC
data. These procedures
will focus on generation of
products useful to mea-
surement and modeling of
global change. Results
will also yield a series of
procedures that can relate
anthropogenic or natural
causes to land cover
change.
These efforts will be
accomplished using a
variety of federal collabo-
rators and university
cooperators. Collabora-
tors include EPA, EMSL-
Las Vegas, the USGS
EROS Data Center, and
other federal agencies.
Certain work will require
assistance from outside
the government, and the
contract or Cooperative
Agreement vehicles will
help to obtain additional
capabilities for data pro-
cessing, and research and
development.
Three large pilot studies
will test the standard land
cover categorization and
change detection proce-
dures. One pilot study
focuses on the 64,000
square mile Chesapeake
Bay Watershed. A second
pilot study will evaluate
these procedures in the
State of Chiapas, Mexico.
The third pilot will be
conducted in the 150,000
square mile Great Lakes
Watershed to evaluate
procedures in the north
temperate and boreal
forest ecoregions. These
data analyses and com-
parative evaluations will
lay the ground work for
efforts with NALC prod-
ucts.
Several major program
outputs are envisioned.
By September T993 a
detailed technical plan will.
be finalized and its ele-
ments will be in place. By
September 1993 pilot
study data sets will be
available, and by Septem-
ber 1994 the NALC stan-
dard product data sets will
be initially available for
North America.
International cooperation
will involve universities of
Mexico, Central America,
and the Caribbean. Early
work will focus on South-
eastern Mexico and issues
related to humid tropical
forests. Later efforts will
involve contributions in the
form of research and in the
form of ground data collec-
tion support activities.
-------
FUTURE EFFORTS
The project goal of devel-
oping products from
satellite data in support of
global change research is
a continuing one. A
principal aim is to move to
a prospective evaluation
methodology based on the
use of Landsat Thematic
Mapper (TM) Data. This
will facilitate detailed
spectral and spatial analy-
ses of ecosystems and
detection of-changes in
land cover in a contempo-
rary timeframe.
The use of TM data will
also facilitate the develop-
ment of products that
represent an entire "swath"
of data across the earth's
surface. Such a swath
would run north to south
across the entire continen-
tal land mass, and stretch
east to west 185 kilome-
ters in width. This product
would supply a great deal
of data over a large area
of the earth in a "same
day" timeframe. It is
anticipated that these
large data sets will be
processed in a
Supercomputer environ-
ment. In addition to
change detection efforts,
these data sets would be
useful in the calibration or
verification of results from
numerical models, or in
support of analyses of
AVHRR and Landsat MSS
products, which have less
spatial and spectral
resolution.
Several proposals have
been initiated within the
Agency to characterize
changes in land cover
types in North America.
There is also an initiative
to acquire data for the
coastal Atlantic forests of
eastern Brazil, and to do
so with a format similar to
the NALC program of data
acquisition. These efforts
and others proposed for
Southeast Asia and tropi-
cal Africa will extend this
land cover analysis ap-
proach to additional re-
gions of interest to Global
Change researchers.
CONTACTS
Points of contacts are
Ross Lunetta (702-798-
2175), EPA Environmental
Monitoring Systems
Moratory - Las Vegas,
id James Sturdevant
(605-594-6511), USGS-
EROS Data Center.
Figure 2. A representation
of a standard NALC data
set for a given scene.
Included are the triplicate
scene elements, a pixel
identity image to indicate
the origin of pixel in a
Reduced Cloud Cover
Composite (RCCC),
digital terrain model data
sets, and a spectral clus-
tered scene of land cover
information.
1991 +/-1
Single Date or
Composite
Images
Triplicate Image
Georeferenced
and Coregistered
1986 +/-1
Image
Four Band MSS Images
Pixel Identity Data Sets
1973 +/-1
Image
Coregistered
clipped to
WRS2
Digital Terrain Model Data
1973 +/-1
1986+/-1
1991+/-1
Images
Spectral Cluster Data Set
-------
R^CEDCLOUD^OVER COMPOSITE
Landsat Multispectral
Scanner (MSS) scenes of
part of the State of Chiapas
in Mexico. The area shown
is 185x 185 kilometers.
Figure 3a is a partially
cloud obscured scene from
3/03/92. Figure 3b is a
partially obscured scene
from 4/04/92. Figure 3c
shows the Reduced Cloud
Cover Composite (RCCC)
image made from Figures
3aand3b. —
Figure 3c.
-------
LAND COVER CHANGE
These Figures present
images used in a change
detection sequence.
Figure 4a is a 1970's
period False Color Com-
posite (FCC) of a portion
of the Chiapas images
presented in Figures 1 and
3. Figure 4b is a FCC of
the 1986 scene. Figure 4c
is the product of an opera-
tion where a normalized
difference vegetation index
(NDVI) image-from Figure
4a was subtracted from an
NDVI image of Figure 4b.
In Figure 4c, note the fire
scars (top two arrows) and
the new reservoir (bottom
arrow) identified by the
change detection proce-
dure. These same fea-
tures are identified on
Figure 1b.
Figure 4c.
-igure 4a.
Figure 4b.
-------
MRLC Consortium
Documentation Notebook
January, 1994
11.6 EROS Data Center
This section contains the following EDC documents related to the
MRLC Consortium.
o Development and Application of a Multi-Resolution Land
Characteristics Monitoring System (prepared by the EDC
during the summer of 1993, and provided to participating
programs at the August 1993 MRLC Consortium meeting in
MN. This enclosure included both a section of text and
a series of supporting overhead projection sheets.
The EROS Data Center is currently preparing a document entitled
"Project Plan for a Multiresolution Land Characteristics Monitoring
System". The current draft was completed November 15, 1993, and is
being circulated within the EDC for internal comments. This
document will be provided to the participating programs early in
1994, and will be included in this notebook at that time. Other
EDC documents related to the MRLC Consortium will be included in
this section as they become available.
-------
clFclFelF KRLC consortium
U Docvimentation Notebook
January, 1994
Development and Application
of a
Multi-Resolution Land Characteristics Monitoring System
I. Introduction
The Multi-Resolution Land Characteristics Monitoring System (MRLCMS) is
presently being developed in order to provide capability for broad-based
research on existing and future condition of physical and biological
resources of the United States. The capability sought includes remote
sensing data and other records as well as the conceptual and analytical
tools for manipulating these data.
The MRLCMS is the result of a cooperative effort by six Federal programs
having similar remote sensing and research needs. These programs are:
U.S. Geological Survey:
Land Characterization
National Water-Quality Assessment
U.S. Environmental Protection Agency:
Environmental Monitoring and Assessment Program
North America Landscape Characterization
National Oceanic and Atmospheric Administration:
CoastWatch Change Analysis Program
U.S. Fish and Wildlife Service:
Gap Analysis Program
-------
The development and application of a Multi-Resolution Lan
Documentation Notebook
(MRLC) Monitoring System for Federal Government environraantalry, 1994
assessment programs and the global environmental sciences community is
essential for furthering our understanding of the Earth-system.
Goal
Provide a current baseline of global multi-scale environmental
characteristics data, and mechanisms for identifying, monitoring,
and assessing environmental changes.
Objectives
1. Develop a global 1-km land characteristics data base using AVHRR
data as the primary satellite source.
2. Develop prototype regional land characteristics data bases using
Landsat MSS and TM data as the primary satellite source.
3. Develop a portfolio of methods for quantifying various types of
environmental change including a system for monitoring synoptic
environmental processes and targeting significant areas of change.
In order to meet these goals and objectives there are immediate and long-
term tasks.
Immediate
Jointly acquire and pre-process multi-temporal Landsat
satellite images of the United States. These images will be
-------
MRLC Consortium
Documentation Notebook
January, 1994
pre-processed/archived at the EDC.
Long-term
These programs are exploring additional mutual benefits to be
had by further collaboration, such as:
Development of a seamless multi-resolution land
characteristics database.
Development of systems for monitoring environmental
processes and identifying areas of significant change.
Developing methods for quantifying various types of
environmental change.
n. Project Participants
The following programs/projects are an integral part of the MRLCMS and
form the foundation for the multi-agency partnership.
A. Land Characterization Project
Mission — The goal of this U.S. Geological Survey (USGS) global
change research project is to develop a multi-resolution global land
cover characteristics monitoring system. Global climate change
research requires large-area (i.e., continental, global) spatial
-------
information describing the distribution and
. . _. ... . _ Documentation Notebook
surface phenomena. The specific information
science community vary considerably, but are all based on
understanding fundamental land characteristics. Land cover,
particularly when defined based on both seasonal and compositional
criteria, provides a starting point for describing land parameters.
The USGS recently completed a prototype land characteristics data
base at 1-km resolution for the Conterminous U.S. Based on the
acceptance of the U.S. prototype by the land process modeling
community, this effort is being expanded to the globe. The three
components of the global activity include: (1) development of six
continental land cover characteristics data bases; (2) validation of
the land cover descriptions based on best available source materials
(such as Landsat imagery) and expert opinions; and (2) research on
critical land charactftriTat^on development issues and applications.
There is a strong need now to begin establishing the framework for
1) including higher resolution data in the land characteristics data
base, and 2) developing approaches to monitor change.
Deliverables — A capability to monitor the land surface, target
changes, and assess those changes is the ultimate goal. Development
a multi-resolution data base, consisting primarily of 1-km AVHRR
data and its derivatives and Landsat data and its derivatives, is the
initial focus. So for the Landsat component of this project, the data
requirements are source Landsat data, georegistered and terrain
corrected Landsat data, and clustered data sets. EDC may assist
user in the generation of specific land cover products. User-
generated land cover products may come back to EDC for archiving
and distribution. All data sets will be cataloged in EDC's information
management system. All data without use restrictions will be
-------
MRLC Consortium
Documentation Notebook
January, 1994
available to the public.
B. National Water-Quality Assessment (NAWQA) Program
Mission — The NAWQA Program of the USGS is designed to describe
the status and trends in the quality of the Nation's ground-water.
The Program integrates information about water quality at a wide
range of spatial scales, from local to national, and focuses on water-
quality conditions that affect large areas of the Nation or occur
frequently within small areas.
Study-Unit Investigators involved in NAQWA will conduct research in
60 major hydrologic basins (Study Units) of the Nation. These
NAWQA Study Units collectively cover a large part of the United
States, encompass the majority of National water use, and include
diverse hydrologic systems that differ widely in both the natural and
human factors that affect water quality facilitating comparative
analysis of significant national water quality issues. Each Study-
Unit Investigation will be conducted in two phases. The first phase
will consist of intensive assessment activity for 4 to 5 years. This
will be followed by 5 to 6 years of low-level assessment activity.
Intensive water-quality assessment within each Study Unit will thus
be conducted on a rotational rather than a continuous basis.
Approximately one-third of the Study Units will be studied
intensively at a given time and the decadal cycle will be repeated
perennially. The first complete cycle of intensive investigations of
all 60 Study Units will be completed in 2002.
r
-------
Deliverables — NAWQA needs four levels of land cover fcifprmattan —
Documentation Notebook
for January, 1994
1) National Synthesis (Pesticides and Nutrients, and in 1994, Volitile
Organic Compounds), where they need very small scale land cover
data. Agricultural pattern distribution by county is the kind of data
needed. The 1-km land characteristics data base will be investigated
for its usefulness.
2) Study Units, where they are currently using LUDA GIRAS data.
There will be 60 active study units.
3 and 4) Within a study unit, transects may be established to assess
what's on the ground. Land cover data will be collected in the field.
Landsat TM-based land cover data will be used in the study unit
analyses, in place of the LUDA data. NAWQA wants clustered data
that they will then use, along with ancillary data, to interpret land
cover.
C. Environmental Monitoring and Assessment Program (EMAP)
Mission —• EMAP, managed by EPA's Office of Research and
Development (ORD) is an innovative research, monitoring, and
assessment effort designed to report on the condition of our Nation's
ecosystems. EMAP is assessing the condition of ecological resources,
e.g., wetlands, surface water, the Great Lakes, agroecosystems,
arid ecosystems, forests, and estuaries. The program is currently
in the pilot phase and is recognized as a critical component of the
-------
MRLC Consortium
Documentation Notebook
January, 1994
overall U.S. Global Change Research Program. When implemented,
EMAP will provide high-quality data on the condition of our Nation's
ecological resources, through the development of monitoring tools to
support our Nation's resources. EMAP will generate new ecological
monitoring and assessment information, which will be combined with
data from other monitoring programs to provide a comprehensive
view of the effectiveness of national and international environmental
and global change policies.
n«.nwraMoc — On a hexagon by hexagon basis for the entire
Conterminous U.S., EMAP scientists are collecting a myriad of
environmental data in the field. Large-area land cover and ancillary
data are needed to provide the framework for the consistent analysis
of the field data. Land cover data must be of known accuracy and
confidence. The various EMAP resource groups have special land
cover interests, so land cover of significant detail is needed. The
land cover classification scheme that GAP has developed may be
suitable for EMAP.
E-MAP land characterization, which is developing the land cover
information, is a high priority activity within the agency. It is likely
to have solid funding for not only land cover product generation, but
also associated research. There is strong interest in EPA for the
"land characterization approach" to land cover generation (also
referred to as the "information-based approach"). This means the
development of a data base of land characteristics from which land
7
-------
cover information may be derived and tailored to specific needs. .
MKLC Consortium
Thus, the requirement is for a clustered
Conterminous U.S., together with source and complex Landsat data
sets and ancillary data.
D. North America Landscape Characterization (NALC) Project
Mission — The U.S. Environmental Protection (EPA) and the U.S.
Geological Survey, through their respective global change research
programs, are conducting a collaborative project that involves
development of multi-temporal Landsat multispectral scanner (MSS)
database of North America spanning the years 1973-1992. The EPA
and USGS are developing geo-referenced MSS triplets corresponding
to 1973, 1986, and 1992 epochs. This effort is designed to provide
the data needed to assess types and rates of landscape change, and
provide detailed land cover data that contributes to EPA carbon cycle
research. This program is considered to be a centerpiece of the
Landsat Pathfinder program headed by the National Aeronautics and
Space Administration (NASA).
E. CoastWatch Change Analysis Program (C-CAP)
Mission — In order to better understand and manage living marine
resources, scientists and managers need up-to-date information on
the distribution and abundance of coastal fisheries habitats and how
these habitats change with time. In accordance with these needs the
Coastal Ocean Program of the National Oceanic and Atmospheric
Administration (NOAA) has initiated the CoastWatch Change Analysis
If
-------
MRLC Consortium
Documentation Notebook
January, 1994
Program (C-CAP). The purpose of C-CAP is to develop a
comprehensive, nationally standardized information system to assess
changes in land cover and habitat in coastal regions of the United
States. C-CAP utilizes both satellite and aircraft based sensors to
map emergent wetlands and surrounding uplands as well as
submerged aquatic vegetation. The goal of the program is to monitor
coastal areas every 1 to 5 years, with time dependent upon the rates
and magnitude of change in each region. Protocols for developing C-
CAP data were drafted through a series of workshops that brought
together approximately 250 technical and regional experts and
representatives of key state and federal organizations. To date a
prototype project using TM data to detect change in land cover in the
Chesapeake Bay drainage area and several projects using aerial
photography to detect change in sea grasses in North Carolina have
been completed. Additional regional projects are underway in South
Carolina and Rhode Island. Plans are now being made to expand
coverage to Florida, the Great Lakes Region and the West Coast.
Deliverables — Watersheds have a big impact on marine resources. A
requirement is to use satellite data to assess land cover change
watersheds in a reasonable time period in order to mitigate impacts.
The satellite image processing requirements for C-CAP have been
documented by a host of NOAA staff, and with the help of Jerry
Dobson of Oak Ridge. John Jenson reviewed the document. It
contains many of the details on C-CAP's approach to Landsat
preprocessing, classification, and change detection. Much research
-------
is still needed in these areas. C-CAP has been fundina others to
conduct the research and this is expected 1
Oak Ridge is expected to continue performing much of the satellite
data processing for C-CAP. They will require geometrically and
terrain corrected Landsat TM data sets from EDC. The ultimate C-
CAP deliverable is a data base of land cover and change information,
archived and distributed by the National Oceanic Data Center in
Washington D.C. The product is a CD-ROM of land cover
classifications for two time periods and land cover losses and gains.
The study area units are Esturine Drainage Areas. National Marine
Fisheries Service is the client. They use the data in models
addressing how marine life and water quality changes per changes in
watershed land cover. ;
F. Gap Analysis Program (GAP)
Mission — GAP was begun in 1988 with research funds from the
National Fish and Wildlife Foundation and the Idaho Department of
Fish and Game. Since then the program has continued and expanded
each year with funds added on by Congress. Although the Service
has maintained the goal of completing GAP in the 48 conterminous
states by 1998, the funds to do so have not, as yet, been sufficient
to reach that goal. The status of GAP nationally is summarized
below:
Research has been initiated in 22 states, which represents 45%
of the lower 48 states' land area.
Initial data-gathering has been completed for three states, five
additional states are expected to complete data-
-------
MRLC Consortium
Documentation Notebook
January, 1994
gathering in 1993.
The institutional and partnerships necessary for continued
development of GAP data and the implementation of GAP
findings on a nationwide basis are established.
• A standardized vegetation HaaaiffnaMfMTi system has been
developed and will be applied across the nation; mapping
protocols have been developed so that the data collected can be
interpreted in a consistent way.
GAP Analysis provides a quick overview of the distribution and
conservation status of several components of biodiversity. It seeks
to identify gaps (i.e. , vegetation types and species that are not
represented in the network of biodiversity management areas) that
may be filled through establishment of new reserves or changes in
land management practices. Gap Analysis uses the distribution of
actual vegetation types (mapped from satellite imagery) and
vertebrate and butterfly species (plus other taxa, if data are
available) as indicators of, or surrogates for, biodiversity. Digital
map overlays in a CIS are used to identify individual species,
species-rich areas, and vegetation types that are unrepresented or
under-represented in existing biodiversity management areas. Not a
substitute for a detailed biological inventory, Gap Analysis organizes
existing survey information to identify areas of high biodiversity
before they are further degraded. It functions as a preliminary step
to the more detailed studies needed to establish actual boundaries for
-------
potential biodiversity management areas. We hypothesize that Gap
MRLC Consortium
Analysis, by focusing on higher levels of
be both cheaper and more likely to succeed than conservation
programs focused on single species or populations.
Deliverables — A standardized methodology for GAP land cover
generation is in development. On-screen digitizing is being tested
and appears to provide more content— rich foiforrnatiop than traditional
automated ^|agg
-------
MRLC Consortium
Documentation Notebook
January, 1994
B. Geo-registxation and terrain correction
C. Clustering
D. etc.
-------
MRLC Consortium
Documentation Notebook
January, 1994
EDC's Fiscal Year 1994 Tasks
I. Ordering and Purchasing
Background
Following the commercialization of the Landsat satellite system by the Land
Remote Sensing Commercialization Act of 1984 (Public Law 98-365), the
U.S. Geological Survey, EDC established a purchasing agreement to assist
Federal agencies in obtaining Landsat products and services from EOSAT.
In 1987, a similar agreement was established with the SPOT Image
Corporation, the exclusive distributor of SPOT satellite data in the United
States. These purchasing agreements eliminated the requirement for each
Federal agency to establish, administer, and maintain separate agreements
for the purchase of civil satellite data. These agreements also provided a
means for each agency to continue to issue funds to another Federal
agency, i.e., government to government money transfer, and provided a
mechanism for each Federal agency to issue a single purchase order for
obtaining both Landsat and SPOT satellite products and services from a
single centralized source. They do not prohibit any Federal agency from
using them, and the use of the agreements is optional. Since 1985, thirty
different Federal agencies have purchased over $22.5 million worth of data
through these agreements. To date, this single centralized source for data
procurement has saved the U.S. Government over $1,200,000.
-------
Given the history and success of these agreements, EDC
Documentation Notebook
collection of multi-agency funds and will procure the TM dat§^tfiEJijghilp^
current USGS/EOSAT Basic Ordering Agreement. It is anticipated that the
current contract would be modified to reflect the agreed-to price schedule.
A complete order, by scene identification, will be prepared by EDC.
Project representatives plan to visit EOSAT to preview each selected scene
to ensure proper cloud cover, therefore, significantly reducing reorders.
A. Status
Moving to discussions on scene selection, Paul Severson went to
EOSAT to screen 530 scenes. About 70 percent of them were OK, but
the rest had problems. Paul selected alternative scenes for the
problem areas and sent their specifications to the cooperators for
review. But in the meantime, EOSAT has now agreed that 1992 and
1993 data are available for our use. Should we go back to square one
and reselect in order to get 1993 data (if available)? Mike keeps
saying that multi-temporal data is critical for certain areas. Mike
needs to go back to the original needs for data, and see if 1993 data
are available. Tom says that we are about 30 days away from legally
ordering the data. Paul will meet with Gail, Don, and Mike tonight to
pin down the strategy to follow next.
Tom explained the current EOSAT proposal. A few facts: 610 scenes
needed, 80 were already available from GAP, 530 needed to
purchase. $1.325 million was the deal based on 530 at $2500. The
first proposal was $3000 for the first 500 and $3500 for every scene
after that resulting in $1.605 million for 530. Now the proposal is
$2500 per scene for the first 430, $3500 for every scene thereafter
for a total of $1.425 million. The window is open for ordering at
these prices in the future. But a key point is that we want to
IV
-------
MRLC Consortium
Documentation Notebook
January, 1994
establish a land cover baseline for as tight a period of time as
possible, because we'll want to establish another baseline in 5 to 10
years.
The feeling is that the longer we wait, the better our chances of
never paying $3500 per scene. This is because negotiations with
EOSAT for Lapd«at 6 may likely result in an across the board price of
$2500 or lower. This will be known by September 30. An approach
would be to buy 430 scenes right now, and wait with the rest of the
order. A word of caution — the deal right now from EOSAT involves
free use of the data among our cooperators. We don't know if the
Landsat 6 negotiations will result in free use of data across
government agencies.
The consortium of programs has agree to EOSAT latest offer.
However, we will agree to procure 430 scenes at $2500 per scene. We
will wait to procure the remaining scenes until at least early in the
next fiscal year. Tom will talk to EOSAT tomorrow.
Satellite Data Pre-processing and Processing
HZ. Data Archiving and Distribution
The National Satellite Land Remote Sensing Data Archive, as established by
the Land Remote Sensing Policy Act of 1992, Public 102-555, which EDC has
It
-------
been directed by the Department of the Interior to inana9e(iRyc^c^^§^^f •
data management and distribution services to the princTp,
their affiliates.
ium
•11
-------
MRLC Consortium
Documentation Notebook
January, 1994
o
Q
cc
e
o
1
cP"
o
0
Q.
O
o
1
i
J
2
5
.
H
> >»
> CO
J
jBO
S
C
a
VI
a
-------
MRLC Consortium
Documentation Notebook
January, 1994
C
5?
** c
"Q.C
>ment and Ap
lulti-Resolutio
CO
OJ
c
• ^^IM
istics Monitor!
haracteri
0
• J J
/)
UJ
OALS AND OBJECTS
O
j
3
O
•o
CO
w"
y
9
*SI
O
CO
CO
o
2
e
o
1
">
global multi-scale en
nt baseline of
i
o
CO
9
!
£
(0
o
01
e
CO
o
s
0)
1
e
o
O)
e
targeting, and assess
r monitoring, 1
^^
g
«
CD
1
^v
*.
«A
V)
UJ
o
UJ
2
o
•
j|%
0)
0)
CO
m
id characteristics dati
£
S
•
CD
a
o
S
&
r
0)
(0
CO
.a
CO
To
•o
—•~-
aj land characteristics
1
i
j
E
«
to
^M
O
i
|
s
*»
lultl-resolutloi
S
o
(0
V
a
(0
a
1
,1
f
i
j
?
t
•o
^
o
i
CD
CO
s
2
1
CD
0)
CO
o
i
s
E
a a 5- 2
^
S
oi
o
9
CO
i
CO
0)
•o
e
CO
(0
V
(0
V)
o
o
Q.
"c
0
c
o
">
i
o
I
O ai
^ C
CD
O) £
C O
1 •
be used for m
significant ar
= §*
Si
s
^n
1 f
a
(A
O
Ifl
-------
MRLC Consortium
Documentation Notebook
January, 1994
c
o
o £
£ C/>
3 O)
LU
i
-------
MRLC Consortium
Documentation Notebook
January, 1994
V
Qi
O
h.
I
To
v
0)
CO
CO
.0
CO
I
CO
^o
I
5
o
§
CO
o
1
o
CO
D
CO
o
a v1
3 M
- a
S ol
2
o
O
Develop
2
Q.
O
o
O)
e
"5.
a
a Q
2 £
S S
e a
o uj
* Z
3
cu
a
(0
o
ffl
r
CO
UJ
Q)
Def
o
1 I
o o
0 S
O e
1 i •
IM
ill
= a z
G)
C
V
O
to
». CO
0) O
c S
cu o
O u.
To E
u «
O) S
O^ CO
CO
o
S2
2«
*5
s
ro
SI
ii
ii
z z
I
Q)
§
I
•o
i
1
Q
UJ
of
l
!
a
o
8
O)
CO
f
CO
CO
CD
Z ffi
CO
1
I
i.
$fl)
jj
^%
«• n
ii
-------
MRLC Consortium
Documentation Notebook
January, 1994
11
<»
&.S£
0)
eQC
•• I 71
i£--
E 3 w
(0
O
o
CO
CO
CO
o
s
i
1
CO
o
CD
«
CO
3
O
•o
as
09
JO
£
Q>
O)
CD
CO
O
CO g
(0 £
A C
•s I
o £
I g
2*2
*8
IA
O
in
3
•o
o
o
8
O
CO
CO
CD
U)
i
o
1|
si
ii
ent, quanttta
8
!
I
CD .
€ |
£ I
O)
£
"CD
ii
co 2
acti
for
s.
CD
CO
•o
CD
11
O
3
«
"5.
£
2 > >»
a« - =
« S
-------
MRLC Consortium
Documentation Notebook
January, 1994
c
Q.C.E
0.0 =
-------
KRLC Consortium
Documentation Notebook
January, 1994
c 2
£-*oJ
Q.C el
•o-x
3.£2
So
o
(0
CO
a
o
o
UJ
JS
CO
.o
I
o
•o
e
(0
0)
V
O)
CO
CO
CO
•o
o
•o
1
a
•o
a
o —
„ I
o 5
W O)
o «
OT ""
CO O
Q o
I!
II
S c
« i
8-1
"O co
c «
(A
o
S
i i
i
o
*
3
N
0
a
o
CD
O)
CO
O)
8tfM JS?
8 «
UJ
3
CO
O
CO
a
o
ac
0 =0
E <2 co
III
* c A
O) — «
£ 5 «
S.5 '
* s^
< 2 D
-------
MRLC Consortium
Documentation Notebook
January, 1994
o £
CO 0)
£ CO
3 0)
5.§
^ 3 i
o'E I
c o
.22
75 g
o .s
^..2
CO CO
50
o
0
Q
•o
c
CO
*
O CO
_ CO
co Q
§ w
5 o
0) CO
cl
a*2
o o
Is
> CO
-------
ac c
Q. O J-
U
e o
co c
"° E
•5 ^
C (0
5» §
(0 2
S; o)
•e §
i ^
g =
O (0
in
O
(A
i i ^
su
g-
o e
o e
o ^
S 2
C0 O)
e §
Si
«i
8."
e a
a e
£ (B
o 8
% v
o «
I8
S "°
i §
§ °>
S £
CD Q}
a 75
i!
•a «
« E
ii
o o
51"
55 a
-------
•2?
73 «c/>
^-3 01
S-g-i
Q. O ^
ill
c°F co
oi o
£30)
0.5 -g
.o®
U w o
SB2
Q (0
6
0>
>»
(0
O)
o>
O)
S
•o
CO
0)
o
a
o
•o
I
0)
CO
9 .
'S
o
0)
•a
I
o
V
I
I
•o
o
i
(0
i§
1 i
51
ii
A O
o ob|ectlv
two-tiered
o
~ To
CC o
DC •=
I £
52-d
< £ o
S o «
0*0)
^j ^ (5
* * «
J; O
Q>
So
CO
o E 5
I 8 2
> w HI
c 5 e
O 5
-------
.2-0
ff e
as£
i§I
III
g 3 CD
Q.^ -g
5^o
§ * o
^o2
Q (0
O
3
(A
CO
LU
CO
UJ
O
UJ
21
O
0
5
2
0
i
O
8
O
CO
S
CO
CO
CO
and acc
£
CO
I
O)
i
(0
T3
* I
"Q >
CO
e
o
nge det
O)
0)
1
CO
a
CO
CO
-------
MRLC Consortium
Documentation Notebook
April 1994
SECTION 12
MRLC CONSORTIUM MEETINGS
This section contains meeting agenda and notes, when available, for Consortium meetings. To
date there have been 5 meetings of the MRLC Consortium, including:
1) Portland, OR - 4/93
2) Las Vegas, NV - 5/93
3) Sioux Falls, SD - 6/93
4) Minneapolis, MN - 8/93
5) Mountain View, CA - 11/93
6) Santa Barbara, CA - 2/94
-------
-------
MRLC Consortium
Documentation Notebook
January, 1994
12.1 Portland, Oregon
The initial meeting of the MRLC Consortium was held near
Portland, Oregon, on March 31 and April l, 1993. The attached
minutes of this meeting were prepared by the EROS Data Center.
-------
MRLC ConsortitM
Documentation Notebook
January, 1994
Minutes
PORTLAND MEETING
March 31 and April 1, 1993
Purpose
To identify areas of potential cooperation among the U.S. Geological Survey
(USGS) EROS Data Center (EDC) and four programs:
1. Environmental Protection Agency (EPA) Environmental Monitoring and
Assessment Program (EMAP)
2. U.S. Fish and Wildlife Service (USFWS) GAP Analysis Program
3. USGS National Water Quality Assessment (NAWQA) Program
4. National Oceanic and Atmospheric Administration (NOAA) Coastwatch - Change
Analysis Program (C-CAP)
Emphasis is placed on assessing the requirements and coordinating acquisition of
Landsat Thematic Mapper (TM) imagery of the Conterminous U.S.
Attendees
EMAP — Dan McKenzie, Denise Shaw, Dorsey Worthy
GAP — Mike Jennings
NAWQA - Gail Thelin
C-CAP - Don Field
University of Ohio — John Lyon
EDC — Tom Holm and Jim Sturdevant
Day 1
Briefings were presented on each of the four Programs and EDC. General areas of
common interest were identified. Requirements for Landsat TM data were
identified. Discussions on data acquisition possibilities and negotiation
strategies with EOSAT were initiated.
Common Activities
Landsat TM requirements
Landsat image preprocessing (geo-registrati on and terrain correction)
Spectral clustering
Ancillary data
Accuracy assessment
Data management
Research and technique development
-------
MRLC Consortium
Documentation Notebook
January, 1994
Unique Strengths of Each Program
EMAP - Ancillary data, data and system interface development software
GAP - Land cover classification, land ownership data, biological data,
state and university connections
NAWQA - project-level data management, multiple-scale analysis,
urban and agriculture
C-CAP - accuracy assessment, change detection, coastal areas
Day 2
Objectives of the Day
1. Define the common goal of the national programs, to be used in negotiation
with EOSAT.
2. Arrive at a common understanding of our negotiating position, the
anticipated government and EOSAT negotiation process, and current contract
limitations.
3. Identify the Landsat pre-processing requirements common among the
programs.
4. Explore opportunities for cooperation on research and technique
development.
5. Discuss impacts of these requirements on EDC. Explore opportunities for
visiting scientists, interns, Post Docs, etc.
6. List and define the actions items.
Objective 1
Four Federal programs are cooperatively developing a national-level land cover
data base.
A major requirement is a Landsat TM data set for the Conterminous U.S. with the
following specifications:
Option A.
— Two complete Conterminous U.S. coverages for 1992, plus or minus one year.
-- Dates of acquisition:
Coverage 1. — At or near the peak of the growing season.
Coverage 2. — Leaf-off with no snow cover.
-- 10 percent or less cloud cover.
Note: Please consider if "peak growing season is optimal for separating broad
vegetation classes (pasture from row crops from forest).
-------
MRLC Consortium
Documentation Notebook
January, 1994
Option B.
One complete Conterminous U.S. coverage for 1992, plus or minus one year,
acquired at peak growing season, and an additional multi-date coverage of
selected areas (i.e.: acquired in 1992 plus or minus one year, but at
another time of year).
Data will be delivered to and archived at the National Satellite Land Remote
Sensing Data Archive at the EROS Data Center. The Archive will distribute data
to the four programs at the cost of reproduction and distribution.
Objective 2
An over-arching issue is whether the current NASA Data Grant can be modified to
satisfy the TM data requirements of the Programs. This approach would be fast
and most easily implemented. The following negotiating scenario was suggested:
Going in position -- The data will be available with no use restrictions
to ail researchers, same as that in the NASA Data Grant.
First fall back position — The data will be available with no use
restrictions to the four agencies (USGS, FWS, EPA, and NOAA) and their
cooperators.
Second fall back position — The data will be available with no use
restrictions to researchers within the four agencies (above).
Third fall back position ~ The data will be available with no use
restrictions to the GAP, EMAP, NAWQA, C-CAP, and the USGS National Mapping
Division Research and Technology Program.
"No use restrictions" is more important than low data cost.
Objective 3
The four programs have common requirements for preprocessing:
Geo-registrati on of all scenes. (One-half pixel accuracy is required.)
Terrain correction for about one third of the scenes.
Maybe common centralized clustering.
EDC was asked to cost out Landsat TM geo-registrati on and terrain correction.
Clustering will be addressed at a subsequent meeting, tentatively scheduled for
late April.
Objectives 4 and 5
Each Program will study options in these areas over the next several weeks and
discuss them at the next meeting.
-------
MKLC Consortium
Documentation Notebook
January, 1994
Objective 6
1. EDC prepare and send plots and associated metadata in digital form (ASCII
format), and a WRS map, to the Programs -- ASAP. Contacts are Dorsey,
Gayle, Don, and Mike.
2. EDC send meeting minutes to the Programs -- ASAP. Each program drafts a
paragraph on their programs and sends to EDC (Jim) — ASAP. These will be
used as the basis for a multi-agency agreement. The agreement may be
useful in negotiations with EOSAT (although it Is unlikely that a multi-
agency agreement can be approved before EOSAT negotiations occur).
3. EDC to brief Center Management and NMD Headquarters — ASAP.
(ISSUE: Whether to bring other agencies into this process.)
4. EDC to informally discuss negotiation options with the NASA and DOD
Landsat Program Office -- shortly after NMD management meetings.
5. Programs to identify the approximate minimum number of scenes to purchase
— shortly after they receive plots from EDC.
6. EDC to assess the cost and feasibility of preprocessing the Landsat TM
imagery of the Conterminous U.S. — by the time of the next meeting.
7. All assess research and technique development requirements -- by the time
of the next meeting..
8. Negotiate with EOSAT — preferably within the next two months.
Final Comments
Just before meeting adjournment, there was discussion about the wisdom of
acquiring two sets of Landsat TM images for the Conterminous U.S. A concern is
that the cost of two sets may leave little funding for the remaining activities
(preprocessing, clustering, accuracy assessment, etc.). Is a second set needed
if funding would not be available to process it? A counter point is that the set
would not need to be processed immediately. If two sets can be acquired for the
price of one, why not go for it? The two sets would be extremely available for
these four programs and many others. Please send additional comments or ideas
on this to Jim.
It was decided that this group should meet periodically for the next year or so.
The next meeting possibly will be the last week in April at the USFWS Regional
Office near the Denver Federal Center on Simms Avenue. Stay at the Doubletree
Hotel. This plan and progress on the action items will be reviewed via telecon
of meeting attendees on April 19. Denise will initiate the telecon.
Special Assignment
Don't forget!! GAP, EMAP, NAWQA, and C-CAP — Provide a paragraph describing
your programs and electronically transmit it to Jim.
(Sturdevant@edcserverl.cr.usgs.gov)
-------
MRLC Consortium
Documentation Notebook
January, 1994
12.2 Las Vegas, NV
The second meeting of the MRLC Consortium was held at the U.S.
Environmental Protection Agency Environmental Monitoring Systems
Laboratory in Las Vegas, Nevada, on April 28-29, 1993. A formal
agenda and set of meeting notes are not available from this
meeting. The following notes on this meeting were compiled from
materials provided by Thomas Holm (EROS Data Center) and Gail
Thelin (NAWQA).
Meeting Dates: April 28-29, 1993
Meeting Location: Environmental Protection Agency Environmental
Monitoring Systems Laboratory, Las Vegas,
Nevada
Meeting Attendees: Don Field, C-CAP
Thomas Holm, EDC
Mike Jennings, GAP
Denice Shaw, EMAP
Gail Thelin, NAWQA
Dorsey Worthy, EMAP/NALC
Primary meeting topics:
1. TM purchase ageement with EOSAT
2. scene selection: criteria and priority
3. scene pre-processing requirements
Purchase Agreement with EOSAT
o EDC will review EOSAT pricing and price reduction
programs for large volume scene purchases
o EDC will take lead in developing mechanism for purchase
of data for MRLC cooperating agencies, to alllow for use
and distribution of the purchased data by these agencies
o The decision was reached at this meeting for EDC to
initiate formal negotiations with EOSAT for the purchase
of the Landsat Thematic Mapper scenes. Tom Holm will
meet with EOSAT representatives on May 10 in Lanham, MD
to begin discussions/negotiations.
Scene Selection
o C-CAP will need approximately 120 TM scenes
o GAP has 52-75 scenes that will fit the 1992 +/- 1 year
time window
o NAWQA and GAP have the most stringent scene selection
criteria based on landcover identification requirements;
C-CAP has very specific criteria prescribed by remote
sensing leads
NAWQA and GAP will take lead in scene
identification for primary scenes; priority will be
given to C-CAP requirements for the secondary
scenes in coastal areas
o EMAP priority states: PA VA WV DE NY MN WA MI IL IN
-------
MRLC Consortium
Documentation Notebook
January, 1994
GAP priority states: AR OR OK NJ WY CO WV DE
— NAWQA and C-CAP agreed with this scheme for
regional prioritization
o Prior to June 7, the participating programs agreed to:
1) define scene selection criteria
2) develop decision rules for meeting criteria
3) list scenes that meet criteria
4) prepare final combined list
5) preview data and prepare final list for ordering
o EDO prepared and provided an initial list of scenes
(1/91-5/92; cloud quality « 0/1/2); will generate new
list for 1991-1993, cloud quality - 0/1
Pre-Processing Requirements
o The following pre-processing steps were discussed and
agreed to by the participating programs:
1) debanding
2) atmospheric and sun angle correction
3) geo-rectification
4) terrain correction (where needed)
o EDO provided a preliminary cost estimate of $600 per
scene to do the pre-processing, substantially less
expensive than that offered by EOSAT.
o GAP and EMAP agreed to contribute funds in FY 1993 and
1994 to pay for pre-processing.
-------
MRLC Consortium
Documentation Notebook
January, 1994
12.3 Sioux Falls, SD
The third meeting of the MRLC Consortium was held at the EROS
Data Center in Sioux Falls, South Dakota, on June 24-25, 1993.
Included in this section is the agenda for this meeting. Initial
TM scene selections were completed during this meeting. No
additional notes are currently available from this meeting.
-------
MRLC Consortium
Documentation Notebook
January, 1994
EROS DATA CENTER
AGENDA
Visit By; Gail The!in, National Water Quality Assessment (NAWQA) Program,
Water Resources Division, USGS, Sacremento, California
Michael Jennings, GAP Analysis Program, Fish and Wildlife Service,
Moscow, Idaho
Denise Shaw, Environmental Monitoring and Assessment Program
(EMAP), Environmental Protection Agency (EPA), Research Triangle
Park, North Carolina.
L. Dorsey Worthy, North American Landscape Characterization (NALC)
Program and EMAP, EPA, Las Vegas, Nevada
Donald Field, Coastwatch-Change Analysis Program (C-CAP), National
Oceanic and Atmospheric Administration, Beaufort, North Carolina
Chuck Dull, Remote Sensing Coordinator, U.S. Forest Service,
Washington, D.C.
Purpose: To identify Landsat TM scenes for the multi-agency Conterminous
U.S. procurement, and to continue discussions on the interagency
project, Development and Application of a Multi-Resolution Land
Characteristics Monitoring System.
Thursday. June 24. 1993
(Airport Holiday Inn, 2nd Floor Conference Room)
8:30 - 10:00 a.m.
10:00 - 10:30 a.m.
10:30 - 11:30 a.m.
11:30 - 1:30 p.m.
Role of Each Program
Status of EOSAT Negotiations
Scene Identification Status/Issues
Lunch in Sioux Falls
Participants
T. Holm
Participants
1:30 - 5:00 p.m.
(Executive Conference Room, EDC)
Scene Identification
Participants
-------
MRLC Consortium
Documentation Notebook
January, 1994
8:30 - 12:00 noon
12:00 - 1:00 p.m.
1:00 - 3:00 p.m.
3:00 p.m.
Friday. June 25. 1993
(Executive Conference Room)
Scene Identification (cont.)
Lunch
Preprocessing, Clustering, Research &
Development, and Data Management
Adjourn
Participants
T. Love]and &
Participants
Distribution:
Senior staff
B. Bailey
T. Holm
D. Binnie
D. Scholz
T. Loveland
D. Greenlee
J. Eidenshink
K. Klenk/G. Johnson
C. Randall (2)
-------
MRLC Consortium
Documentation Notebook
January, 1994
12.4 Minneapolis, MN
The fourth meeting of the MRLC Consortium was held in
Minneapolis, Minnesota, on August 15-18, 1993. A copy of the
agenda for this meeting is included in this section. A set of
meeting notes is currently being compiled from multiple sources.
Upon its completion, it will be included in a future notebook
update.
-------
MRLC Consortium
Documentation Notebook
January, 1994
AGENDA
Satellite Scene Purchase Meeting
August 15-18. 1993
Minneapolis, NN
Sunday. August 15. 1993
1:00 - 5:00 p.m. Introduction of Team Members
Overview of Project Goals and Objectives
Overview of Related/Relative Projects
• Global 1-km Project EDC
• GAP
• EMAP
• NA8W-
Scene Selection Status
• Summary of Paul Severson's Trip to EOSAT
• Status of EOSAT Data Partnership Proposal
Monday. August 16. 1993
8:00 - 5:00 p.m. Preparation of White Paper
• Define Purpose of Paper
• Develop Outline
• Draft Report by Agency
Tuesday. August 17. 1993
8:00 - 12:00 noon Review White Paper Draft
1:00 - 5:00 p.m. Detailed Discussion on Development of Regional Land
Characteristics Data Base
Data Preprocessing Requirements
• Noise Removal
• Debanding
• Detector Striping
• Bit Drops
Radiometric Corrections
• Haze Removal
• Sun Angle Corrections
-------
Geometric Corrections
• Ground Control Point Source
• Accuracy Requirements
• Map Projection
• Resampling Techniques
• Terrain Corrections
Spectral Clustering
• On-Going Research Activities (SPECTRUM, etc.)
Image Mosaicking Requirements for Seamless Data Sets
• Resolution of Data to be Mosaicked
• CAGIS Software Upgrade
Wednesday. August 18. 1993
8:00 - 12:00 noon Ancillary Data Integration
• Discussion of Ancillary Data Included
Data Archive Issues
Data Base Management System
1:00 - 5:00 p.m. Project Summarization
• Cooperator's Final Product Requirements
• EDC Processing Flows and Data Management System
-------
12.5 Mountain View, CA
MRLC Consortium
Documentation Notebook
January, 1994
The fifth meeting of the MRLC Consortium was held at the NASA
Ames Research Center in Mountain View, California, on December 7-9,
1993. A copy of the meeting notes is included in this section.
-------
MRLC Consortium
Documentation Notebook
January, 1994
MRLC Consortium Meeting
NASA Ames Research Center
Mountain View, CA
11/16/93 to 11/18/93
Meeting Notes
Meeting Participants:
Thaddeus Bara (TB) - ManTech Environmental, NC
Susan Benjamin (SB) - USGS (Ames), CA
Ed Bright (EB) - Oak Ridge National Laboratory, TN
Jeff Eidenshink (JE) - Hughes STX, SD
Don Field (DF) - NOAA, NC
Len Gaydos (LG) - USGS (Ames), CA
Ray Harris (RH) - San Jose State University, CA
Joy Hood (JH) - Hughes STX, SD
Mike Jennings (MJ) - FWS, ID
Dave Peterson (DP) - NASA, CA
Denise Shaw (DS) - USEPA, NC
LaRue Smith (LS) - USGS, NV
Gail Thelin (GT) - USGS, CA
Dorsey Worthy (DW) - USEPA, NV
Notes by: TB, Draft-11/24/93
Meeting Agenda: see Appendix A
Tuesday, 11/16/93
MEMORANDUM OF UNDERSTANDING
o JE provided a copy of National Mapping Division guidelines for MOUs. Every agency
has own format, and the difficulty in integrating multi-agency formats was recognized.
o There is a need for agencies to determine who will sign MOU; signatories should be at
comparable level in respective agencies. It would be easiest to do at the Program level
(ie., Martinko with EPA, or Lauer with EDC), but with the number of agencies and the
amount of money involved, higher-level signature may be necessary.
- MOU currently being reviewed by head of NOAA COP who signs all of the
agreements (DF).
- DS indicated that an EPA AAO would be the probable EPA signatory.
— JE thought Don Lauer might sign it, alternately it could go to Alan Watkins,
Chief of NMD.
- Within USGS, there has been talk that it might have to go higher into Interior,
as inter-agency MOUs tend to get reviewed at higher level (GT).
-------
MKLC Consortium
Documentation Notebook
January, 1994
o DS suggested that since USGS appears to have the strictest criteria for approval, that
information on the level required should be guided by USGS. Once information from
USGS is available re level of signing, other Agencies can plan for signatures and set a
date.
o The effects of no signed MOU include impact on the transfer of resources between
groups (though some lAGs are operational, such as EPA/EDC), and the development and
implementation of consortium details. DF indicated that NOAA COP is facing a $1
million cutback that is not yet allocated between programs (including C-CAP), that MOU
may help C-CAP.
AMENDMENTS TO MOU
o The items discussed at MN meeting relating to proposed amendments to be included in
MOU were listed:
Data Purchase
First Regional Implementation
Evaluation of Clustering Technique
Pre-processing and Data Management (National level)
Product Sharing and Distribution
Accuracy Assessment
Information Management
Change Detection
o DW suggested that these items not be considered amendments to MOU, as this would
require review process. Rather, they can be formalized as a "working protocol".
Meeting participants agreed to this approach.
OTHER TOPICS
o DW expressed opinion that Consortium should be ready to deal with specifics. Data are
being received and there was a need for a plan of implementation of activities.
o Florida
— JE indicated that EDC was in process of creating a statewide TM coverage of
Florida using MRLC data. The effort was mandated by Watkins.
- DW indicated that a "1 inch = 1 foot" landuse map of Florida was being
generated.
— DF indicated that C-CCAP will be working in a 4-scene area of South Florida as
part of a statewide project next year.
o DS indicated that EMAP will be going into Region 3 next year.
o There was general, agreement that Accuracy Assessment effort needs to be made more
specific than the general outline developed at MN meeting. MJ talked later (see below)
of GAP Accuracy Assessment meeting.
EDC MRLC Project Re-organization and Land Characteristics Proposal (JE)
o JH will lead EDC effort for pre-processing TM data, including clustering, classification,
accuracy assessment to allow integration of TM/AVHRR as a monitoring tool.
o EDC will have 16 new staff, and $1.2 million in resources for next year on project.
Babbitt has expressed a strong interest in program, and presentations have been made to
the Secretary.
-------
MRLC Consortium
Documentation Notebook
January, 1994
o Loveland's work on AVHRR is far in advance, re intensity of work and resources, of
work on regionalization. Consequently he will be involved with development of land
characteristics database, and will take the lead on the landcover generation effort. He
has ideas on how to approach landcover issues, but is not interested in technical
processing issues, such as choice of clustering algorithm.
o Loveland envisions an ecosystem-based perspective (not political or other arbitrary
boundaries). EDC will merge different datasets (ie., spectral, terrain) so that landscape
can be characterized for particular applications. Thematic coverages of landscape units
would not be included in this proposal, but rather the supporting data for the
development of units appropriate to a particular application.
o Loveland has a research proposal for land characterization under review within EDC.
Comments are expected by the week after Thanksgiving, and the proposal should be
ready for review by other Consortium members in mid-December. The plan calls for
a 2-year development period for the database.
o This proposal does not attempt to develop a prototype, but to reach the goal of a land
characteristics concept (EDC does not want to independently develop details/specifics).
Data will be provided for program needs, but a major prototype is viewed as premature
by EDC, until data pre-processing issues are addressed. GT and DW disagreed, in that
advance experimentation and research on database issues are necessary.
o (JH) The proposed database will be made available to different programs. The database
would be consistently processed before going out The programs would return derivative
datasets to EDC that can be cross-tracked, and merged into a set of "super-structure"
datasets.
o As part of development, JH indicated that EDC will need to know what other programs
will require. A list of questions pertaining to these requirements was circulated
(Appendix B).
o DS thought that it was important for all agencies to be involved on an equal basis, and
not for EDC to work independently. JE said the plan was for EDC's benefit to "get own
thoughts together" regarding process, and will be looking for comments from other
programs for integration in mid-December.
PROJECT DOCUMENTATION
o TB provided an overview of the plan to develop documentation for previous Consortium
activities in 1993, and a proposed bi-monthly update of continuing activities. Overheads
used in presentation are included in Appendix C.
o TB will be visiting EDC in early December. He will collect information from
EDC/Hughes SIX personnel who were previously active in Consortium (including Tom
Holm and Jim Sturdevant), and will collect detailed information on the EDC archiving
and processing protocol, including the results of related research conducted ar EDC, and
relevant algorithms and supporting documentation (JH).
PRE-PROCESSING
o EDC has ordered 200-300 scenes; 34 have been received as of 11/15/93. More details
were to be provided later (see below).
o geo-referencing
— To get +/- 1 pixel accuracy will require terrain correction.
-------
MRLC Consortium
Documentation Notebook
January, 1994
i. EB did not think it would be necessary for the flat areas, and GT thought
terrain correction would introduce error (from merging 60 m DEM with
30mTM).
ii. DW and JE thought than TM data should be co-registered with DEM.
This would allow for seamless merging of multiple scenes.
- (JH) EDC's current approach involves trying to use DLGs and TM Control Point
Library: a) select control location, b) record elevation at point, c) apply geo-
correction grid or polynomial (first order). Terrain correction will be done at
time of resampling.
o clustering
— CCAP does not want clustered data, according to protocol.
o DW thought these details are necessary to work out in advance of developing a long-term
environmental database. Different pre-processing and processing regimes would affect
the ability to merge derivative datasets with EDC land characteristic database (IE).
METADATA
o (DW) There is no standard metadata content or format for raster data. The ultimate
source for the standard will have to be the FGDC. Metadata standards from each of the
programs should be provided to consortium members. DW recommended that the format
of the EDC distributed dataset be considered as a standard.
o C-CAP had prepared a metadata document for Ches Bay data. The parameters included
in the document were specified by the NODC.
o JE indicated that there is no metadata standard available. EDC will comply with
whatever is appropriate. JE noted the difference between what metadata is compiled and
what would be appropriate for a front-end sheet with a dataset that someone can use.
o Documentaml was described as being difficult to use. There 106 pages to each
document, there would be many blank fields for a typical application, and the document
is difficult to page through to find what is important GT said it should be possible to
produce a narrative containing the critical information contained in the metadata.
o EDC will record image metadata, and needs to have technical contacts with each program
to obtain metadata requirements (JH).
— As the .ddr file is not complete, additional information will be provided (sample -
Appendix). EDC can build the Pathfinder dataset into MRLC data (JH).
— EDC metadata will comply with FGDC as close as possible (JE).
- EDC will compile data and seek ways to include as much other pertinent
information as necessary.
ACCURACY ASSESSMENT
o (MJ) As a result of recent review panel and discussion between primary participants,
GAP is committed to a serious funding of accuracy assessment and the development of
a technical paper on accuracy assessment for GAP.
o (MJ) A meeting will be held February 2-3 at UCSB with the intent of providing written
statements that would serve as a basis for a technical accuracy assessment document.
GAP wants to include the needs of other agencies, and is seeking to maximize
information sharing.
o The document will develop statistical performance reviews of GAP'S map products. The
-------
MRLC Consortium
Documentation Notebook
January, 1994
meeting will seek agreement over the standards (i.e., what standards and how they are
developed) for thematic and position accuracy and sampling (field, airborne) issues. The
number of information fields will be determined, and a common set of fields will be
developed. The focus will be on independent datasets to measure map performance.
OTHER TOPICS. II
o DS indicated that EMAP is working on 4 ongoing projects with DoD:
— Ches Bay: fusion of TM data with high spatial resolution sensors
- National Stream Survey: sampling frame is DLG, these include non-stream (i.e.,
arroyo, etc) features as apparent streams, DoD will use archived data to assess
actual flow status in these features at specified times
- CA vernal pool study
— Estuaries: interest in submarine landscape ecology
o (MI) The Pathfinder meeting in Durham NH will discuss several topics including what
to do with the loss of Landsat 6. The group wants future sensor design to be user driven
so wants to stay in touch with user community. MRLC members were invited to
participate as MRLC is perceived is perceived as a major user.
— Apparently a major multi-agency purchase is supposed to happen in 1994.
CONSORTIUM PROGRAM DESCRIPTIONS (FOR BENEFIT OF NASA AMES/USGS
(AMES)
o DW described NALC
— Major change in program: NALC will not be labeling change products in
conterminous US, and will only do so in humid sub-tropics, due to budget
limitations
- Currently three pilot areas: Ches Bay, Chiapas, Oregon Transect
- dataflow:
i. data assembly at EDC (to be completed EOY 1995 or early 1996)
ii. clustering at EMSL-LV
iii. labeling by cooperators
iv. accuracy assessment at EMSL-LV (photo-interp., except for Mexico
where TM will be used)
v. data archived at EDC, and available to researchers and non-profits
- NALC is cooperating with geologic survey in Mexico and University of Mexico
o Ml provided overview of GAP
- GAP will now include aquatic ecosystems (though not estuarine or marine). GAP
will work with C-CAP over the next 6-month time frame to evaluate
compatibilities
o DF provided an overview of C-CAP
o GT provided an overview of NAWQA
- The Merced County Landuse Map provided a discussion on the complementarity
of the GAP and NAWQA classification schemes. The LU map had higher order
classification of agricultural lands, with a general class of "native vegetation".
MJ indicated that it is these areas which GAP will classify in more detail.
- GT saw a need for a broad land classification framework with which more
detailed information could be integrated. This additional information would
-------
MRLC Consortium
Documentation Notebook
January, 1994
include finely detailed ag classes or land management practices (including
silviculture practices or grazed rangeland, as examples).
— LG provided some comments regarding his experience in a classification exercise
in Alaska. After much work went into developing an acceptable standardized
classification, few researchers actually used this product. Most wanted direct
access to the original database in order to re-interpret data for own particu_ar
purposes.
DEM ERROR /^ND EFFECT ON TERRAIN CORRECTION
o A detailed discussion was held on DEM errors and their effect on terrain correction of
TM data. William Acevedo (WA) of USGS (Ames) participated in the discussion.
o WA discussed a paper he had prepared comparing and evaluating error in several DEM
types (DMA, NCAR 15 and 30 second, l:250kDEM, l:24k DEM, and 10 and 15 meter
resolution grids interpolated from 7.5 minute hypsography surface).
— Study was in San Mateo Co., and included Santa Clara Mt. range
— Study was mostly qualitative, with quantitative analysis across SC mountains.
- WA evaluated the presence and severity of error, but not the affect of error on
particular applications.
- Differences in elevation were attributed to spatial resolution differences and
artifacts from interpolation algorithms and stereo-profiling (including striping and
directional biases)
— The DMA datasets were characterized by flat ridges and valley floors. The
contour interval of 200 feet "lost" intermediate ridges. There was a bias towards
contours as evidenced in histograms, and appearing as benches or steps in data.
- l:100k products had fewer artifacts than l:24k products.
— The maximum DEM error for DMA products was 200 feet
- Sue Jenson is working in 3 pilot areas to develop a multi-resolution DEM
database with an application to global research analysis.
o EB indicated that terrain correction will not affect C-CAP analysis.
o JH indicated that EDC would not archive geo-referenced image. Control points and p-
code images would be archived. Control point selection is the time intensive process (see
below).
o WA thought that difference in spatial resolution between DEM and TM would have more
impact on terrain correction than DEM artifacts. JE indicated that for 1 km. data set,
terrain correction used ETOPO5 elevation data; so terrain data can be lower resolution
than image data and maintain effective correction.
o Terrain correction for the TM data was agreed to by all consortium programs.
IMAGE PROJECTION
o Group needs: CCAP required by protocol to use UTM (DF), NAWQA has no problem
with UTM, except for distortions near edge of zones (GT), GAP would accept UTM as
standard - cooperators could re-project (MJ).
o As spatial extent increases, the effect of error from projection decreases (JE). Therefore,
recognizing that the data will frequently be subset for finer scale studies, the projection
should be as accurate as possible, i.e., UTM. For coarser scale studies, the data can be
re-projected (DW).
-------
KRLC Consortium
Documentation Notebook
January, 1994
The decision was made to use the 83 datum, though EDC will maintain the 27 datum as
an option, since much of analog data is referenced to this datum.
Summary of defaults to be used by EDC:
- 30 m. pixels
— UTM projection (with data archived in p-code)
- 1983 datum (with 1927 option)
— terrain correction (with option for no terrain correction)
(IE) EDC will add so much value at default parameters than ordering agencies will have
little incentive to change defaults
The issue of projection has a significant effect on the compatibility of the returned
derivative datasets in long-term EDC database.
— if users re-project data, or request data in different projection from default, will
not be consistent with longterm database.
— on national level, error will be minimized (IE), but derivative data from study
areas of smaller extent will be problematic
- (JE) Not practical to enforce a single standard on all users
— most users of the original and derivative data in longterm database will not care
if data had been previously re-projected
— conclusions
i. EDC will retain returned derivative data in returned projection
ii. EDC will reconstitute data to a national level database
iii. metadata in longterm database will include information on reprojection of
returned derivative datasets
-------
MRLC Consortium
Documentation Notebook
January, 1994
Wednesday, 11/17/93
o DP provided an overview of NASA Ames, Earth Sciences research. Handouts were
provided to each participant.
KHOROS/SPECTRUM
o JG provided an overview of Spectrum. A paper by the developers of the Los Alamos
clustering algorithm (White and Kelly) was provided (Appendix D).
- represents a "smart" compression of data, with a toolbox for display and analysis
— components:
i. clustering into large number of classes
ii. original multi-band data is discarded
iii. clustered data is displayed
iv. display is "super" photo-interpreted
- key is a) clustered image and b) codebook containing mean vectors for each
wavelength in each cluster.
— codebook drives analysis and display through use of look-up tables; direct and
custom transformation of mean vectors possible through codebook
- Spectrum is currently limited to spectral data only; not possible to integrate
ancillary data in interpretation
— Spectrum resides in Khoros environment and is currently dependent on it, though
this should not be necessary
— Clustering done in Khoros using Los Alamos algorithm provided as part of
toolbox
o SB provided more detailed information on Khoros/Spectrum. Her overheads are included
in Appendix E. Additional notes are contained below.
- Current version of Khoros is V.I, Patch 5. V 2.0 is under development
- Khoros has been ported to DG. Spectrum is currently limited to SUN.
- Because of Khoros limitations, cannot process whole TM scenes with less than
64 MB of memory.
o LG provided a demonstration of Spectrum-based image interpretation using spectral data
from the SF Peninsula.
o PH provided a demonstration of interpretation results using Spectrum and ERDAS for
the Elkhorn Slough, CA. Handouts were provided and included in Appendix F.
o SB provided an example of the use of Spectrum with multi-temporal data from the
Willamette Valley, OR.
POST DEMONSTRATION DISCUSSION
o (LG) If a standard clustered product is decided on by Consortium then it is likely that
private vendors will seek to integrate these datasets and codebooks in their systems to
advance business development. For example, ESRI might develop, at their own expense,
Spectrum-like tools. The time frame would be uncerte jn, but LG speculated that it might
be a year for development.
o (JE) There are other tools for organizing and analyzing data, including DDL and Infobase,
that might be worth investigating. The focus on Spectrum might be too limiting.
Spectrum has many problems, including the lack of statistics carry-over.
-------
MRLC Consortium
Documentation Notebook
January, 1994
DW thought there really wasn't time to wait on evaluation or private vendor development
because of pressure for products. It would be easier to select an algorithm now and
change it later if future research shows this to be appropriate.
LG thought that a stand-alone version of Spectrum could be generated either at Ames or
UNM. The code is available for modifications. MJ thought that a stand-alone version
is important as soon as possible. Their constituent group would be dependant, due to
high resource cost of Khoros, for stand-alone. MJ did not see the consortium able to pay
for the development, however.
LG indicated that a more solid version of Spectrum within Khoros would be available
more quickly than a stand-alone. A strong expression of interest to Los Alamos and
UNM with additional funding for a stand-alone Spectrum, however, should accelerate the
process. It would take perhaps 6-9 months to develop a stand-alone version.
The Consortium should take the lead in contacting UNM and Los Alamos. In particular,
EDC should contact them (see below). Additionally, it should be possible to import
ERDAS or PCI clustered products into Spectrum, but this needs more investigation,
particularly regarding codebook construction.
Whether stand-alone or coupled with Khoros, the need to incorporate ancillary data for
interpretation is critical (DW).
LG indicated that cluster algorithm developer (White) advocates using 4096 clusters,
though LG indicated that 240 should be acceptable based on Ames work. Current
hardware does not support the display of more than 240 classes. JE indicated that 4096
clusters don't exist in a dataset, anyway. 240 clusters are available and workable now
(LG).
Ames and EDC (through JH) will work together to 1) get Los Alamos cluster algorithm
in Khoros installed at EDC, and 2) to talk at system level about what algorithm is doing.
- EDC should be able to get from Los Alamos a version of Khoros ported to the
SG.
Ames has prepared and submitted a list of Spectrum improvements to Los Alamos and
UNM (Appendix G). The Consortium member agreed with the list. LG hopes that most
of this list will be implemented in the next version of Spectrum.
Regarding multi-temporal scenes, EDC will use 12 bands as input and one band as output
as product. It will be possible to back out the original individual scenes by subsetting
out the codebook.
-------
KRLC Consortium
Documentation Notebook
January, 1994
Thursday, 11/18/93
EDC PREPROCESSING OH!
o Overheads were provided by JH and are included in Appendix H. Additionally, JH
provided a list of scenes (Appendix I) with annotation by Paul Severson (Hughes STX),
and a list of scenes showing the order status (Appendix J).
o There have been major changes in scene selection, and EDC has flagged a number of the
multi-temporal images that looked questionable in their judgment.
- MJ has talked with 40 different remote sensing specialists involved with GAP,
and has had a number of conference calls with GT and Severson. The growing
seasons must be conterminous, as he had provided for.
— GT has tried to stay with the same seasons MJ had wanted in her selection.
o Regarding scene review, EOSAT has been very cooperative, and has been reviewing
ordered scenes on own initiative.
o To date 430 scenes have been ordered, 100 still to be ordered, plus ISO scenes
previously ordered by GAP.
o The deadline for ordering scenes under the EDC/EOSAT agreement has passed, but
EOSAT has been flexible. They do want additional scenes to be ordered by
Thanksgiving.
- Assuming that all 150 GAP holdings meet requirement, Consortium only needs
to order 25 by Thanksgiving.
- The question was raised whether the FL mosaic could be qualified as pre-existing
holdings.
- EDC has prepared a list of 67 scenes, and needs guidance from Consortium as
to whether these scenes should be ordered.
i. EDC has flagged all multi-temporal scenes from 2 years with a p
(proposed to purchase) or a q (questions on scene selection).
ii. GT agreed to review all of the q's, review through GLIS and decide with
MJ before Thanksgiving on which scenes to be ordered.
o EDC is receiving 8 mm tape from EOSAT, and are archiving on 3480 format (JH), A
custom computer program is necessary to make format conversion. The source code was
completed last week, so will being the conversion/archiving process.
o JH provided an overview of proposed EDC processing protocol (see Appendix H for
details; note that based on discussions today, this protocol will be modified [the modified
protocol will be included in the documentation notebook - TB]).
- prior to point selection will run preview.pdf then print b/w prints of all 6 bands
- control point files will be archived (to allow response to request for alternate
projections); EDC will add point locations to control point file metadata at request
of Consortium (JE)
— clustering can either precede or follow registration (EDC will default to cubic
convolution; as C-CAP will use nearest neighbor resampling); a choice will be
provided to agency).
i. C-CAP will take p-code data and register and pre-process (DF,EB)
ii. image-to-image registration can subsequently be performed on C-CAP
derivative products (DW)
ill. Derivative product needs to be well-documented; EDC should not be re-
-------
MRLC Consortium
Documentation Notebook
January, 1994
processing derivative data
- JH estimated 8 hours of pre-process time per scene per operator
o Tom Holm (EDC) will provide information on how processing amounts will be billed,
and level of EDC contribution.
o Processing will begin by the end of December. New hires at Hughes STX will be
trained using real data (IE).
ISSUE OF PRIOR GAP HOLDINGS
o Under purchase agreement with EOS AT, Consortium must identify a complete coverage
of US, and provide back to EOS AT. Because 23 scenes (from Montana) and as many
as 37 other scenes (including New England) were ordered from EOSAT with different
pre-processing, Consortium may have to face situation of purchasing additional scenes
to make up for these.
- MJ expressed strong reservations about the Consortium having to do this. The
scenes are in a format which GAP can use, and they should not have to buy new
scenes to meet agreement.
- Both JE and DW thought that the Montana scenes in current form represented
copyrighted products, and thus fell under the agreement. Alternately, EOSAT
could go back to their own archives, or to the unprocessed data that was delivered
to make up for these scenes.
- It is not clear in what format the GAP scenes were ordered. MJ thought they
were p-code. The scenes are being sent to EDC for review, but only a few have
been received. Those scenes were terrain-corrected. But until all GAP scenes
are in, EDC will not know what type of format.
PILOT STUDIES
o LG and SB offered to assist the Consortium in classification of the pilot areas.
o MJ indicated that there will be a Region 3 Technical Meeting in March or April, 1994;
this meeting sets a deadline for Consortium action. MJ thought it would be a good idea
to have the first clustered datasets ready for this meeting.
o Two study areas were identified - Ches Bay and Oregon
- All scenes but one of the 15/33 multi-temporal scenes, have already been ordered
for Ches Bay area. Initially 6 path/row pairs (12 images) were considered as
priority. As an initial focus, EDC clustering will emphasize the NAWQA Lower
Susquehanna Study Area, contained in 14/32 and 15/32 (note: the 15/32 scene is
in October, at GAP's request, and will contain peak biomass, meeting C-CAP's
interests). The SA is entirely contained within Region 3.
i. EDC will start with these data, since already in-house or enroute.
ii. EDC will start processing these data no later than 1/1/94.
- Scenes in Oregon were selected to include C-CAP study area at mouth of
Columbia (C-CAP will have one scene done by mid to end of 1994), as well as
NALC study area, and GAP area along WA border.
i. Data along WA border already in GAP holdings; other data in OR part of
NASA Data Grant (total of 4 path/rows); also have NALC triplicates for
some of study area
ii. EDC will provide corrected data to Ames for clustering
-------
MRLC Consortium
Documentation Notebook
January, 1994
iii. OR paths 45, 46, 47; OR rows 28, 29 will be included
iv. LG indicated that all 6 path/rows can probably not be completed by 3/94;
but will do what can be done
v. NAWQA and GAP staff can have access to Ames to do clustering (LG).
EDC has identified Kent Hegge as customer service contact for Consortium. Hegge
shares office with Severson, who has been moved from customer service to product
support. A product tracking protocol (referenced by scene) will be developed by Hegge
and Severson as part of product support.
-------
MRLC Consortium
Documentation Notebook
January, 1994
LSI OF
Meeting Notes
MRLC Consortium Meeting
NASA Ames Research Center
Mountain View, CA
11/1693 to 11/18/93
APPENDIX DESCRIPTION
A Meeting Agenda (GT)
B EDC MRLC Requirements (JH)
C Consortium Documentation Efforts Overheads (IB)
D Spectrum Paper (LCD
E Spectrum Overview Overheads (SB)
F EEkhom Slough Demonstration (PH)
G Spectrum Upgrade Requests (LG)
H EDC Scene Status and Preprocessing (JH)
I Scene Selection Notes - EDC (JH)
J Scene Ordering list - EDC (JH)
-------
MRLC Consortium
Documentation Notebook
January, 1994
APPENDIX A
-------
MRLC Consortium
Documentation Notebook
January, 1994
Interagency Consortium for Land Cover Mapping
NASA/Ames Research Center, Building 242, Room 206
November 16-18, 1993
Tuesday November 16
9 - 9:30 Welcome from EROS/Ames Research Center (Gaydos)
Introductions
9:30 - noon -Status of Interagency MOU
•Project Documentation
•Discussion: Metadata standards for raster data
•Accuracy Assessment - GAP meeting in December
•EPA Chesapeake Project (using high resolution classified
data in conjunction with TM data)
•NASA Landsat Pathfinder Meeting, December 1993
noon -1
1 -2
2-5
Lunch in Galileo Room
Overview of programs for Ames participants
Each program representative gives a 10 minute program
overview and program requirements for land cover data
•Statistics on final Scene Selection (# of 1991, 1992, 1993
scenes; Path/Rows with multi-temporal coverage; which
Path/Rows need to be ordered, etc.)
•How many scenes have been delivered to EDC
•Description of data base used to catalogue and track status
of data processing
•Path / Row Priorities for processing
•Update of proposed preprocessing methodology - geometric
correction and terrain correction (evaluation of technique
in various environments)
•Schedule for delivery of data to projects -
-------
MRLC Consortium
Documentation Notebook
January, 1994
Wednesday. November 17
8:45 - 9:00 Welcome from SG/SGE (Lawless/Peterson)
9:00-11:00 Spectrum system overview (Benjamin/Gaydos)
Clustering/Classification methods review
LANL approach to image classification
Spectrum - interpretation into land cover units
11:15- 3:00 Workstation Demonstrations
LUDA interpretation - Santa Clara Co. (Gaydos)
C-CAP interpretation - Elkhom Slough (Harris)
(break for NAWQA Cropland Interpretation - Willamette Valley (Benjamin)
lunch) Tools for interpretation (Benjamin)
Use of Ancillary Data (Benjamin)
3:00-3:15 Break
3:15 - 4:30 Future of Spectrum (Gaydos)
Khoros 2.0
Changes for Spectrum
Possibilities for commercialization/CRADA
November 18
Hands-on demonstrations of Spectrum using SGE workstations.
-------
Attendees
MRLC Consortium
Documentation Notebook
January, 1994
Who
Ed Bright
Don Field
Denise Shaw
Dorsey Worthy
Thaddeus Bara
Mike Jennings
Gail Thelin
LaRue Smith
Jeff Eidenshink
Joy Hood
Len Gaydos
Susan Benjamin
Ray Harris
Dave Petersen
Jim Lawless
Organization
NOAA
NOAA
EPA
EPA
EPA (ManTech Environmental)
USF&WS
USGS WRD
USGS WRD
Hughes SIX (USGS NMD)
Hughes SIX (USGS NMD)
USGS NMD
USGS NMD
SJSU
NASA
NASA
Location
Oak Ridge Nat. Lab
Beaufort, NC
Research Triangle Pk
Las Vegas
Moscow, ID
Sacramento
Carson City, NV
EDC, Sioux Falls
EDC, Sioux Falls
EROS Ames
EROS Ames
Ames
Ames - SGE
Ames - SG
-------
MRLC Consortium
Documentation Notebook
January, 1994
APPENDIX B
-------
MRLC Consortium
Documentation Notebook
January, 1994
MRLC REQUIREMENTS
bduction
The goal of this document is to define "what" functionality is
required for the MRLC project in the areas of database maintenance,
product ordering, and searching capabilities. The issues of "how"
the required functionality is implemented will be covered in the
MRLC design document.
The requirements listed below were gleaned from the Land Science
Data Archive Data Base Design Review document. These requirements
can be used as a starting point for the MRLC Requirements document.
MRLC requirement questions for each section are listed after the
requirements.
1. Database maintenance requirements 1
- Create a MRLC database using the ORACLE database management system.
- Load with data from Landsat/AVHRR/Aircraft/SLAR/SPOT production
datasets
- Increment links by 1 when records are added
- Load the date_entered field when the record is added
- Load the date_updated field when the record is modified
- Provide ability to write metadata to ascii files for plotting
purposes
- Calculate corner points for Landsat, SPOT, Aircraft, and AVHRR data
(Will need recording technique for Landsat data, May not be possible
for other types of data)
- Convert all incoming derivative data to 3480 archive media
- Write all complex data to 3480 archive media
- Stuff complex link in derivative dataset by searching the complex
dataset with the path, row, and acquisition date
- Automatically cross reference the tape library's media location
(We are working with two different database management systems
here - ORACLE & UNIFY - may require large development effort)
- Derivative products with more than one media location need a
derivative media dataset
- Provide periodic management reports
estions:
- How much information is duplicated or copied from the original source
dataset? Can the original source dataset be accessed to get the
corner points?
-------
MRLC Consortium
Documentation Notebook
January, 1994
- If more than one project uses the same original source will
updates to the original source require automatic updates to all
the project basic datasets?
- How are automatic cross references to the tape library's media
location handled with the original source datasets?
- Are there going to be derivative products with more than one
media location?
- What kind of management reports are needed and what kind of
information do they provide?
2. Product ordering requirements ., .-•'
- Interface with DORRAN to provide product ordering
- Have first two characters in ordering_id identify the dataset
Questions:
- Can DORRAN handle products with more than one media location?
- Are there any special price breaks, ordering restrictions,
special processing, purchasing/brokerage agreements needed
for ordering MRLC data that DORRAN currently does not have?
i. Searching requirements
- Provide for pa-rh/row and geographic queries
- Search metadata by path/row and/or study region for each level of
derivative products generated.
- Index higher level datasets to the original source or to the
National Satellite Land Remote Sensing Data Archive
- Provide a tracking system for contributor derivative data
Questions:
- How will the consortium members access the information system
at EDC? What kind of equipment or network capability do they have?
INTERNET? X windows?
- What kind of user interface is wanted? ASCII? GUI?
- Will MRLC data be restricted to consortium members or can anyone
order and research it?
- Is browse capability of original source data wanted?
- Is browse capability of derived data needed?
- Will everyone provide data processing information for derived
data? At what level will it be provided? dataset level?
granule level?
-------
MRLC Consortium
- Are geographic queries of derived data needeS?cWSt!^^Ia?y. -•
corner point information? Will there be more than 4 points'?'
- Are queries by study regions needed? Who defines the study region?
- Is a tracking system of contributor derivative data needed?
Who would use it? Who needs access to it?
- Do we want to include MRLC data in GLIS or IMS Version 0 (NASA)?
-------
MRLC Consortium
Documentation Notebook
January, 1994
APPENDIX C
-------
MRLC Consortium
Documentation Notebook
January, 1994
DELIVERABLES AND DISTRIBUTION
TYPES OF DELIVERABLES
1. INITIAL NOTEBOOK
O CALENDAR YEAR 1993 DOCUMENTATION
o DESCRIPTION OF CONTINUING EFFORTS INCLUDING STATUS
AND TIMELINES
2. BIMONTHLY UPDATES
O SUMMARY OF CONTINUING ACTIVITIES WITH SUPPORTING
DOCUMENTATION
3. CENTRAL FILE SYSTEM
INTENDED AUDIENCE
1. PROGRAM STAFF (DIRECT DISTRIBUTION ON BIMONTHLY BASIS)
2. HIGHER-LEVEL AGENCY STAFF (THROUGH PROGRAM STAFF)
3. OUTSIDE WORKSHOPS AND REVIEW PANELS (TO SERVE AS
BACKGROUND INFORMATION)
-------
MRLC Consortium
Documentation Notebook
January, 1994
INFORMATION REQUESTS FOR INITIAL NOTEBOOK
1. MEMORANDUM OF UNDERSTANDING - MOST RECENT DRAFT
(RECEIVED 11/8)
2. TM PURCHASE AGREEMENT
O MODIFICATION 16 TO BASIC OPERATING AGREEMENT (BOA)
BETWEEN EDC AND EOSAT, AND LETTER OF CONCURRENCE
O COPY OF BOA (?)
o UPDATES ON EARTHSAT ISSUE
3. SCENE SELECTION CRITERIA
4. SCENE ORDERING AND DELIVERY SCHEDULES
5. PRE-PROCESSING
o RESULTS OF RELATED RESEARCH
O PROTOCOL DOCUMENTATION (ALGORITHMS AND PROCEDURES)
o STATUS OF SCENE PRE-PROCESSING
6. DATADASE DESIGN AND MANAGEMENT
O DATABASE DESIGN REPORT
o METADATA STANDARDS
O PRE-PROCESSED DATA ARCHIVING AND AVAILABILITY
7. COPIES OF NOTES TAKEN PREVIOUS MEETINGS BY
CONSORTIUM PARTICIPANTS
o COMPILATION INTO FORMAL MEETING NOTES (TO BE CIRCULATED
FOR REVIEW, EDIT, CONCURRENCE PRIOR TO INCLUSION IN
NOTEBOOK)
O CONFERENCE CALL NOTES AS AVAILABLE
8. COPIES OF COMMUNICATIONS
-------
MRLC Consortium
Documentation Notebook
January, 1994
OBJECTIVES OF DOCUMENTATION EFFORT
1. DEVELOP A DOCUMENTATIVE HISTORY OF CONSORTIUM
ACTIVITIES IN CALENDAR YEAR 1993
O DOCUMENTATION OF PREVIOUS MEETINGS AND CONFERENCE
CALLS
O PROGRAM GOALS AND NEEDS AS RELATED TO CONSORTIUM
o MEMORANDUM OF UNDERSTANDING
O TM PURCHASE ORDER AND ACQUISITION
O PRE-PROCESSING RESEARCH ACTIVITIES AND DECISIONS
O DATABASE DESIGN, ARCHIVING, AND DISTRIBUTION
2. PREPARE AND MAINTAIN A CONTINUING HISTORY OF
ON-GOING CONSORTIUM ACTIVITIES
O DEVELOP PROTOCOL FOR INFORMATION TRANSFER
TYPE OF INFORMATION TO BE TRANSFERRED
MODE OF TRANSFER
-------
KRLC Consortium
Documentation Notebook
January, 1994
APPENDIX D
-------
MRLC Consortium
Documentation Notebook
January, 1994
Preprocessing remotely-sensed data for efficient analysis and classification
Patrick M. Kelly, James M. White
Los Alamos National Laboratory, Computer Research Group
MS B-265, Los Alamos, NM 87545
ABSTRACT
Interpreting remotely-sensed data typically requires expensive, specialized computing machinery capable of stor-
ing and manipulating large amounts of data quickly. In this paper, we present a method for accurately analyzing
and categorizing remotely-sensed data on much smaller, less expensive platforms. Data size is reduced in such a way
as to retain the integrity of the original data, where the format of the resultant data set lends itself well to providing
an efficient, interactive method of data classification.
1. INTRODUCTION
A Landsat Thematic Mapper (TM) quarter scene consists of approximately 12 million pixels, each being repre-
sented by seven spectral reflectance values between 0 and 255. Each quarter scene, therefore, occupies 84 megabytes
of storage, and performing even simple data manipulations for analysis or display purposes requires a large number
of operations. By preprocessing the data by a technique known as vector quantization or clustering, computational
requirements necessary for image analysis and manipulation are greatly reduced.
The advantages to clustering large data sets are numerous. Many times when scientists work with multispectral
image data, they are interested in grouping together sets of similar data - something that clustering algorithms do
Clustered data also has a number of properties that simplify data analysis and categorization. Data
is a very desirable by-product of the clustering process, reducing the computational resources necessary
to manipulate the data. Additionally, because pixels belonging to the same cluster are intrinsically associated with
one another, sets of pixels in an image which share common characteristics can be manipulated simultaneously.
Statistics for each cluster can easily be calculated during the clustering process, allowing many properties of the
original data to be retained. For many applications, we have found that once clustering has been performed, the
original data is no longer needed.
Each pixel in an image is commonly categorized according to its spectral signature. Many methods are used
for classifying multispectral data, including both supervised and unsupervised classification methods [1, 2]. When
using supervised methods for data classification, a user selects training areas representative of several types of
land cover, and a classifier is developed to discriminate between different classes. This classifier is then used to
categorize the remaining pixels in the scene. Numerous pattern recognition algorithms of this type exist, including
nearest neighbor algorithms, discriminant function techniques, artificial neural networks, and statistical methods. An
overview of these techniques can be found in standard pattern recognition textbooks [3, 4]. Statistical methods such
as maximum likelihood classifiers [3] have always been popular for this type of problem. In general, although these
techniques often work well, they are very time consuming both in computer time and operator effort. Additionally,
they do not tend to allow easy classifier adjustments (or "fine-tuning") for the system.
Unlike supervised methods of classification, which require a user to define training sets, unsupervised techniques
require no training sets at all. They instead attempt to automatically find the underlying structure of multi-
dimensional data, by "clustering" the data into groups sharing similar characteristics. Unsupervised classification
is an off-line process, requiring very little time of the system user. A user simply needs to specify a number of
clusters to find, and allow the classification program to do the rest. This technique assumes, however, that the
number of natural categories present in the data is known a priori, with data from different category clusters being
-separated.
aril-
-------
MKLC Consortium
Documentation Notebook
January, 1994
Figure 1: Clustered Representation of Multispectral Image Data
When using clustering methods for analyzing multispectral data, many people attempt to define a relatively
small number of clusters - between 5 and 100 clusters, for example. Our technique relies on the fact that many
clusters (between 256 and 4096) can be defined for the data. The method of data analysis and classification presented
in this paper first preprocesses the data using a fast clustering algorithm. We cluster the data using a relatively
large number of clusters (as compared to the number of categories we wish to define for the data), and then use
the clustered data for analysis and classification. For many applications, there is no need for the original data after
clustering is performed. Using the clustered data, we can efficiently manipulate computer displays as well as analyze
and categorize data.
2. CLUSTERING METHODOLOGY
The basic principle of clustering (or vector quantization) is to take an original image (for our example, containing
around 12,000,000 pixds with each pixel being represented by a seven-dimensional vector), and represent the same
image using only a small number of unique pixel values. A codebook of N "best pixel values" to represent the image
must first be generated by some iterative method (the "construction" phase of the clustering algorithm). Once we
have generated these values, we step through the original image and assign each pixel to the cluster of the closest
match existing in our codebook (the "projection" phase of the clustering algorithm). Figure 1 shows the clustered
image representation, as compared to the original image representation.
In processing the data this way, two things have occured. First, we have reduced the volume of data needed to
represent the image by a factor of seven. This is reflected by the fact that we now need only a single band of image
data which contains indices into the codebook of reference vectors. Second, we have done a preliminary classification
of the data: similar pixels in the image are now intrinsically associated with one another.
Since we would like the clustered data to adequately represent the original data, the selection of the codebook
vectors is very important. By increasing the number of clusters, the accuracy of image representation can be
improved. Depending on the application, we use between 256 and 4096 clusters for a typical TM quarter scene. The
time required to cluster the image increases as the number of clusters increases. After clustering has provided a set
of clusters, the statistics for each cluster are computed and stored in the codebook along with the cluster reference
vectors. This is an important step because from these statistics, the combined statistics of the original data can
easily be computed.
As an extra step, the cluster indices are sorted according to values stored in the mean vectors. Before this step
is performed, the single two-dimensional band of cluster indices representing the data is meaningless unless it is
associated with its codebook. By sorting the clusters according to values in a single dimension, or by the sum of
multi-dimensional components in each one, a physical meaning is associated with each index. Bright pixels in the
-------
MRI/C Consortium
Documentation Notebook
January, 1994
iginal data set will be associated with larger cluster indices than the darker pixels. The result will be an image
which, when not associated with its codebook, can easily be displayed as a black and white image of the current
scene.
3. CLUSTERING ALGORITHM
Many types of clustering methods have been developed and analyzed for use with different types of data [3. 5]. In
general, many of these algorithms attempt to find a partitioning of a given data set that minimizes a predetermined
cost function. The k-means clustering algorithm [4] attempts to minimize a squared error cost function by manipu-
lating a set of k cluster centers. In particular, this algorithm tries to partition the data into k clusters, denoted by
C,, with the representative vector for each cluster (i,) being defined as the within-cluster mean:
Xf =
(1)
This algorithm iteratively moves vectors between clusters in such a way as to minimize the total squared error:
*' ^ ll^-iill1
Error =
(2)
This algorithm, however, becomes painfully slow when using very large data sets. One basic problem is that a
tremendous number of vector distance calculations must be performed during both the "construction" and "projec-
tion" phases of the algorithm. Several methods have been developed to improve this situation [6, 7, 8]. Many of
these schemes work very well in lower-dimensional spaces, but still tend to have a difficult time as the dimension of
the problem and number of clusters increase.
5-
TIMINGS POK MOSCOW SCENE
TIMINGS POM ALBDQUCftqUE SCXNC
1-
Nuabvr of Clu«t,«ri
O-V-C Contract
• Total
B«B Ca»traa~t •-*-* Project
• Totml
Figure 2: CPU Timings for Moscow and Albuquerque Scenes
-------
MRLC Consortium
Documentation Notebook
January, 1994
We use a. version of the nearest neighbor algorithm proposed in [9], where cluster positions are sorted along one
of the axes for the data. This algorithm, like many others, does not continue to work effectively as the problem
dimension increases. To combat this, we use the first principal component of the data as the axis on which to do the
sort. This axis gives the best possible separation of the data.
Another major hindrance with the k-means algorithm is that the "construction" phase can require many passes
through our tremendous data set to build the codebook. But this extra work is not necessary; the data has large
amounts of redundant information. We use a monte carlo method for passing through the data, and only sample
about 10 percent of the actual data.
Our overall clustering technique yields the same results as the k-means algorithm, but converges much faster.
Clustering times for a TM quarter scene (seven-dimensional data. 3000 rows by 3500 columns) of the Moscow and
Albuquerque areas are shown in Figure 2. These were calculated on a desktop SUN SPARCstation IPX with 16
MB of RAM. and show CPU time required for clustering the data into 256, 512, 1024, 2048, and 4096 clusters. It
is important to note that the execution time grows linearly as the number of clusters is increased. This is not a
property of the algorithm in general, but it has seemed to hold true for the vast majority of real-world multispectral
data sets (as well as most others) that the authors have encountered.
4. DATA ANALYSIS AND CLASSIFICATION
Once our TM scene has been clustered, it requires only one-seventh of the storage originally required, and the
new clustered representation provides an opportunity to use common computer displays very efficiently. Since there
are only N unique "vectors" representing the image, it takes on the order of N operations to manipulate the data as
compared to 12 million operations before the clustering was performed. Calculating the vegetation vigor of pixels
in a TM scene shows an example of the savings incurred by clustering. One measure of vegetation vigor commonly
used by remote sensing specialists is (Band 4 - Band 3) / (Band 4 + Band 3). This transformation results in large
values (bright pixels) for pixels representing healthy vegetation, and requires three operations at each pixel, or 36
million operations for the entire scene. If we first cluster the data to 256 clusters, we can use 8-bit computer displays
effectively. Since the clustered image contains only 256 unique values, 768 operations are required for calculating the
vegetation vigor, and the results can be directly mapped into the computer display look-up-tables (LUTs). While
this is a simple type of operation, the same holds true for very complicated transformations such as the Tasseled Cap
transformation, Karhunen-Loeve transformation, principal component analysis, etc.
Using a display package called SPECTRUM, developed by Los Alamos National Laboratory and the University
of New Mexico, we are able to use any desktop workstation running Unix and Xwindows to analyze and categorize
clustered data. Figure 3 shows a clustered TM scene of Moscow as displayed in SPECTRUM. A user can design and
manipulate a legend that specifies categories of land cover, labels for each category, and pseudocolor representations
to be used when categorizing geographic areas in the clustered image. SPECTRUM can manipulate the color map
for the computer display using any transformation of the clustered data, and can display cluster positions on a
two-dimensional scatter plot. Using these features, users are able to analyze data in a variety of ways. Data can
be categorized by selecting areas with a known type of land cover, causing all associated pixels in the image to
be given the same pseudocolor representation. Using the TM data, for example, a user could locate a wheat field,
highlight the pixels in that field, and all other wheat fields hi the entire image would be highlighted immediately.
After categorization, an image can be written out showing the different geographic areas for the scene.
Using the scatter plot, cluster positions can be displayed in a two-dimensional space with axes specified by
the user. Scientists can use this feature to interpret and categorize data by looking at different mathematical
transformations of the cluster positions, while results of the process are updated in the currently displayed clustered
image.
-------
MRLC Consortium
Documentation Notebook
January, 1994
Figure 3: Manipulating Moscow Data with SPECTRUM
5. ERROR ANALYSIS
To examine the accuracy of the clustering relative to the number of clusters used, we will look at the average
error per pixel introduced by the clustering, the distribution of these errors, and a Chi Square goodness-of-fit measure
for different land cover training areas.
An 800 x 800 subsection was extracted from the original 3000 X 3500 original image of Moscow and the 3000 X
3500 clustered version of the image. An error image was created by averaging, for the 7 spectral bands, the absolute
difference between the original image and the clustered image data. In the clustered image, each pixel is represented
by the mean vector of the cluster to which it is assigned. It should be noted that errors for each of the individual
bands is similar in magnitude and distribution to the average between the 7 spectral bands. The first plot in Figure
4 shows a plot of the average error per band per pixel and this error ± one standard deviation. The average error
for 256 clusters is less than 2 digital numbers (DN) and drops to less than 1.25 DN average error for 4096 clusters.
The maximum error over the subsection was much larger. There were a few popcorn clouds in the subsection and
the error for the center pixel in the clouds ranged from about 70 DN for the 256 clusters image to about 30 for the
4096 clusters image but these outliers in the data set were few and it is an easy process to isolate them as outliers
during the clustering process. The second plot in Figure 4 shows a histogram of the per pixel errors. The histograms
show that even for the 256 clusters image almost all the pixels have an error within ± 3 DN.
Finally, we chose three training sites for each of 4 land cover types in the 3000 x 3500 Moscow image representing
. soil, water, and forest. The training sites were located in the center of large uniform land covers and chosen
-------
MRLC Consortium
Documentation Notebook
January/ 1994
Error
Error Hi .too
•f Clu.tar.
Ab.oluta Xr
1 1U -t-r-f - 1 *U
•II
Figure 4: Per Pixel Errors for 800-by-800 Subsection of Moscow Scene
•••*•••• •* rtt Mt.atl.tle.
Crsai
i r.».t
Figure 5: Chi-Squared Goodness of Fit for 7 DOF
-------
MRLC Consortium
Documentation Notebook
January, 1994
if they were to be used in a traditional supervised classification. We then did a Chi Square goodness-of-fit test
to determine what our confidence was that the mean vectors representing the clustered data came from the same
process which generated the statistics from the training sets in the original data. The results are shown in Figure
5. A Chi Square test with 7 degrees of freedom has a value of less than 2.83 for greater than 90% confidence and a
value of less than 2.17 for a greater than 95% confidence. For an image with 4096 clusters all land covers had greater
than 95% confidence. For 256 clusters, the goodness-6f-fit values were much worse for the water training sets than
for other land covers. The training sets for water were extremely uniform with a variance in each spectral band of
less than 1.5. This means that even small differences between mean vectors yield large Chi Square values.
The errors introduced in a fine grain clustering of the multi-spectral data were not large enough to affect a level
one land use classification. With 4096 clusters, the clustered image could be used to effectively represent the original
data. Each land cover type was identified as easily as with the original image data.
6. CONCLUSIONS
Using a clustering method to do a preliminary classification of multispectral data provides data sets that can
be rapidly categorized hi an interactive fashion. A desktop workstation can be used to manipulate and analyze
the preprocessed data in real time. Unlike present uses of clustering, where scientists attempt to find relatively
small numbers of clusters in the data, our techniques define a large number of clusters to use. This data contains a
relatively small number of unique representative vectors that must be categorized, as compared to millions of pixels
in the raw data.
7. ACKNOWLEDGEMENTS
This work was performed under a U.S. Government contract (W-7405-ENG-36) by the Los Alamos National
Laboratory, which is operated by the University of California for the U.S. Department of Energy.
8. REFERENCES
[1] Paul M. Mather. Computer Processing of Remotely-Sensed Images. St. Edmundsbury Press Ltd., Bury St.
Edmunds, Suffolk, 1987.
[2] Robert A. Schowengerdt. Techniques for Image Processing and Classification in Remote Sensing. Academic
Press. New York. New York, 1983.
[3] R.O. Duda and P.E. Hart. Pattern Classification and Scene Analysis. Wiley, New York, NY, 1973.
[4] J.T. Tou and R.C. Gonzalez. Pattern Recognition Principles. Addison-Wesley, Reading, MA, 1974.
"51 A.K. Jain and R.C. Dubes. Algorithms for Clustering Data. Prentice Hall, Englewood Cliffs, NJ, 1988.
[6] Jerome H. Friedman, Jon Louis Bentley, and Raphael Ari Finkel. An algorithm for finding best matches in
logarithmic expected time. ACM Transactions on Mathematical Software, 3(3):209-226, 1977.
[7] J.L. Bentley. B.W. Weide. and A.C. Yao. Optimal expected-time algorithms for closest point problems. ACM
Transactions on Mathematical Software, 6:563-580, 1980.
[8] M.E. Hodgson. Reducing the computational requirements of the minimum-distance classifier. Remote Sensing of
Environment. 25:117-128. 1988.
'9] Jerome H. Friedman. Forest Baskett, and Leonard J. Shustek. An algorithm for finding nearest neighbors. IEEE
Transactions on Computers, pages 1000-1006, October 1975.
-------
MRLC Consortium
Documentation Notebook
January, 1994
APPENDIX E
-------
MRLC Consortium
Documentation Notebook
January, 1994
o
p co
r- 111
0 ft
LU
CC
CL
OC
P t D
DC
DC Q o
UJ _j in
> < Q.
O DC CO
° i3 CD
Q S ?
Q. CO
— CO D
--. H
LU
CO
-------
MKLC Consortium
Documentation Notebook
January, 1994
Khoros
"Software integration and development environment that emphasizes
information processing and data visualization."
X-windows image processing environment and system.
Contains programs to manipulate, enhance, and interpret images.
Maintains a programming environment to:
add new functionality
customize existing functions
proceduralize common tasks
store and retrieve records of complex processing sessions
System Size
363 Separate applications programs
Requires 220 Mbytes of disk storage for system
Written by John Rasure and students at University of New Mexico.
Copyright transferred to Khoral Research, Inc. in May 1993.
Open Software Package - Khoros can be used and modified only for
internal use in the organization obtaining it. The organization cannot
redistribute khoros unless the organization is a member of the Khoros
Consortium and has signed a redistribution license agreement.
Khoros Consortium - group of agencies and companies who fund
khoros development and maintenance. USGS has been a member.
Available through anonymous ftp over the Internet from site
ftp.eece.unm.edu (129.24.24.119).
Los Alamos programs available from this site (as the C3 Cluster
toolbox) or from Jim White at LANL (jwhite@lanl.gov)
Spectrum program available from this site (as the Classify toolbox)
-------
MRLC Consortium
Documentation Notebook
January, 1994
khorbs 2.0
Major upgrade of khoros image processing environment
In alpha release now (to members of khoros Consortium)
Beta release expected in mid-December
Public release scheduled for second quarter 1994
Active development on:
HP 9000/700 HP-UX 8.07
SGI Indigo OS 4.0.4
DEC Alpha OSF1.2
SUN SparcStations SUN OS 4.1.3 (Solaris 1.1)
SUN SparcStations SUN OS 5.1,5.3
(Solaris 2.1,2.3)
New Features
•Able to handle large images efficiently
•Removing reliance on Athena widget set - choose widget set at
compile time
•Image format more object oriented. Will recognize and deal with
non-Viff image formats
•Display programs will handle 16-bit images
•User can customize environment - select order within menus,
special help files, etc.
•Georeferencing information will be provided in the viff header
•Able to display irregular areas of interest (areas, points, polygons)
via the annotation layer. Eventually hope to have GIS file formats
directly supported by khoros (currently unfunded).
LANL will port construct, modify codebook, and project to khoros 2.0
UNM group estimates they could port khoros 2.0 to DG for about $20k
-------
MRLC Consortium
Documentation Notebook
January/ 1994
Executing Khoros programs
Khoros programs can be run in several different ways
• cantata - visual programming environment
-programs are selected through pull down menus
-placed on workspace as "glyphs"
-linked to transfer output from one glyph to the next
-executed singly or as a unit
-workspace can be stored and retrieved
Requires Xwindows execution
• Batch mode
-command line specification of all program options
-programs may be executed sequentially, but output cannot
be "piped" between programs
• Command-line prompts
-user is prompted for program options, including defaulted
items
• Xv routines
-program name is entered by user
-program runs with pull down menus and options
-requires Xwindows execution
-can also be run through cantata
-------
MRLC Consortium
Documentation Notebook
January, 1994
Hvper-Clusterina
•Execution Time
Projection is the longest step to execute
Affected by system load, amount of uniform area (background
in input, data volume (# pixels, # bands)
•Run over the same geographic area with different band
combinations:
Construct Project Total Seconds/
# Pixels # Bands Seconds Seconds Seconds Mbvte
8,073,000 6 1013 2688. 3701 80.126
8,073,000 5 833 2160 2993 77.750
8,073,000 6 1281 3623 4904 106.170
•Limitations
-Memory
Sufficient memory (or swapspace) to store full multispectral
image
Ames limitation of 64 Mbytes
-Disk
Temporary files may eat up available free space
Input data needs to be in viff format and interleaved
Procedure is: transform each input band into viff, then
combine separate bands into one multispectral dataset
Requires three separate stores of images to disk.
-------
MRLC Consortium
Documentation Notebook
January, 1994
-------
MRLC Consortium
Documentation Notebook
January, 1994
O)
••H
0)
o
CO
csfl
-
e u.
i-^
* r
Q. j->
U.
o
LU
I
LU
O
U
o
ICO
0)
CD
o
o
0)
4J
CD
o
0)
,1
«
'5
is
o
3
"S
CD
S
«>
£
O
l£
. "I
I 0-1
tn
CD
CD
ID
. »—i
!O
-------
MRLC Consortium
Documentation Notebook
January, 1994
4->
§
5
_< CJ
UJ
0)
«o
o
CD
-------
MRLC Consortium
Documentation Notebook
January, 1994
SPECTRUM
An interactive program to visually interpret land use / land cover from
classified multispectral images
Input: "Clustered" file written by project
Output: "Legend" file describing land cover units
"Image" file with header information to assign clusters to
the legend units
"Colormap" file of RGB values for land cover units
Interpretation is a visual process
Image is displayed
Interpreter outlines polygons of contiguous land cover
Clusters within that polygon can be:
-assigned to a new or existing unit
-ignored
-transferred from a current assigned unit to a new or different
one
Use of codebook statistics on cluster mean values (stored in the
image header) lets the program treat the classified image as through
it was still a multispectral image.
Can display different band combinations, functions of bands, or
transforms of bands.
Hardware requirements are simple:
Unix and Xwindows-compatible platform
8-bit color display
mouse and keyboard
-------
MRLC Consortium
Documentation Notebook
January, 1994
Spectrum Processing Files
Raw Image Single band per file. No header information.
VIFF-format Khoros image format. Band interleaved with header
Codebook
Binary file of per-band cluster means. Created by
the program "Construct" and modified by the
program "Modify Codebook" to include class 0.
Cluster Image Image file created by the program "Project"
Single-band, each pixel has a value from 0 to
maximum number of clusters. Header contains a
copy of the codebook file, modified to reflect the
pixels assigned to each class.
Input to Spectrum for land cover interpretation.
Clusout Image Image file created by Spectrum. Header contains
a "count column" indicating # of pixels in
each class.
If the image has been interpreted, the header
contains a "class column" indicating which classes
are assigned to each land cover unit.
Legend File File of land cover unit names and colors, created
by Spectrum.
Colormap File File of Red-Green-Blue color values used to display
each land cover unit. Created by Spectrum. Ascii text.
Image with
Colormap
Viff-format image file created by Spectrum. Header
contains a color map with the color assignments made
during interpretation. Pixel values range from 0 to
# of classes.
-------
MRLC Consortium
Documentation Notebook
January, 1994
Spectrum Output
To transfer interpretation back to khoros (or to another system) the
interpreted image is written out with its colormap as a standard viff format
file with colormap stored in the header
This file can be converted to a "raw" format file (no header) for transfer to
another system.
The color map can be written to an ascii file for transfer to another system.
Within khoros, the colormap from one viff image can be applied to another.
This transfers interpretation of one section of an image to another section.
-------
MRLC Consortium
Documentation Notebook
January, 1994
-------
MRLC Consortium
Documentation Notebook
January, 1994
} Inpu
'. Input File :
Output File:
it/Output
Khoros Image Files
I HELP 1 Closej
Inputs the 'clustered" rnage from project <*
<>
Legend files descrfce Land Cover Unts,
; Legend Files assigned to each, and colors assigned t<
I Input File :
Output File:
Output Image &
Output Colormap
>;g^^^*^^-'^w**^»>yww-*w^^*«>»x^«w
t,
1^
«*
«*
Colormap Output Saves the interpreted Land Cover Map
Colormap: ^<
Only:
i«"
!<*
lasses
iach.
The Spectrum Input/Output Window
-------
MRLC Consortium
Documentation Notebook
January, 1994
Control Display of Image
HELP I {Close
Type Of Normalize [ Local
When To Normalize
How To Normalize
Contrast Stretch Control
when Necessary
< n0rm < MaxColors .
Change Hap Columns Currently Displayed as Red, Green, Blue:
I GREEN
I BLUE | M2
Define Red, Green & Blue as Functions of map columns:
Controls for rnage band combinat
function display.
Can be changed at any stage of in
BLUE
File to View [^ Shows a text file (wth function parameters^
etatfen.
-------
MRLC Consortium
Documentation Notebook
January, 1994
s
5
£fi
.S>g
u £
Si
£-3
"I
S^
£
Cui
MJS|gs
w J.5^
Sfi «£^
Q£ta.l2 =S
-------
MRLC Consortium
Documentation Notebook
January, 1994
Legend (HELP
Color-space models
RGB n CMY PI HSV
Close]
Control for class and Land Cover Unit
HL?0l0Pl GREY
1 Clear Polygons frow Image | | ftbort Polygon Creation |
\ Delete Categonj(s) j
i Empty Categories) |
{ Catch-fill Categora |
\ Shou Selected Categories) j
I Hide Selected Cate9onj
-------
MRLC Consortium
Documentation Notebook
January, 1994
Examine
clusters in
spectral
space.
Add or
delete
clusters from
land cover
unite
-------
MRLC Consortium
Documentation Notebook
January, 1994
Tools for Interpretation
•Standard Legends
•Function and Transform Files
•Image Stratification Using ARC/INFO
•Image Stratification Using khoros Thresholding
•Use of Ancillary Data with Classifications
-------
MRLC Consortium
Documentation Notebook
January, 1994
Standard Legend Flies
Have written "empty" legend files for established/proposed
classification systems:
LUDA
new USGS
UNESCO/GAP
C-CAP
NALC
Legend File = list of land cover units and standard colors to be used
with them.
Using Standard Legend File:
Input File = Spectrum-created image file with "count" column
Input Legend = standardized legend file
Legend initially appears colorless
As clusters are assigned to each unit, standard colors appear
Function/Transform Files
Spectrum "Display Form" allows scrolling display of ascii file while
interpreting land cover classes
Function and transform equations can be input to
Band Display (red, green, or blue color guns)
Scatterplot
TM and MSS-specific equation files have been written
Equations are in "map column" form
-------
MRLC Consortium
Documentation Notebook
January/ 1994
Standard Legends
Anderson Level 1 Legend NALC Level 1 Legend
250 0 0 'Urban or built-up land1 1 220 0 220 '1.0 Developed Land1 1
200 150 0 'Agricultural land' 2 200 150 0 '2.0 Cultivated Land1 2
255 200 0 'Rangeland' 3 250 200 0 '3.0 Grassland
(herbaceous)' 3
0 200 88 'Forest land' 4 0 200 0 '4.0 Woody' 4
0 0 250 'Water1 5 200 200 200 '5.0 Exposed Land1 6
0 150 200 'Wetland' 6 255 255 255 S.O Snow and Ice1 5
200 200 200 'Barren land1 7 0 200 255 7.0 Wetland1 7
200 225 200 Tundra1 8 00 200 '8.0 Water and submerged
land1 8
255 255 255 'Perennial snow or ice' 9
USGS.new Level 1 Legend
220 0 220 'Developed Land' 1
200 150 0 'Cultivated Land' 2
250 200 0 'Grassland1 3
0 200 0 'Woody Land' 4
0 0 200 'Water1 5
0 200 255 'Wetland' 6
200 200 200 'Exposed Land' 7
200 225 200 Tundra1 8
255 255 255 'Snow and Ice1 9
-------
MRLC Consortium
Documentation Notebook
January, 1994
Function / Transform Files
Thematic Mapper Equations for Spectrum Display
NDVI : ((M3-M2) / (M3+M2))
TVI : (SQRT (((M3-M2) / (M3+M2)) +0.5) )
WATER-BODIES: ((M4 - Ml) / (M4 + Ml))
KAUTH-THOMAS (Tasseled Cap) TRANSFORM
Brightness
Greenness
Wetness
NormStress
NonnDiff
(MO* 3037)+(M1*.2793)+(M2*.4743)+(M3*.5585)+ (M4* .5082)+ (MS* .1863) )
(M0*(- 2848) )+ (Ml* (-. 2435) )+ (M2*(-. 5436) )+C6- ((MO*(0.071733)) + (Ml*(-0.833486)) + (M2*(0.521473)) +
-------
MRLC Consortium
Documentation Notebook
January, 1994
Imade Stratification Using ARC/INFO
Spatial stratification prior to clustering
One band of raw image data selected for strata delineation
Convert image to ARC "Image" format by:
Storing in ARC workspace with file extension .BIL
Create a .HDR file of # of rows, # of columns, pixel resolution,
and georeferencing information
Stratification Procedure:
Display image file in ARCEdit
Draw strata boundaries as arcs
Convert arcs to polygon coverage
Run POLYGRID
Run GRIDIMAGE and output file as BIL to form a strata image
Convert strata image to viff format
Turn into bit masks - within strata and outside strata
Apply each bit mask to each input band of multispectra! data
•multispectral within-strata image
•multispectral outside-strata image
Construct and Project can then be run separately on stratified images.
A similar procedure can be run for post-interpretation stratification of
a classification. The khoros colortable is converted to a .CLR file for
ARC display of the classification as an image.
-------
MKLC Consortium
Documentation Notebook
January, 1994
Image Stratification Using khoros
Spectral stratification
One band selected to best discriminate between desired strata
A strata reflectance DN "threshold" chosen
Bit masks created: within-strata and outside-strata
Each bit mask is applied to each band of multispectral input image
•multispectral within-strata image
•multispectral outside-strata image
Construct and Project can then be run separately on stratified images.
Choose Selection
oUDSanip l c
Threshold I
mmmmmmmmmmmmmmmfmmfmfmmt
Extract Sub Image
Dilation
Erosion
Median Filter
Invert Image
Print Image
Sun2VIff
Warp Image
Simple Uarp
HELP
QUIT
File Based Image thresholding utility*
Incut Inaoe
Output Image
U Threshold Level 128^ IPifttlft^^^^
uutput oata type»
• _
Byte
Q Bitmap
•Invert (False '
Execute Help
-------
MRLC Consortium
Documentation Notebook
January/ 1994
Use of Ancillary Data with Classifications (ARC/INFO)
Existing ARC/INFO datasets can be combined with images in the
hyper-clustering and interpretation process
•Before clustering, for stratification
•Before clustering, as an information band in the multispectral
image
•After interpretation, for clarification, plotting, selection by
feature
Classified Image to ARC/INFO
•Input as an ARC "Image" file
build .HDR file
convert khoros colormap to .CLR file
Allows image display
Allows vector overlay
•Convert to ARC "GRID" format using IMAGEGRID
Allows image display
Allows vector overlay
Allows value query and selection from GRID
-------
MRLC Consortium
Documentation Notebook
January, 1994
APPENDIX F
-------
MRLC Consortium
Documentation Notebook
January, 1994
APPLICATION OF
NOAA'S C-CAP PROTOCOL
IN THE
ELKHORN SLOUGH WATERSHED
NASA
NOAA'S COASTWATCH CHANGE ANALYSIS PROJECT
(C-CAP)
• Establishes a set of methodologies for monitoring
habitat and landuse change in coastal watershed
Identifies a standard classification scheme for
wetland and upland areas
• Establishes criteria for digital data processing,
interpretation, and publication
ru/\sA
-------
MRLC Consortium
Documentation Notebook
January, 1994
THE ELKHORN SLOUGH AND WATERSHED
> Elkhorn Slough if the main estuarine area of a
•lough complex. It covers approximately 1070
ha. The primary habitat is estuarine emergent
wetland.
1 Known problems in the slough are sedimentation,
erosion from tidal action, pesticide contamination
and nutrient loading.
' The watershed for the slough covers
approximately 582 *q. km. Primary land use is
grassland and irrigated row crops.
fUASA
LANDS AT THEMATIC MAPPER DATA
Scene Selection
• Limited to available imagery
Date of Acquisition
• June 20,1990
Precipitation Conditions
• Below normal (drought year)
TJdfi
• 3.7 feet above Mean Lower Low Water at
Moss Landing
ru/\s/\
-------
MRLC Consortium
Documentation Notebook
January, 1994
DATA PROCESSING AND CLASSIFICATION
Data Processing
• Geometrically corrected
• Rectified to UTM
• Extracted subset of TM scene
• Used 6 of 7 TM bands (no thermal)
Clustering
• Generated 240 clusters using SPECTRUM
(construct, mod.ebook, project)
unsupervised clustering algorthym.'
• Generated 150 clusters using ERDAS 7.5
unsupervised clustering algorthym.
NASA
DATA PROCESSING AND CLASSIFICATION (cent.)
Classification
• Classified the 240 clusters using SPECTRUM
interactive on-screen tools
• Classified the 150 clusters using ERDAS
IMAGINE on-screen tools
Special Technique
•Attempted 'cluster busting" procedure using
KHOROS and ERDAS
NASA
-------
MRLC Consortium
Documentation Notebook
January, 1994
USER OBSERVATIONS
Spectrum Advantages
• Fast clustering. Elkhom Watershed Image (807 rows
x 626 columns) took -1000 seconds total for
construct, mod.ebook, and project
• Extremely easy on«ecreen classification
•easy to select single cluster
• Interactive "Print Color Class* gives each band
response
• Interactive 'Scatter Plot* shows cluster and
classification pattern
NASA
USER OBSERVATIONS (cont)
Spectrum Disadvantage*
• No Documentation
• Not currently able to display clusters beyond 240
• No resizing of zoom window
• No display of map coordinates
• Random crashes
• Limited file-based manipulations
• Must export to KHOROS for image manipulations
(which might be o.k. if documentation were
available for KHOROS)
NASA
-------
MRLC Consortium
Documentation Notebook
January, 1994
USER OBSERVATIONS (cont)
Comparison of SPECTRUM & ERDAS
• For a novice, SPECTRUM is less time intensive for
classification than ERDAS.
• Clustering in both software packages resulted in
"confused clusters" (may be a result of the TM
data)
• If supplying classified images to a GIS is the goal,
data conversion is documented in ERDAS but not
documented in KHOROS.
NASA
-------
MRLC Consortium
Documentation Notebook
January, 1994
-------
MRLC Consortium
Documentation Notebook
January, 1994
APPENDIX. G
-------
MRLC Consortium
Documentation Notebook
January, 1994
Proposed Spectrum Enhancements
ADD MORE FUNCTIONS
1) Interface with ancillary data (GIS, particularly ARC/INFO)
2) Improved access to cluster statistics (printing, sorting, etc.)
Scatterplot Window
•Get info on cluster stats for individual clusters
•Click on point to get ID in margin
•Click on ID to get table of stats
Print Color Class Window
•Identify cluster number in this window
3) Interpret by strata
4) Color palette to pick colors from rather than sliders or RGB numbers
5) Display form:
•Change the default on "How to normalize" to +- 2 st. dev.
•Allow options for defaults in display (standard band combinations)
•Customize display options
6) Handle more than 255 classes in spectrum
7) Paint interpretation (like on a Mac)
8) Trackball" mapping in Display form to allow on-the-fly stretch
9) Assign ground coordinates to data (UTM)
ADD NEW OUTPUT CAPABILITIES
1) Output a "grouped" file - an image with as many classes as are
defined in the legend file. Pixel values are unit numbers from legend.
2) Write cluster statistics out to ASCII file
-------
MRLC Consortium
Documentation Notebook
January, 1994
3) Add list of clusters assigned to each unit to the legend file
ADD MORE INFORMATION / CHANGE PRESENTATION OF INFORMATION
1) Band identification
•Display (show up in "(descr)")
•Scatterplot (show up in "(descr)")
•Print color class (show up along margin)
2) Rename "Print color class" to "Spectral Response Curve"
3) Customize help files
4) Changes for scatterplot form:
•Use a white background with black and colored dots
•Bigger dots
•Different way (not orange) to show current class (change shape, outline,
flash, etc.)
BUG FIXES
1) Spectrum crashes when legend file and original file from project are read in.
2) Bug in legend subform with color sliders - color flashes on and off. Have to
set colors by typing in RGB values
3) Bugs seen only at Ames:
•Scatterplot: add/delete of clusters to and from units doesn't work
•Drawing polygons on main display (not Zoom window) will crash
program (sometimes throws you out of openwin)
•Spectrum won't run under cantata, have to start from csh prompt
4) Bugs when running on 24-bit monitor:
•Zoom-window - horizontal lines, not correct image -
•"Black" menus - they work, but you cant read them. Sub-panes are
correct
-------
MRLC Consortium
Documentation Notebook
January, 1994
APPENDIX H
-------
MRLC Consortium
Documentation Notebook
January, 1994
SCENE STATUS
As of 11/9/93 a total of 323 scenes have been ordered
1 scene acquired in 1990
80 scenes acquired in 1991
138 scenes acquired in 1992
104 scenes acquired in 1993
136 of these scenes are multi-temporal
187 are one time coverage
Of the 68 multi-temporal pairs ordered
58 are same year
10 are different years
Paul has prepared a list of 67 scenes proposed to order next. These
scenes represent mostly multi-temporal coverage where different
years were selected. The 67 scenes proposed for ordering represent a
preliminary selection of the "best" scene for each path/row pair.
150 previously purchased scenes have been identified and are avail-
able for use. Some of these scenes were ordered as geometrically /
terrain corrected / mosaic quality images from EOSAT. A decision
must be made as to how to handle these data where no TM-P level data
will be available.
-------
MRLC Consortium
Documentation Notebook
January, 1994
MRLC Landsat TM Geometric Processing Tasks
o Develop Terrain Correction Algorithms for TM-P Data in the Space
Oblique Mercator (SOM) Projection (Done)
- Software development in implementation/testing phase
o Address Variable Output Frame Processing Issues: (in work antici-
pating completion of analysis Dec. 1, some additional software inte-
gration may be required)
- Handle multiple output projections without changing product
accuracy
- Handle multiple output pixel sizes without changing product
accuracy (accuracy will be stated in meters)
- Handle both NAD27 and NAD83 datums
o Investigate the use of Landsat TM-P data in the UTM map projection.
SOM is prefered because image scan lines are very close to what
was originally imaged by the satellite. It is not known how far the
UTM products are from this original scan-line relationship. The ter-
rain correction process requires along-scan corrections. (Analysis
scheduled to begin Mid-December)
o Investigate the use of the GSFC TM Ground Control Point Library.
Slightly more than half the U.S. is covered, mainly in the east and
midwest. (Analysis will be begin after investigation of TM-P UTM
study)
-------
MRLC Consortium
Documentation Notebook
January, 1994
PRODUCT.PDF
Restore pre-processed image or clustered image to disk, all tape loca-
tion information will be available in the COMPLEX data set.
Restore control point file to current working directory
Prepare geometric correction grid
Resample either clustered image or orginal TM bands depending on
order requirements. If clustered data are being registered nearest
neighbor resampling will be used, if the orginal TM bands are being
registered cubic convolution resampling will be used.
The resampled image will be written to digital archive tape and the
COMPLEX data base updated with tape storage location.
A tape copy will be written and sent to the requesting agency.
Estimated cost of $0 /scene per the agreement that all consortium
members will receive one time complete coverage of the entire data
set. After each member has received a total of 630 scenes, cost will be
calculated on a per scene basis for reproduction.
-------
MRLC Consortium
Documentation Notebook
January, 1994
PREVIEW.PDF
(1) FSTFMTIN - Load the 6 TM bands
(2) DEBAND - Attenuate the banding pattern found in TM imagery
(3) Detector-to-Detector noise removal - Research is currently
being done to develop techniques to perform detector-to-detector
noise removal
(4) GPYRAMID - Down sample the image by a factor of two. GPYRA-
MID calculates the average value of a 4x4 pixel area and writes it to the
output image.
(5) GPYRAMID - Down sample the previously down-sampled image
by 2.
(6) FILTER_HIGH - Apply a high-pass filter to the down sampled
image created in step 5. above
(7) REDIST2 - Apply a contrast enhancement to the filtered image
(8) CONCAT - Combine the enhanced bands into a single output
image
(9) QLP_ADD - Write the preview image to the quick-look printer.
TOTAL COST = $114 - 300 depending on the results of the detector-to-
detector noise removal research and additional noise identified in the
scenes
-------
MRLC Consortium
Documentation Notebook
January, 1994
PROCESSING FLOW
PREVIEW THE QUICK-LOOK PRINT TO DETERMINE IF FURTHER
NOISE REMOVAL IS NECESSARY
ARCHIVE the cleaned image
PERFORM CLUSTERING - Clustering algorithm yet to be determined
ARCHIVE the clustered image
PICK CONTROL POINTS -
Image-to-Map - Control points will selected using the technique
determined to be most efficient and meet registration require-
ments i.e. 7.5-minute topographic maps, DLG's, or GCP libraries
Verification - A single band (normally band 2) will be registered
using the control points selected to the SOM projection. The reg-
istered band will be verified. All images failing to meet the
requirement of geometric accuracy of between -1 and +1 pixel
will be rejected and new control selected.
Image-to-image - If multi-temporal data has been requested for a
particular path/row all subsequent images will be registered to
the above registered image.
Verification - A single band will be registered and verified against
the reference image created above.
ARCHIVE CONTROL POINT FILES - The control point files will be writ-
ten to a special CTP.ARCHIVE directory. The approriate control point
information will be passed to Archive Management Section for inclu-
sion in the the COMPLEX data base.
CLEANUP and ARCHIVE WORKING DIRECTORY - Upon successful
registration all temporary files will be removed from the working direc-
tory and remaining information restored to tape.
-------
MRLC Consortium
umentation Notebook
January, 1994
-------
MRLC Consortium
Documentation Notebook
January, 1994
APPENDIX I
-------
PAT
10
12
12
13
13
13
14
14
14
14
14
14
14
14
14
14
14
15
15
15
15^^
15
15
15
15
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
17
17
1~ M
1 V
17
17
17
ROW
29
30
31
30
31
31
32
32
29
30
32
32
33
33
34
34
35
36
36
29
30
31
32
}32
33
34
34
37
37
30
31
32
32
33
33
34
34
35
35
36
36
37
38
38
39
30
31
tfl
C2
32
33
33
DATE
06/25/1991
09/04/1991
09/04/1991
09/27/1991
09/27/1991
10/06/1992
04/13/1992
09/20/1992
09/09/1991
09/09/1991
03/17/1991
05/20/1991
05/04/1991
09/09/1991
05/04/1991
09/09/1991
10/11/1991
06/23/1992
10/13/1992
06/14/1992
10/18/1991
06/14/1992
06/17/1993
10/20/1992
09/16/1991
04/11/1992
10/18/1991
02/04/1991
10/18/1991
05/20/1992
05/20/1992
06/24/1993
08/24/1992
03/01/1992
05/20/1992
03/01/1992
05/20/1992
03/01/1992
09/09/1992
03/01/1992
05/04/1992
11/26/1991
05/07/1993
11/26/1991
05/07/1993
10/02/1992
05/11/1992
10/02/1992
05/14/1993
10/02/1992
07/17/1993
10/02/1992
C
1
0
0
1
0
0
0
0
1
1
0
0
1
0
1
0
0
1
0
1
0
0
0
1
0
0
0
1
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
0
0
0
0
Q
9
9
9
9
9
9
9
9
8
8
9
9
9
8
9
8
8
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
MRLC Consortium
COMMENTS Documentation Notebook
__ January ,.-1334
no hard copy-need to see it-not much up here -
checked the microfiche looks ok
clouds in quads a&c-
slight clouds in a&c
mike wants this one- looks good
m wants this one - looks great
mike wants this one - looks good
bit of clouds in quads a & b
eosat ? pick-cloudy & pixel noise-we dont want it
m wants this one - okf some snow in ql
only picked one-others looked unusable
no quality rating, no image - eosat sent image looks
good
there is no quality rating, so cannot look at image
eosat sent image - looks good-per m try 4/23/93
a few clouds mostly in Canada - only 1 good pick her
had a look at it on fiche - only one i could find
that was good
probably only want one here considering available
choices
very slight clouds in q3-good scene!
try to get another one
only one good pick here
need to verify the quality of this one
mike wants to try three here
there is some snow
picked from the images good early summer
scan line defects visable on image maybe not on cct
bit of clouds -not bad
looks good on mss microfiche - had judy collins chec
qual and quick look - looks good
looks good
m didn't want this one-best choice we have
minor clou in quad b
good looking scene
-------
17
17
17
17
17
17
17
17
17
17
18
18
18
18
18
18
18
18
18
18
18
18
18
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
34
34
35
35
36
36
37
37
38
39
31
31
32
33
35
35
36
37
37
37
37
38
38
31
32
32
33
34
34
35
36
36
37
37
38
38
39
39
39
29
30
30
31
32
32
33
33
34
34
35
36
36
37
37
38
38
39
05/11/1992 1 9
10/02/1992 0 9
05/11/1992 0 9
11/03/1992 0 9
05/11/1992 0 9
11/03/1992 0 9
04/28/1993 0 9
06/20/1992 0 9
06/20/1992 0 9
10/16/1991 0 9
08/09/1993 0 9
10/23/1991 0 9
08/06/1992 0 9
08/06/1992 1 9
06/06/1993 0 9
10/25/1992 0 9
04/19/1993 0 9
05/18/1992 0 9
06/06/93 0 9
10/12/93 0 9
12/10/1991 0 8
04/16/1992 0 9
12/10/1991 0 8
04/07/1992 1 9
04/07/1992 1 9
08/11/1991 1 9
09/30/1992 1 9.
04/23/1992 0 9
08/11/1991 1 9
09/30/1992 0 9
07/12/1992 0 9
12/17/1991 0 9
07/31/93 0 9
11/17/1992 0 9
04/10/1993 0 9
10/14/1991 0 9
04/10/1993 0 9
07/31/1993 0 8
10/03/93 0 9
07/22/1993 0 9
05/16/1992 1 9
09/05/1992 0 9
05/16/1992 0 9
05/16/1992 0 9
10/21/1991 0 9
08/02/1991 0 9
10/21/1991 0 9
02/10/1992 0 9
08/02/1991 1 9
11/06/1991 0 9
01/25/1992 0 9
07/22/1993 0 9
01/25/1992 0 9
07/22/1993 0 9
02/10/1992 0 9
07/22/1993 0 9
07/01/1991 1 9
MRLC Consortium
a few clouds in all quads-metuiteafiatearii ^dawbyep
January, 1994
looks good on core
looks good on mss microfiche
clouds in quad d
good scene
good looking scene
some haze in gl
- one cloud in bottom center
gail wants this one-per eosat looks good
gail pick-looks pretty good-bit of clouds
looked very good as an image
checked it out-not bad cl in qu 2
look at this one-adequate-but clouds in q2
bit of clouds quads a c d
looked at all images-this is the best but some cloud
picked from images-good scene
some clouds in all quads not the best scene ever
bus the only summer scene available
good looking scene
gail pick-looks good per eosat
mike wants this one - looks good
good scene
good image
had jim n at eosat look at it-no clouds, acouple of
scan line drops upper portion of scene
gail pick-looks good per eosat
good one
look at this one - good scene
good looking scene
picked from visual look at images,also there is a
goodlO/24/92 scene not on list checkitout
check this one out - looks good very slight clouds
no quality rating - eosat sent image - looks good
check for snow
need qual rating - eosat sent image - looks good
look for summer of this p/r
good scene a few small clouds
see what this looks like-looks very good
some clouds on land area in quad b
-------
20
21
21
^^^k
~A
V
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
21
22
22
22
22
22
2^^^
' ^^^^^H
2*
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
23
23
23
23
23
23^
2'A
2 ^
23
23
23
39
28
29
29
30
30
31
31
32
32
33
34
34
35
36
36
37
37
38
38
39
39
40
28
29
29
30
30
31
'31
32
32
33
33
34
34
35
35
36
37
37
38
38
39
39
40
28
29
30
31
31
32
p2
33
34
34
11/06/1991
06/11/1993
05/10/1993
08/25/1991
05/10/1993
08/14/93
05/10/1993
08/30/93
05/10/1993
08/25/1991
09/12/1992
04/05/1992
10/14/1992
07/10/1992
01/16/1992
09/12/1992
02/03/1993
09/26/1991
01/16/1992
10/12/1991
05/07/1992
10/14/1992
10/12/1991
08/21/1993
08/21/1993
10/05/1992
07/01/1992
10/05/1992
07/01/1992
10/05/1992
07/20/1993
10/05/1992
07/20/1993
10/05/1992
06/18/1993
08/16/1991
06/18/1993
11/04/1991
03/09/1991
01/25/1993
10/21/1992
01/25/1993
10/21/1992
01/25/1993
10/05/1992
10/05/1992
05/05/1992
05/05/1992
05/05/1992
04/17/1991
10/12/1992
06/09/1993
10/12/1992
04/22/1993
10/12/1992
07/06/1991
10/10/1991
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
MRLC Consortium
looks ok Documentation Notebook
January, 1994
minor clouds in al
looked at all images of area- this one is best
good scene
gail pick-looks ok per eosat
gail pick - bit of clouds but ok
good no clouds
look at this one per mike-pretty good slight cl
looks better - no cloud-lets go with 91 scene-very
little difference
slight popcorn scattered about
not a bad scene bit of clouds on east edge
slight clouds in quad a
looked at all images these 2 look best
good scene
extension of miss delta in quad a-rest is mostly h2o
some clouds in c&d not real bad
very slight clouds in q4-s edge
a couple of scan line defects are apparent
good scene
slight clouds in quad b
good scene
bit of clouds in all quads-b more than others
check this one - get it if good- its good
fws - shifted south 50% - full scene looks good
good one
fws - shifted south 50% - full scene looks good
good scene
look at this one if ok change-its good
looks good on image-nothing else we saw is good
at all
picked from viewing all images of area april looks
interesting here
m wanted to switch-per eosat it looks good-10/28pas
good scene
slight popcorn in q2
slight clouds quad a
maybe a bit of haze in a&b
-------
MRLC Consortium
23
23
23
23
23
23
23
23
23
23
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
14
25
25
25
25
25
25
15
IS
15
15
!5
!5
:5
5
5
5
5
5
5
5
5
5
5
5
6
6
35
35
36
36
37
37
38
38
39
39
27
28
28
29
29
30
30
31
31
32
32
33
33
34
34
35
38
39
27
27
28
29
29
30
30
31
31
32
32
33
33
34
34
36
36
37
38
38
39
39
40
40
26
27
04/22/1993
07/27/1993
04/22/1993
07/27/1993
04/22/1993
07/24/1992
04/22/1993
10/12/1992
04/01/1991
10/12/1992
07/13/1991
04/24/1991
07/31/1992
04/24/1991
07/31/1992
04/24/1991
07/31/1992
04/24/1991
10/17/1991
05/15/1993
10/17/1991
08/14/1991
10/17/1991
04/10/1992
10/03/1992
06/16/1993
10/19/1992
10/19/1992
05/06/1993
09/24/1992
05/06/1993
05/19/1992
09/08/1992
05/19/1992
09/08/1992
05/03/1992
09/08/1992
05/03/1992
09/24/1992
05/03/1992
09/24/1992
05/03/1992
09/24/1992
07/25/1993
10/05/1990
07/25/1993
02/10/1991
07/06/1992
02/10/1991
09/22/1991
02/10/1991
07/06/1992
05/10/1992
05/10/1992
0
0
0
0
0
0
0
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
9
9
9
8
9
9
9
9
9
9
9
8
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
9
9
9
9
gail pick-looks ok per
good scene
few clouds in g4
good looking scene
few clouds in quad d
great image!
cannot look at it - checked microfiche-its fine
a few clouds on land area-not too bad
checked microfiche it looks fine
a few clouds scattered not bad acouple of bad lines
another one that we could not see, - checked
microfiche - it looks fine
little bit of clouds central area
couldn't see this one-no quick look- looked at it
on microfiche - its ok
perhaps a couple of bad lines lower half
was unable to see this image - microfiche of it
it looks fine
mike wanted to switch-had eosat look its ok-10/28pas
good scene-little bit of haze g4
some clouds in quad d - mike wants this path row don
first
good scene
see if we have enough land area to be worth it -it
would be worth it
few clouds q34 not very bad
see what it looks like-looks good
would realy like to see this one -can't - checked
mss fiche - good scene
cannot see image - looked at mss fiche - good
good scene
check this one out-looks really good
fairly good size cloud in quad b
some clouds in quad c-not too bad
look at it - looks great
look at this - looks great!
-------
MRLC Consortium
26 27 08/12/1991 0 9
05/13/1993 0 9
10/01/1992 0 9
05/13/1993 0 9
10/01/1992 0 9
05/13/1993 0 9
08/30/1992 0 9
10/01/1992 0 9
03/23/1992 0 9
10/01/1992 0 9
04/06/1991 0 9
09/13/1991 0 9
03/23/1992 0 9
08/14/1992 0 9
07/13/1992 0 9
03/07/1992 0 9
07/13/1992 0 9
03/07/1992 0 9
07/13/1992 0 9
02/01/1991 0 9
05/13/93 0 9
08/17/1993 0 9
01/21/1993 0 9
11/02/1992 0 9
01/21/1993 0 9
10/01/1992 0 9
03/10/1993 0 9
11/02/1992 0 9
04/11/1993 0 9
11/02/1992 0 9
05/01/1992 0 9
06/16/1991 1 9
06/16/1991 0 8
09/22/1992 0 9
27 29 06/16/1991 0 7
27 29 09/20/1991 0 9
27 30 05/01/1992 0 9
27 30 09/22/1992 0 9
27 31 05/01/1992 0 9
27 31 09/04/1991 0 9
27 32 05/01/1992 0 9
27 32 08/21/1992 0 9
27 33 03/14/1992 0 9
27 33 08/21/1992 0 9
27 34 03/14/1992 0 9
27 34 09/22/1992 0 9
27 35 03/14/1992 0 9
27 35 08/21/1992 0 9
27 36 03/14/1992 0 9
27 36 08/21/1992 0 9
27^37 03/14/1992 0 9
2 ^^37 08/21/1992 0 9
2 ^^38 03/14/1992 0 9
27 38 08/05/1992 1 9
27 39 02/08/1991 0 9
27 39 10/24/1992 0 9
26^
^^
^^r
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
26
Zv,
27
27
27
27
28
28
29
29
30
30
31
32
32
33
33
34
34
35
36
36
37
37
38
38
38
39
39
40
40
41
41
42
42
27
27
28
28
some clouds but mostly ove^tfiF^tP11- Nf )gi at
ntSS fiche January, 1994
good scene
look at this one - looks great
good scene
look at it-looks good
good scene
picked from images good late aug agriculture
change to this one of ok - its fine
change to this if ok-this one is good
clouds in quad b-about half of the quad
cannot see image - mss fiche - good scene
picked from visual look at all available images
looked at mss fiche - good scene
scattered popcorn
eosat sent the image - it looks good
gail pick-bit of clouds but ok
good scene if date is ok
missing quick look-looks great on core
picked visually
the only other one i like here is the 10/1 scene & i
appears that there would be very little difference
look at it- looks pretty good
looked at fiche - good scene
need to see the image - looked at tm fiche - good
only good looking scene avail to see 91 scenes are
quality 8
this one has a qu rating of 7, but it looks very goo
on quick look and the date is just right-lets try it
look at this - good one
see if there is a 6/2 scene, if so it might be bette
bit of popcorn in quad d
this looks good
bit of popcorn in quad c
bit of popcorn in quad a&b
a few clouds but pretty good summer scene
not much choice here
-------
41
41
42
26
26
27
27
28
28
29
29
30
30
31
31
32
32
33
34
35
35
36
37
38
39
39
40
41
26
26
27
28
28
29
29
30
30
31
32
33
34
35
36
37
38
39
40
26
26
27
27
28
28
29
30
31
32
33
34
35
04/18/1993
07/23/1993
07/23/1993
05/11/1993
08/10/1991
05/11/1993
07/30/1993
05/08/1992
08/26/1991
05/08/1992
08/26/1991
05/08/1992
08/26/1991
05/27/1993
08/26/1991
04/09/1993
08/26/1991
04/09/1993
07/30/1993
03/08/1993
07/09/1991
05/08/1992
08/15/1993
05/11/1993
03/08/1993
08/15/1993
05/11/1993
08/10/1991
05/15/1992
09/04/1992
08/17/1991
05/15/1992
08/17/1991
05/15/1992
08/19/1992
08/19/92
10/22/1992
08/19/92
08/19/92
08/01/1991
08/01/1991
08/01/1991
08/22/1993
08/22/1993
08/22/1993
09/04/1992
08/06/1993
05/06/1992
07/23/1991
05/06/1992
08/10/1992
05/25/1993
08/10/1992
05/25/1993
07/28/1993
07/28/1993
07/28/1993
07/25/1992
07/28/1993
07/28/1993
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
1
1
0
0
0
0
0
0
0
0
0
0
0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
MRLC Consortium
slight clouds in corner of alimentation Notebook
good scene
y'
good scene
eosat screened scene & provided an image - looks g
-------
31
31
31
31
31
31
31
31
31
31
31
31
31
32
32
32
32
32
32
32
32
32
33
33
33
33
33
34
34
34
34
34
34
34
35
35
35
35
35
35
35
36
36
36
36
^>
36
36
36
36 08/26/1992 1 9
37 08/13/1993 0 9
38 07/28/1993 0 9
39 07/28/1993 0 9
40 07/28/1993 0 9
26 07/14/1991 1 9
27 06/12/1991 0 9
28 06/12/1991 0 9
29 07/14/1991 0 9
30 07/14/1991 0 9
31 07/14/1991 0 9
32 07/14/1991 0 9
33 07/30/1991 0 9
34 07/30/1991 0 9
35 06/12/1991 0 9
38 04/27/1992 0 9
39 04/27/1992 0 9
40 04/27/1992 0 9
26 07/23/1992 1 9
27 08/08/1992 1 9
28 08/27/1993 1 9
29 09/09/1992 0 9
30 08/06/1991 1 9
31 08/06/1991 1 9
32 08/06/1991 1 9
33 08/06/1991 1 9
38 03/31/1991 0 9
39 03/31/1991 0 9
26 07/17/1993 0 9
27 05/27/1992 0 9
28 08/18/1993 0 9
29 08/15/1992 0 9
30 08/15/1992 0 9
31 08/15/1992 0 9
33 07/06/1992 0 9
26 08/09/1993 0 9
27 09/23/1992 0 9
28 08/09/1993 0 9
30 08/09/1993 0 9
31 09/21/1991 0 9
32 09/23/1992 0 9
33 09/07/1992 0 9
29 08/16/1993 0 9
30 08/16/1993 0 9
31 08/16/1993 0 9
32 08/11/1991 0 9
33 06/18/1992 0 9
37 06/13/1993 0 9
38 06/13/1993 0 9
26 08/07/1993 0 9
27 08/07/1993 0 9
30 08/23/1993 0 9
31 08/23/1993 0 9
32 08/23/1993 0 9
'. 33 08/23/1993 0 9
34 08/23/1993 0 9
35 08/23/1993 0 9
36 08/23/1993 0 9
MRLC Consortium
some clouds through quads ^
-------
36
37
37
37
37
37
37
37
37
37
37
38
38
38
38
38
38
38
38
39
39
39
39
39
39
39
39
40
40
40
40
40
40
40
40
41
41
il
11
11
11
12
12
12
.2
:2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
38
26
30
31
32
33
34
35
36
37
38
30
31
32
33
34
35
36
37
29
31
32
33
34
35
36
37
28
31
32
33
34
35
36
37
31
32
33
34
35
36
28
31
32
33
33
33
34
34
35
35
36
27
29
30
31
32
33
34
34
05/14/1991
08/14/1993
08/27/1992
08/14/1993
08/14/1993
06/27/1993
06/22/1991
06/22/1991
06/22/1991
06/22/1991
06/22/1991
07/20/1993
07/20/1993
07/17/1992
07/20/1993
06/15/1992
06/15/1992
07/01/1992
05/28/1991
07/24/1992
07/27/1993
07/27/1993
07/27/1993
07/27/1993
07/27/1993
07/27/1993
07/27/1993
08/19/1993
08/19/1993
08/19/1993
08/19/1993
08/19/1993
08/19/1993
08/19/1993
04/26/1992
08/26/1993
08/26/1993
08/26/1993
08/10/1993
08/10/1993
08/26/1993
06/14/1993
06/14/1993
06/14/1993
05/13/93
07/16/1993
08/01/93
05/13/93
07/16/1993
07/16/1993
08/17/93
05/29/1993
08/08/1993
07/02/1991
08/08/1993
08/08/1993
08/08/1993
08/24/1993
05/01/1992
07/20/1992
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
MRLC Consortium
Documentation Notebook
January, 1994
looked at image-great looking scene
good scene slight cloudsl
good scene-very slight clouds
looks good
see what this looks like on this date-looks ok
good scene-slight clouds on east edge
a few clouds more in quad a-but not too bad
good scene
a few clouds in q2-not bad
looks good either this or the 6/13/92 are fine
looked at it on core-good scene
very slight clouds
good scene - better of two here in 93
good looking scene-minimal clouds
slight clouds
gail pick -per eosat looks ok
gail pick - per eosat looks ok
gail pick - per eosat looks ok
ice in high mnts
good scene
gail pick - per eosat-looks ok
looks good
looks good
good scene-bit of ice corner of q2=
mike wants these two over the area-looks good
-------
43
43
44
44
44
44
44
44
44
44
44
45
45
45
46
46
46
46
46
46
46
47
47
47
48
35
35
29
30
31
31
32
32
33
33
34
34
27
32
33
28
29
29
30
31
32
32
26
27
30
27
05/01/1992
07/20/1992
06/09/1992
10/13/1991
06/12/1993
10/13/1991
06/12/1993
07/30/93
06/09/1992
10/15/1992
01/03/1993
06/12/1993
09/18/1991
08/06/93
08/22/93
08/29/93
05/06/1992
08/29/93
05/06/1992
10/11/1991
07/12/1993
10/11/1991
08/04/1993
09/16/1991
07/30/1991
09/09/1992
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
1
0
0
0
0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
9
9
9
9
9
looks like some haze in qu
^
clouds around monterey bay
MRLC Consortium
tion Notebook
January, 1994
we were looking at June-its cloudy-this oct scene
looks really good
looked at all images, oct is away from desired seaso
but best looking scene available
do we want this on with the 8/30/93?
gail pick-per eosat looks ok
mike wants this one-some haze-not bad
mike wants this one-again some clouds-not terrible
January but really clear over san francisco-thats
rare
this one is good
gail pick - per eosat looks ok
gail pick - per eosat looks ok
gail pick - per eosat looks ok
do we want this one with the 8/29/93?
gail pick - per eosat looks ok
good scene -no clouds at all
little bit of clouds in quad d
good scene
we have one-look at this anyway-looks very good
check out this one-looks great
-------
MRLC Consortium
Documentation Notebook
January, 1994
APPENDIX J
-------
PAT ROW DATE
C Q ORDER DA
10
j.2
12
13
13
13
14
14
14
14
14
14
14
14
14
14
14
15
15
15
15
15
15
15
15
15^^
^B
^^
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
17
17
17
17
17
17
17
17
i -^^
w
I/
17
17
29
30
31
30
31
31
32
32
29
30
32
32
33
33
34
34
35
36
36
29
30
31
32
32
33
34
34
37
37
30
31
32
32
33
33
34
34
35
35
36
36
37
38
38
39
30
31
31
32
32
33
33
34
k34
'35
35
36
36
06/25/1991
09/04/1991
09/04/1991
09/27/1991
09/27/1991
10/06/1992
04/13/1992
09/20/1992
09/09/1991
09/09/1991
03/17/1991
05/20/1991
05/04/1991
09/09/1991
05/04/1991
09/09/1991
10/11/1991
06/23/1992
10/13/1992
06/14/1992
10/18/1991
06/14/1992
06/17/1993
10/20/1992
09/16/1991
04/11/1992
10/18/1991
02/04/1991
10/18/1991
05/20/1992
05/20/1992
06/24/1993
08/24/1992
03/01/1992
05/20/1992
03/01/1992
05/20/1992
03/01/1992
09/09/1992
03/01/1992
05/04/1992
11/26/1991
05/07/1993
11/26/1991
05/07/1993
10/02/1992
05/11/1992
10/02/1992
05/14/1993
10/02/1992
07/17/1993
10/02/1992
05/11/1992
10/02/1992
05/11/1992
11/03/1992
05/11/1992
11/03/1992
1
0
0
1
0
0
0
0
1
1
0
0
1
0
1
0
0
1
0
1
0
0
0
1
0
0
0
1
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
0
0
0
9
9
9
9
9
9
9
9
8
8
9
9
9
8
9
8
8
9
9
9
9
9
9
9
9
9'
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
1109
1109
1109
1109
1109
1015
1109
1109
1015
1015
1015
1015
1015
1015
1109
1109
1109
1109
1109
1015
1015
1015
1015
1015
1015
1015
1015
1109
1109
1015
1109
P
q
1109
1109
P
P
1109
1109
1109
1109
1109
q
q
1109
1109
1109
1109
P
q
p
q
1109
1109
1109
1109
1109
1109
MRLC Consortium
Documentation Notebook
January, 1994
-------
1
1
37 04/28/1993
37 06/20/1992
38 06/20/1992
39 10/16/1991
31 08/09/1993
31 10/23/1991
32 08/06/1992
33 08/06/1992
35 06/06/1993
35 10/25/1992
36 04/19/1993
37 05/18/1992
37 06/06/93
37 10/12/93
37 12/10/1991
38 04/16/1992
38 12/10/1991
31 04/07/1992
32 04/07/1992
32 08/11/1991
33 09/30/1992
34 04/23/1992
34 08/11/1991
35 09/30/1992
36 07/12/1992
36 12/17/1991
37 07/31/93
37 11/17/1992
38 04/10/1993
38 10/14/1991
39 04/10/1993
39 07/31/1993
39 10/03/93
29 07/22/1993
30 05/16/1992
30 09/05/1992
31 05/16/1992
32 05/16/1992
32 10/21/1991
33 08/02/1991
33 10/21/1991
34 02/10/1992
34 08/02/1991
35 11/06/1991
36 01/25/1992
36 07/22/1993
37 01/25/1992
37 07/22/1993
38 02/10/1992
38 07/22/1993
39 07/01/1991
39 11/06/1991
28 06/11/1993
29 05/10/1993
29 08/25/1991
30 05/10/1993
30 08/14/93
31 05/10/1993
31 08/30/93
32 05/10/1993
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
1
1
1
1
0
1
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
1
0
0
0
1
0
0
0
0
0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
9
8
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
q
q
1109
1109
P
q
1109
1109
p
q
1109
q
p
p
q
p
q
1109
q
p
1109
q
p
1109
p
q
p
q
p
q
p
q
p
1109
1109
1109
1109
p
q
1109
1109
q
p
1109
q
p
q
p
q
p
1109
1109
1109
p
q
1109
p-
p
p
1015
MRLC Consortium
Documentation Notebook
January, 1994
-------
21
21
21
2^W
^B
^L
21
21
21
21
21
21
21
21
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
22
~A
w
22
22
22
22
22
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
23
2"A
2 V
2j
24
24
32
33
34
34
35
36
36
37
37
38
38
39
39
40
28
29
29
30
30
31
31
32
32
33
33
34
34
35
35
36
37
37
38
38
39
39
40
28
29
30
31
31
32
32
33
33
34
34
35
35
36
36
37
37
38
38
39
39
27
28
08/25/1991
09/12/1992
04/05/1992
10/14/1992
07/10/1992
01/16/1992
09/12/1992
02/03/1993
09/26/1991
01/16/1992
10/12/1991
05/07/1992
10/14/1992
10/12/1991
08/21/1993
08/21/1993
10/05/1992
07/01/1992
10/05/1992
07/01/1992
10/05/1992
07/20/1993
10/05/1992
07/20/1993
10/05/1992
06/18/1993
08/16/1991
06/18/1993
11/04/1991
03/09/1991
01/25/1993
10/21/1992
01/25/1993
10/21/1992
01/25/1993
10/05/1992
10/05/1992
05/05/1992
05/05/1992
05/05/1992
04/17/1991
10/12/1992
06/09/1993
10/12/1992
04/22/1993
10/12/1992
07/06/1991
10/10/1991
04/22/1993
07/27/1993
04/22/1993
07/27/1993
04/22/1993
07/24/1992
04/22/1993
10/12/1992
04/01/1991
10/12/1992
07/13/1991
04/24/1991
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
9
9
9
9
9
9
9
8
1015
1109
1109
1109
1109
1109
1109
q
q
q
p
1109
1109
1109
1109
1015
1015
1109
1109
1109
1109
q
p
1015
1015
P
q
p
q
1109
q
q
q
q
q
q
1109
1109
1109
1109
q
p
q
p
q
p
1109
1109
1109
1015
1015
1015
1015
1015
P
q
p
q
1109
q
KRLC Consortium
Documentation Notebook
January, 1994
-------
28 07/31/1992
29 04/24/1991
29 07/31/1992
30 04/24/1991
30 07/31/1992
31 04/24/1991
31 10/17/1991
32 05/15/1993
32 10/17/1991
33 08/14/1991
33 10/17/1991
34 04/10/1992
34 10/03/1992
35 06/16/1993
38 10/19/1992
39 10/19/1992
27 05/06/1993
27 09/24/1992
28 05/06/1993
29 05/19/1992
29 09/08/1992
30 05/19/1992
30 09/08/1992
31 05/03/1992
31 09/08/1992
32 05/03/1992
32 09/24/1992
33 05/03/1992
33 09/24/1992
34 05/03/1992
34 09/24/1992
36 07/25/1993
36 10/05/1990
37 07/25/1993
38 02/10/1991
38 07/06/1992
39 02/10/1991
39 09/22/1991
40 02/10/1991
40 07/06/1992
26 05/10/1992
27 05/10/1992
27 08/12/1991
28 05/13/1993
28 10/01/1992
29 05/13/1993
29 10/01/1992
30 05/13/1993
30 08/30/1992
31 10/01/1992
32 03/23/1992
32 10/01/1992
33 04/06/1991
33 09/13/1991
34 03/23/1992
34 08/14/1992
35 07/13/1992
36 03/07/1992
36 07/13/1992
37 03/07/1992
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
P
q
P
q
P
1109
1109
q
P
1109
1109
1015
1015
1015
1109
1109
q
P
1015
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1015
1015
1015
1015
1015
q
q
1109
1109
q
q
1109
q
p
1015
1015
q
p
q
p
p-
1109
1109
1109
1109
1015
1015
1109
1109
1109
1109
MRLC Consortium
Documentation Notebook
January, 1994
-------
26
26
26
^^
^B
^0
26
26
26
26
26
26
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
27
' ^^k
^B
27
27
27
27
27
27
27
27
27
28
28
28
28
28
28
28
28
28
28
28
28
28
28
2P^^
: 9
2o
28
28
37
38
38
38
39
39
40
40
41
41
42
42
27
27
28
28
29
29
30
30
31
31
32
32
33
33
34
34
35
35
36
36
37
37
38
38
39
39
41
41
42
26
26
27
27
28
28
29
29
30
30
31
31
32
32
33
34
35
35
36
07/13/1992
02/01/1991
05/13/93
08/17/1993
01/21/1993
11/02/1992
01/21/1993
10/01/1992
03/10/1993
11/02/1992
04/11/1993
11/02/1992
05/01/1992
06/16/1991
06/16/1991
09/22/1992
06/16/1991
09/20/1991
05/01/1992
09/22/1992
05/01/1992
09/04/1991
05/01/1992
08/21/1992
03/14/1992
08/21/1992
03/14/1992
09/22/1992
03/14/1992
08/21/1992
03/14/1992
08/21/1992
03/14/1992
08/21/1992
03/14/1992
08/05/1992
02/08/1991
10/24/1992
04/18/1993
07/23/1993
07/23/1993
05/11/1993
08/10/1991
05/11/1993
07/30/1993
05/08/1992
08/26/1991
05/08/1992
08/26/1991
05/08/1992
08/26/1991
05/27/1993
08/26/1991
04/09/1993
08/26/1991
04/09/1993
07/30/1993
03/08/1993
07/09/1991
05/08/1992
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
9
7
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
1109
q
p
p
q
q
q
q
q
q
q
q
q
P
p
q
1109
1109
1109
1109
q
p
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
q
q
1109
1109
1109
1015
1015
1109
1109
q
p
q
p
q
p
q
p
q
p
1109
1109
q
p
1109
MRLC Consortium
Documentation Notebook
January, 1994
-------
37 08/15/1993
38 05/11/1993
39 03/08/1993
39 08/15/1993
40 05/11/1993
41 08/10/1991
26 05/15/1992
26 09/04/1992
27 08/17/1991
28 05/15/1992
28 08/17/1991
29 05/15/1992
29 08/19/1992
30 08/19/92
30 10/22/1992
31 08/19/92
32 08/19/92
33 08/01/1991
34 08/01/1991
35 08/01/1991
36 08/22/1993
37 08/22/1993
38 08/22/1993
39 09/04/1992
40 08/06/1993
26 05/06/1992
26 07/23/1991
27 05/06/1992
27 08/10/1992
28 05/25/1993
28 08/10/1992
29 05/25/1993
30 07/28/1993
31 07/28/1993
32 07/28/1993
33 07/25/1992
34 07/28/1993
35 07/28/1993
36 08/26/1992
37 08/13/1993
38 07/28/1993
39 07/28/1993
40 07/28/1993
26 07/14/1991
27 06/12/1991
28 06/12/1991
29 07/14/1991
30 07/14/1991
31 07/14/1991
32 07/14/1991
33 07/30/1991
34 07/30/1991
35 06/12/1991
38 04/27/1992
39 04/27/1992
40 04/27/1992
26 07/23/1992
27 08/08/1992
28 08/27/1993
29 09/09/1992
0
0
0
1
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
1
1
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
0
9
9
9
9
8
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
1109
1109
1109
1109
1109
1109
1109
1109
1109
q
P
1109
1109
P
q
P
p.
1109
1109
1109
1109
1109
1109
1109
1109
q
. P
1109
1109
q
P
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
MRLC Consortium
Documentation Notebook
January, 1994
-------
32
32
32
33
33
33
33
33
33
33
34
34
34
34
34
34
34
35
35
35
35
35
35
35
36
36
36
36
36
36
36
36
37
37
37
37
37
37
37
37
37
37
38
38
38
38
38
38
38
38
3°
39
39
30 08/06/1991
31 08/06/1991
32 08/06/1991
33 08/06/1991
38 03/31/1991
39 03/31/1991
26 07/17/1993
27 05/27/1992
28 08/18/1993
29 08/15/1992
30 08/15/1992
31 08/15/1992
33 07/06/1992
26 08/09/1993
27 09/23/1992
28 08/09/1993
30 08/09/1993
31 09/21/1991
32 09/23/1992
33 09/07/1992
29 08/16/1993
30 08/16/1993
31 08/16/1993
32 08/11/1991
33 06/18/1992
37 06/13/1993
38 06/13/1993
26 08/07/1993
27 08/07/1993
30 08/23/1993
31 08/23/1993
32 08/23/1993
33 08/23/1993
34 08/23/1993
35 08/23/1993
36 08/23/1993
38 05/14/1991
26 08/14/1993
30 08/27/1992
31 08/14/1993
32 08/14/1993
33 06/27/1993
34 06/22/1991
35 06/22/1991
36 06/22/1991
37 06/22/1991
38 06/22/1991
30 07/20/1993
31 07/20/1993
32 07/17/1992
33 07/20/1993
34 06/15/1992
35 06/15/1992
36 07/01/1992
37 05/28/1991
9 07/24/1992
1 07/27/1993
32 07/27/1993
33 07/27/1993
34 07/27/1993
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
9
9
9
9
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1015
1109
1109
1109
1109
1109
1109
1109
1015
1109
1109
1109
1109
1109
1015
1109
1109
1109
MRLC Consortium
Documentation Notebook
January, 1994
-------
35 07/27/1993
36 07/27/1993
37 07/27/1993
28 08/19/1993
31 08/19/1993
32 08/19/1993
33 08/19/1993
34 08/19/1993
35 08/19/1993
36 08/19/1993
37 04/26/1992
31 08/26/1993
32 08/26/1993
33 08/26/1993
34 08/10/1993
35 08/10/1993
36 08/26/1993
28 06/14/1993
31 06/14/1993
32 06/14/1993
33 05/13/93
33 07/16/1993
33 08/01/93
34 05/13/93
34 07/16/1993
35 07/16/1993
35 08/17/93
36 05/29/1993
27 08/08/1993
29 07/02/1991
30 08/08/1993
31 08/08/1993
32 08/08/1993
33 08/24/1993
34 05/01/1992
34 07/20/1992
35 05/01/1992
35 07/20/1992
29 06/09/1992
30 10/13/1991
31 06/12/1993
31 10/13/1991
32 06/12/1993
32 07/30/93
33 06/09/1992
33 10/15/1992
34 01/03/1993
34 06/12/1993
27 09/18/1991
32 08/06/93
33 08/22/93
28 08/29/93
29 05/06/1992
29 08/29/93
30 05/06/1992
31 10/11/1991
32 07/12/1993
32 10/11/1991
26 08/04/1993
27 09/16/1991
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
1
0
0
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
8
9
9
9
1109
1109
1109
1109
1109
1109
1109
1109
1109
1109
1015
1109
1109
1109
1109
1109
1015
1109
1109
P-
P
q
P
P
P
q
P
1015
1109
1109
1109
1109
P-
P
1109
1109
1109
1109
1015
1015
1015
1015
q
P
1109
1109
1109
1109
1109
P
P-
q
q
q
1109
1109
P
q
1109
1109
MKLC Consortium
Documentation Notebook
January, 1994
-------
MRLC Consortium
Documentation Notebook
April 1994
12.6 Santa Barbara, CA
The sixth meeting of the MRLC Consortium was held in Santa Barbara, California, on
February 1-2, 1994. A copy of the meeting notes is currently under review, and will be
included in this section in a future update.
-------
MRLC Consortium
Documentation Notebook
January, 1994
SECTION 13
MRLC CONSORTIUM CONFERENCE CALLS
This section contains notes, when available, for Consortium
conference telephone calls held to date.
-------
MRLC Consortium
Documentation Notebook
January, 1994
MRLC CONFERENCE CALL DISCUSSION
DATE: 11/8/93
PARTICIPANTS:
Thaddeus J. Bara (TB) - ManTech Environmental
John Dwyer (JD) - Hughes STX
Jeff Eidenshink (JE) - Hughes STX
Karl Hermann (KH) - ManTech Environmental
Joy Hood (JH) - Hughes STX
Denise Shaw (DS) - U.S. EPA
Gail Thelin (GT) - USGS
Dorsey Worthey - U.S. EPA
NOTES BY: TB
Image Selection and Acquisition
o Image selection is an on-going process involving GT,
Michael Jennings (MJ) and EDC.
o 1993 data is being added to order list.
o EDC reported that 50 scenes are currently on order; nine
have received and 30 were shipped by EOSAT on 11/5/93
o (GT) Apparently GAP coordinators in the NW have ordered
a number of mosaic-quality (i.e., geo-referenced, terrain
corrected, edge tie-points) scenes from EOSAT. New
scenes will need to be ordered to make up for these in
order to meet MRLC needs.
According to GT, Thomas Holm does not want to ask
for anything additional from EOSAT under current
purchase agreement, so these make-up scenes will
need to come out of additional 50 scenes.
Affected scenes are in MT, also some in ID
MJ is talking with cooperators to find out what
scenes were ordered.
Acquisition and Processing Time Line
o EOSAT should be able to deliver more than 80 scenes per
month.
o EDC should be able to process 1 scene per day er
operator. Currently have a single operator (in
training) , by beginning of 1994 will have three and
possibly four.
Print Ordering
o Currently EDC charges $150 for developing
transparency, then $65 for 40-inch sheets
(1:250,000 for TM scenes).
o There is no provision for discount reductions unless the
number of ordered prints exceeds 25. The major cost,
however, is associated with preparing transparency. Once
the transparency has been made, then costs for prints
will be limited to $65, provided same projection is used.
o There is currently a 2-3 week backlog in photo lab for
-------
MRLC Consortium
Documentation Notebook
January, 1994
prints; special requests can be done with faster
turnaround with additional cost.
o Agencies currently can ac ruire prints through standard
EDC procedures, through customer service center. Only
requires a charge number.
— EDC will designate an individual to serve as point
person for MRLC requests, and will develop protocol
for ordering prints once that person is identified
(JE).
EDC will provide at CA meeting order information
under current conditions.
o EDC asked that the anticipated number of photographic
prints be provided at CA meeting, since this has not been
factored into EDC resource allocation.
DS indicated that EMAP-Forest expressed an interest
in having a number of prints available for field
work.
o EDC recommended that EPA develop an accounting system
that would designate the allowable users of prints to
indicate who can place orders.
Mississippi River Flood (MRF) Science and Assessment Strategy Team
(SAST)
o The SAST includes individuals from FEMA, USGS, EPA, USGS
interested in effects of MRF. The SAST will develop a
detailed CIS-based evaluation of the MRF, including a
landcover database of soil, terrain, DLG's, focusing on
conditions before, during, and after the flood.
Change detection techniques will be employed and
MRLC data .can, or will, be used as baseline (pre-
flood) dataset.
o SAST will spend 8 weeks at EDC starting 11/15/93;
temporary computing facilities are being established at
EDC.
— EDC not sure what their role will be. SAST may
have experts who know what they want to do and how
they want to do it, or they may expect EDC to show
the way.
JE anticipates that his staff will be 50% committed
to SAST efforts during the next 8 weeks.
o JE wanted to know if MRLC wanted to actively support SAST
program. This would entail re-prioritizing image
selection to emphasize MRF-affected areas (these images
would need to be ordered ASAP to meet 11/15 start-up).
The advantage to MRLC would be accelerated processing of
these images.
(DS) The SAST processing would need to comply with
MRLC procesing protocol, and fit into the national
database.
GT and DS also expressed reservations that MJ has a
priority list for on-going GAP efforts that may be
difficult for him to modify. So MJ needs input on
prioritizing data selection.
-------
MRLC Consortium
Documentation Notebook
January, 1994
(GT) Other efforts with short-term deadlines should
not drive long-term MRLC processing scheme.
Terrain Correction in Pre-Processing
o JH indicated that there is still an issue regarding
accuracy of terrain correction. She is talking with Sue
Jenson. According to Don Steinlawn(?) - scientist
working on terrain correction - it is better to do
correction even with DEM error than no terrain
correction.
o GT indicated that best paper is an unpublished one by
Acevedo comparing 4 systems (?) ; she will bring to CA
meeting.
Other Business
o IAG is in place between Vegas and EDC for Product Support
(JE). There is nothing to indicate that GAP will not be
able to make contribution (JE), but there is no budget
yet (GT, JE).
o EDC will provide a cost breakdown of processing flow at
CA meeting
o EDC will put MRLC list into ArcView once final scene list
is complete.
-------
MRLC Consortium
Documentation Notebook
January, 1994
MRLC CONSORTIUM CONFERENCE CALL
DATE: 12/17/93
PARTICIPANTS:
Thaddeus Bara (TB) - ManTech Environmental
Jeff Eidenshink (JE) - Hughes STX
Karl Hermann (KH) - ManTech Environmental
Mike Jennings (MJ) - GAP
Tom Loveland (TL) - EROS Data Center
Paul Severson (PS) - Hughes STX
Denise Shaw (DS) - U.S. EPA
Gail Thelin (GT) - USGS
Dorsey Worthey (DW) - U.S. EPA
NOTES BY: TB
o (PS) The TM scene order list will be constantly updated over
the next several weeks. Future purchases will be on a scene-
by-scene basis, rather than large blocks.
o Currently 403 scenes have been ordered. 180 scenes have been
received, including 68 today. Approximately 25 scenes have
been rejected to date based on EOSAT preview.
o PS reported that only a few GAP scenes have been received from
cooperators to date. MJ indicated that he will follow-up this
issue with the cooperators. PS requested that the data be
delivered without requiring special unpacking or programming.
MJ agreed to request this from cooperators.
o PS indicated that EDC will have CORE system up shortly to
easily review scenes before ordering from EOSAT.
o MJ indicated that GAP cooperators are very interested in
knowing about the availability of data for their states. Many
of the state projects are in idle mode until pre-processed
scenes are available. MJ had prioritized the list of scenes
to spread them out across the states, so that each state would
have something to start work on.
JE indicated that there are software issues remaining at
EDC that will take several weeks to resolve. Production
should start in January.
Production rate is anticipated to be l
scene/day/operator, with 3 operators.
o Region 2/3 scenes have the highest priority for pre-
processing. MJ indicated that GAP has scenes for NY, only, in
the Region (8 or 9 scenes). Region 2/3 scenes were at the top
of MJ's prioritization.
o DS requested information on photo products of selected scenes.
EDC has a 40"x40" product on its standard order sheet (JE).
-------
MRLC Consortium
Documentation Notebook
January, 1994
o The US6S NMD has produced a survey called User Needs Asessment
for Cartographic Products (6T).
o TL described a recent USGS meeting which dealt with hisotry of
development of land characteristics database for landcover
datasets. The meeting closed with discussion of the MRLC.
The Division now recognizes the need for land characteristics
database, and the concept is supported through Watkins'
office.
o TL invited MRLC Consortium agencies to a 2-day seminar on
InfoBase to be held 1/26-27/94 in Res ton. The seminar will
include USGS, National Image Display Lab, and the Dual Use
Program, and will discuss the continued funding and
development of InfoBase. USGS is interested in getting
potential users at meeting to discuss their programs and
program needs.
o TL also suggested a meeting on the EDC Project Plan in early
1994, to ensure that plan is compatible with participating
agencies.
o GT indicated that the MOU signatary will be Bob Hirsch, Acting
Director of USGS; JE indicated same for EDC side. MJ said
Gene Hister (Acting Director of NBS) is reviewing MOU. High
level signataries are appropriate for this document.
o MJ expressed an interest in getting the USFS involved to
provide resources for clustering and development of SPECTRUM.
He has a FS contact in Salt Lake City who is interested.
o MJ proposed that programs send personnel to Ames to do a pilot
study and evaluate SPECTRUM. The group would be tasked to
perform a joint classification and produce a report in the
form of a paper ready for publication. GT and DS both
expressed interest. MJ will fax an outline of approach (copy
of fax attached).
o Regarding Region 2/3 implementation, DS met in Philadelphia
with GAP, Region reps, Delaware Bay Program. She reported a
general interest in proceeding with the implementation.
o The next MRLC Consortium meeting will be held February 1-2,
1994, in Santa Barbara before the GAP accuracy assessment
meeting (February 3-4).
-------
DEC-20-1993 09:34 FROM IDCFURU
TO 19541495889535727194 P.01
MRLC Consortium
Documentation Notebook
January, 1994
COLLEGE OF FORESTRY, WILDLIFE AND RANGE SCIENCES
UNIVERSITY OF IDAHO
MOSCOW, ID 83843
Fish and Wildlife Resources
Idaho Cooperative Fish and Wildlife Research Unit
Cooperative Park Studies Unit
A FACSIMILE MESSAGE
PLEASE DELIVER PROMPTLY
TO:
FROM:.
DATE:
TIME:
NUMBER OF PAGES TRANSMITTED
(INCLUDING COVER PAGE)
IF YOU DO NOT RECEIVE ALL OF THE PAGES, PLEASE CALL
Phone: 208-885-6336 FAX No.: 208-885-9080
-------
DEC-20-1993 09=35 FROM IDCFURU
TO 19541495889535727194 P. 02
MRLC Consortium
Documentation Notebook
January, 1994
Preliminary Thoughts on MRLC Pilot Projects
Michael Jennings, Gap Analysis Coordinator
December, 1993
Here are some quick thoughts on putting the projects together. They
should be the first addendum to the MOU, and workplans should be written
under that framework. The pilot areas are in the Bailey province-level
ecoregions of Eastern Deciduous Forest and Pacific Forest. Overall, these
projects should demonstrate:
1. The scientific and technical capability - "It can be done"
2. The imperative for cooperation - "It can be done reasonably"
3. How complex ecosystems can be interpreted and described for
democratic decision-making processes - This is why it is important
and how it can be useful"
COOPERATIVE ASSESSMENT OF ECOSYSTEMS
BY LAND CHARACTERISTICS
Project One: An Application and Evaluation of Spectrum Software by the
MRLC Partners.
Objectives - A. Generate a land cover map of each pilot area.
R Evaluate the utility of Spectrum for this task.
Methods -
A. Send experts from each program to Ames for a
week to work with Susan and Len to classify and
label the MRLC TM data for the pilot areas.
Potential participants:
Susan Benjamen, Ames
Len Gaydos, Ames
EMAP-LC, to be designated (Carl Herman??)
Ed ?? (C-CAP)
Gail Thelin, NAWQA
Ann Rasberry, GAP-MD/DE/NJ
Wayne Myers, GAP-PA
Kelly Cassidy, GAP-WA
Tom O'Neill, GAP-OR
EDC, to be designated
-------
DEC-20-1993 09=35 FROM IDCFURU
TO
19541495889535727194 P.03
MRLC Consortium
Documentation Notebook
January, 1994
Products -
Conclusion •
B. Evaluate Spectrum for this application.
C Describe uses and limitations of products.
A. Pilot area data sets.
a Technical report suitable for peer-reviewed
journal (with color plates).
C. Hard copy maps.
If this project is successful, then use its data for
Project Two. If not successful, then have each of
the partner projects generate preliminary land
cover maps for the ecosystems that they are
concerned with (i.e., C-CAP does coastal, NAWQA
does agricultural and urban, GAP does natural
terrestrial). Then bring these data sets together
into a single land cover map.
Project Two: An application of the MRLC pilot project land cover data set
to the NALC triplicate MSS data sets.
Objectives -
Methods •
A. Demonstrate how the MRLC land cover data set
can be applied to historical data, showing in
visual and statistical dimensions, the
geographic degree of land cover change in
specifically labeled categories. For example: a)
the Quercus alba / Pinus ridgera natural
community (or its coarser-level classification
of "Pine-Oak Forest"); b) the Estuarine
Intertidal Emergent Persistent Regularly
Flooded Spartina foliosa / Salicornia spp. (or its
coarser-level classification of Estuarine
Intertidal Emergent Wetland); c) the habitat of
certain vertebrate species (such as the Saw-
Whet Owl).
B. Demonstrate how present technology and inter-
agency cooperation can be used to show trends
in landscape configuration.
A. Apply the Project One data set to the NALC data
sets.
-------
DEC-20-1993 89:39 FROM IDCFURU
TO 19541495889535727194 P. 02
MRLC Consortium
Documentation Notebook
January, 1994
Products -
Lead Person - Dorsey Worthy
Visualization / Animation - William
Technical writing - Thaddius
Ames
A. A land cover change data set.
B. A technical paper suitable for peer-reviewed
publication (Ecological Applications...?).
C. An animated interpretation of findings.
D. Hard copy maps.
Conclusion - Briefing at one of the Smithsonian museums.
-------
MRLC Consortium
Documentation Notebook
April, 1994
MRLC CONSORTIUM CONFERENCE CALL
DATE: 1/21/94
PARTICIPANTS:
Thaddeus Bara (TB) - ManTech Environmental
Jeff Eidenshink QE) - Hughes STX
Don Field (DF) - C-CAP
Karl Hermann (KH) • ManTech Environmental
Joy Hood (JH) - Hughes STX
Mike Jennings (MJ) - GAP
Tom Loveland (TL) - EROS Data Center
Paul Severson (PS) - Hughes STX
Denice Shaw (DS) - U.S. EPA
Gail Thelin (GT) - USGS
Dorsey Worthy (DW) - U.S. EPA
NOTES BY: TB
Santa Barbara Meeting
o DS and MJ will put together an agenda for the MRLC meeting to be held at Santa Barbara on
2/1-2.
o GT indicated that Susan Benjamin of Ames would like to attend the Santa Barbara meeting for
one day to provide an update on SPECTRUM work.
- By 1/28, Los Alamos will have a stand-alone version of SPECTRUM running on a Sun.
This will be the current version of SPECTRUM, without the enhancements discussed at
the Ames MRLC meeting.
- Any additional conversion efforts should go through University of New Mexico and Los
Alamos. Although Los Alamos has not pursued any requests for resources to date,
additional conversion on their part may require mem to do so.
o JE indicated that EDC will be sending 1 person (Brad Reed) to GAP Accuracy Assessment
workshop. If MRLC meeting will be fully attended, Reed will be there as well.
o MJ will provide an accuracy assesment thought paper to all programs prior to the GAP
workshop.
MRLC Classification Prototype
o GT indicated that SB wanted to start planning for the Ames prototype workshop, including the
identification of who will be involved and when it will be held.
o MJ thought a decision still needed to be made as to whether sub a workshop should be held. TL
questioned its need, because programs are already running. Ideally, it should be possible to take
results from regional programs and subsequently resolve the consistency issues. Focussing on
what system to use for classification would take emphasis away from ongoing work; it would be
better for cooperators to continue working on the paths with which they are more familiar.
-------
MRLC Consortium
Documentation Notebook
o If a workshop is held, then it might make sense to have it on the East Coast (DS afif GT). The
Reston lab would be a good alternative to Ames, if software can be installed and supported there.
Region 2/3 Implementation Meeting
o DS described the regional implementation meeting scheduled for 3/2-3 in Annapolis for
generating joint efforts between CCAP, EDC, GAP, NAWQA, and EPA. The meeting would
set roles and responsibilities, status of each program, delivery dates for programs* products.
o GT stressed that it would be important to have a plan going into this meeting, rather than trying
to use the meeting forum for the purpose of developing an implementation plan.
TM Scene Processing at EDC
o JH indicated that production has started on the first 2 Path/Rows. These first four scenes will
be sent to Susan Benjamin for forwarding to Los Alamos by 1/25.
— There have been some hardware problems which has delayed start of processing.
- EDC is still trying to implement Los Alamos clustering algorithm on SG machine at
EDC.
o EDC has prepared a prioritization list for preprocessing remainder of scenes and submitted to
DW and MI. Additional copies will be sent to GT and DS.
o Morgan Sarges (EDC) will send MI a list of what has been received from GAP states.
- EDC is still waiting for scenes, mostly from NW
— • EDC has ordered replacement scenes for MT
- For other states:
1. LA scenes were all scene-shifted and not usable,
2. FL sent 14 scenes packed onto a single 8 mm tape; EDC has not been successful
in deeding it (MI will have FL send mem separately)
3. EDC is interested hi NM scenes, which have not been received as yet, because
there are a large number of scenes.
o IH indicated that Kent Hegge (EDC) has prepared and sent out 2 weeks ago a sample order form
on how to order data, and track cooperators. Input from the other programs was requested.
o MI indicated that GAP funding for preprocessing is now available. He will expedite resource
transfer which should gor through soon.
MRLC Documentation Notebook
o DS requested that comments be provided by 1/26. Otherwise, notebook production will be
pushed back until well after the Santa Barbara meeting (TB).
o Both MI and GT complimented the draft version of the notebook.
o MI indicated that NBS is developing its own logo. Once available, it will be incorporated in the
notebook (TB).
-------
MRLC Consortium
Documentation Notebook
April, 1994
Memorandum of Understanding
o (GT,TL) USGS will go to Director for signature, when a Director is appointed (should be in
April, based on Congressional confirmation hearing).
o DF indicated that MOU will be signed by COP director.
o MJ indicated that MOU would be signed at least at the level of the acting director of the NBS.
o DS indicated mat MOU has been sent to Browner's assistant with request for Browner's
signature.
o TL indicated that for parallel tracks, USGS (NAWQA and EDC) would require Babbitt's
signature. MJ indicated that NBS would then also require Babbitt's signature.
CCAP Business (DF)
o Classification and change analysis in the CCAP Columbia River study area will be completed in
February by a West Coast researcher.
o CCAP has updated the classification document with a January 20 version. This is the version
that will be printed by NOAA.
o Jerry Dobson will probably attend the Wed. session of the Santa Barbara meeting and should have
Khorram's change analysis accuracy report with bun.
Other Business
o JE wanted to know what the exact paper titles and the number of talks for Reno CIS meeting,
MRLC session.
o DS indicated that DW will be preparing a slide show mat will display all of the standard MRLC
products. This will be availablelater this year.
-------
MRLC Consortium
Documentation Notebook
April 1994
. SECTION 14
MRLC REGIONAL IMPLEMENTATIONS
The MRLC Consortium is currently pursuing regional implementations to facilitate the
cooperative efforts of the participating programs. This section will include relevant information
pertaining to these efforts.
-------
-------
MRLC Consortium
Documentation Notebook
April 1994
14.1 Region 2/3 Implementation
On March 2-3, 1994, the MRLC held a meeting for the Region 2/3 Implementation
effort. This section includes a set of notes summarizing what was discussed during this meeting.
The Appendices containing the overhead projection material used by the primary speakers are
not included in this notebook because of their length, but are being held in the MRLC file
system. Additional information relating to this implementation effort will be included in future
updates.
-------
-------
MRLC REGION 2/3 IMPLEMENTATION MEETING
MEETING NOTES
Meeting Site: Ramada Inn, Annapolis, Maryland
Dates: March 2-3, 1994
Moderators: Denice Shaw (U.S. EPA) and Mike Jennings (NBS)
Notes compiled and edited by: Thaddeus J. Ban, Senior Scientist
ManTecb Environmental Technology, Inc.
A copy of the meeting agenda is included in Appendix A. Appendix B lists the name of meeting invitees
and participants
March 2,1994 - Morning Session
INTRODUCTION
Denice Shaw of the U.S. EPA provided introductory remarks to the meeting. Overheads used in her
presentation are irefadgd in Appendix C.
The MRLC Interagency Consortium began in April, 1993, and consists of
EPA - Environmental Monitoring and Atieiiment Piograiu (EMAP-LC)
Norn America Land Characteristics Project (NALQ
NBS • Gap Analysis Program (GAP)
USGS - National Water Quality Assessment Program (NAWQA)
Norm America Land Characteristics Project (NALQ
NOAA - CoastWatcb Change Analysis Program
The EROS Data Center (EDC) is the central coordinator of many of the Consortium activities.
Although the different agencies have different purposes and requirements the common
dominator was a need for land cover information. Advantages were **p*srfrrt for cooperative
efforts to address the acquisition, processing, quality assurance, applications and management of
landcover data in a unified way. Although the initial model for cooperation envisioned divergent
labelling requirements, mere is now a recognition of convergence in mis area.
Although the pmcemc has not hem wfrhmtf fat pmhleme, th» enat tavinpt haw Wn tnh»a^a|
with a conservatively esthnatBd saving of 30 million dollars through the following: data
purchase, image processing, data.
The Region 2/3 EPA MRLC Land Cover Implementation will be divided up into various areas
by the participating programs. Each program working within the region will work together to
be consistent and we are working together to facilitate these activities. There are several
additional potential participants which may join these efforts.
OBJECTIVES
Mike Jennings delivered comments about the objectives of the meeting.
-------
o Hie reason for meeting today is to modify the framework by which resource management and
protection is implemented. Historical efforts, such as the Clean Water Act, etc. are ad hoc
approaches which have failed in many ways. Comprehensive land use planning was previously
rejected or not well implemented, because of institutional and funding constraints, and competing
ideology. Hie results have been ineffective management, and the inability to develop and
integrate ecological information of critical importance.
o We need to work toward developing information on landcover and land use condition and change.
There is now an opportunity to begin building a complex data set mat relates to an entire
geographic area. *n»fc MM} of tfifaijr ha? not happaigd hymn** of "ytfottfo"^ constraints If the
different programs can coordinate acihrWes, we can devdop greater understandings, iiialdng this
an exciting «°d positive effort to break down barriers between agencies and programs . Already,
the MRLC TM purchase is the largest civilian Landsat purchase ever.
— For example, if we can agree to a set of geographic boundaries and basic definitions men
mere can be complementary labelling of landcover polygons, and the opportunity for
shared field work, metadata development.
• If the Region 2/3 effort is successful, then the concept wfll be applied to other regions
of the U^.
MRLC GOALS AND OBJECTIVES
Tom Loveland of the USGS EDC provided comments on the goals and objectives of the EDC MRLC
Monitoring System (MRLCMS) project.
o In spite of nun than 20 years of Landsatariodiersatellh^
picture of the surface of the earn. We lack a good adequate baseline at multiple scales from
which to get information. We need to develop global and national environmental baselines and
environmental monitoring purposes. In the U.S. the effort has evolved into many programs
national as weU as mteraational. We also need to understand how laijdsuAces is changing over
time.
The MRLCMS is planing to provide current frwlhif of global multi-scale characteristics, and
marhaninm of monitoring, targeting and ttMsiing environmental changes. MRLCMS has 3
objectives:
1) develops a baseline of global land characteristics,
2) development of a regional database,
3) a monitoring and assessment stage.
Objective 1 wfll develop a global land characteristics database, consisting of seasonal land cover
regions. USGS is committed to product global landcover classification. Efforts are going to be
based upon a hud surface characteristics database prototype that wfll be international in scope.
EDC is developing a strategy for a land cover regions multi-purpose database mat can provide
data to users wim a wkie range of possible applications. The prototype is being developed with
potential uses in mind as well as existing programs.
- EDC has tried to regionalize landcover based on 30 variables that describe landcover
condition. Statistical techniques are used to categorize the areas. There are 159 classes,
of which 1 12 are unique
- The regionalization is a database approach. All data used to generate the classes are
retained as attributes in the coverage. Regions have attributes including climate
characteristics, temperatures, elevation, spectral responses, and linkages to existing
-------
classes.
Several examples of the landcover data were presented including greenness, vegetation
seasonal responses, and input to landscape pattern and process models.
For U.S. work data set is available at USGS for $32 and includes all of the attributes that
have been developed. Layers include land characteristics data, source raster files,
derived raster files, descriptive and statistical attributes. The attributes represent an audit
trail used in defining regions as well as supporting data for die regions.
A global land cover data set will be completed by 1998.
Currently doing "m*1"' greenness calculations. Completing a profile for landcover and
have just finalized information for 1990. Full set should be available in the next four
months.
Cooperative efforts with universities have experimented with national scale and global
scale changes. Experiments conducted by mis group represented many types of
Most of the work has mdiratad positive results in mesoscale modeling.
Objective 2 involves me development of a regional prototype land characteristics database using
the raqiitrgniftiflp ^if ftia jirriyrMiM partiriparing in the MRLC Commtium. The emphasis is On
in "^"i processing *"^ the collection and development of landscape characteristics
data to aUow product to be used mother applications. Based on ongoing multiagency landcover
wort in Alaska provides evidence that this effort can be highly successful.
- Meeting the MRLC land characteristics data base research objective of refining and
adapting the land characteristics consistent with Landsst TM data; and developing and
improving methods for landsat-based large-area land characterizaticmwffl be a continuing
challenge. Some of me issues include
1) evaluating the limns to spectral signature artansion,
2) developing strategies for selecting classification parameters that provide consistent
spatial structure from scene to scene, and identifying a strategy for astfinhlinf
scene-based special regions in seamless regional data set.
3) dftrnnhimg the role of coarse resolution time-series data for improving the
region labeling and documental ion, and assessing change.
- The use of common source materials, particularly aerial photos or remotely sensed
imagery, field data, and a clear understanding of potential applications, consistent
documentation of data bases, use of appropriate validation strategies wfll solidify success
of activities.
o Objective 3 reflects the interest at USGS in national and mtenationalinontoring of landcover
condition and change, The main problems are identifying change and analyzing change. Current
change detection algorithms are only adequate for characterizing a limited set of the many types
of landscape rfiangat,
- AVHRR wfll be used to provide daily looks to determine changes and conditions in
surface activities TM and other data wfll quantify magnitude- of changes occurring.
USGS LAND USE/LAND COVER PROJECT
Kathy Lins provided information on the ongoing USGS LU/LC project Overhead projections used in
her presentation are included in Appendix D.
Objectives of the current USGS national land use and land cover program:
1 . Develop capability to produce nationally consistent LU/LC data
-------
2. Develop classification standard to meet multiple needs for a nationally consistent dataset.
3. Develop methods for collection i"d automation of LU/LC database.
4. Work with the user community to define the product.
USQS is not in business of production data for own use but want to collect land use land cover
data for other users.
Former LU/LC program provided nationwide coverage available in graphic and digital form at
Anderson Level U classification systems, at scales of 1250,000 and 1:100,000 in GIRAS format.
- used high-altitude aerial photography and manual interpretation
- is still a widely used standard but is now out of date
Current program is moving from GIRAS format.
USGS and USEPA co-hosted a LU/LC Forum for federal and state program representatives
"riHmj USFWS, USGS, USEPA NOAA, and state programs as producers of data. We also
need to reach the user tt"""""*fry also so we sent out a questionnaire; 397 responded to the
survey, including 218 federal, 133 state, 24 local, 16 academic, 6 private.
Primary mnarf^g fppKeafoi. inrfiiri* mtmr jpality, «i«rianHB planting v*gt*atinn fhatig*
runoff, water use, wildlife, urban development aquatic/marian habitat analysis
LU/LC retnirements and preferences from survey
1. digital vector data (66% of respondents)
2. 90-percent accuracy
3. 2.5 acres or less minimum mapping unit
4. 124,000-1:59,000 scale product
5. register to an accurate base map
6. 3-5 year updates
7. Classification rrpferfpm — Anderson Level n compatible and Cowardin wetlands
Actual LU/LC parameter*:
1. 1:100,000 scale digital vector data
2. DLC format
3. LU/LC data registered to hydrography and transportation DLGs
4. 10-tcreMMU
5. compatible with Anderson Level n (*j*f»ig*» not an **yt match) t1*^ Cowardin
6. Land cover mapped as a continuous surface
7. Land use mapped as additional attributes
8. More man one description possible for any parcel.
Currently trying to move prototypes to digital mode. We are working with ESRI in the
Vancouver/Portland area and intend to develop a TM-based dataset using Spectrum software;
looking at digital data for i«pd use data for one composite product p**d to evaluate
to derive the MRLC datasets. A Proof of Concept document wfll be created.
March 2, 1994 - Afternoon Session
Hie afternoon session consisted of presentations of MRLC Consortium programs, an update of EDC
-------
processing of MRLC data, and a presentation of MRLC clustering approach.
GAP - Mike Jennings
o GAP is interested in identifying the gaps in current conservation effort lands. GAP is a terrestrial
vegetation-community based approach mat looks at where they are they and how well they are
represented m conservation lands.
o Good information on vegetation and land cover is of critical importance.
o Aquatic analysis is also being started in GAP, and will be implemented as much as possible like
vegetative cover.
o GAP analysis maps present day status of common, nsirve plants and animals, to develop status
of plants and wildlife based on information on habitat and species, and distribution of habitat
relative to land ownership/land use.
NOAA C-CAP - Don Field
o CoastWatch Change Analysis Program supported by NOAA Coastal Ocean Program managed
in Beaufort, NC. Monitoring and research for understanding, prediction and «""i^pman» of
human changes, land use/change, management, regulations planning, effects land cover and
habitat change, which in effect effects fisheries. Based on recognition mat standard mapping is
not geared for ImHttt change.
Purpose: land cover change by aerial photo
Scope: U.S. estuarine drainages
Frequency: 1-5 years
Data Source: Landsat Imagery
Approach: Regional Projects arranged with states
Methods: Inmlemem plan devdopedmrough a series of regkmal and topical workshops
Contact: Dr. Ford Cross, NOAA.
Protocol issues are addressed in a guidance document prepared by Jerry Dobson and Ed Bright
at the Oak Ridge National Laboratory.
— Coverage*mland,ofishore *IK* interregional boundaries
- Tidal Considerations
Vegetative State
— Quality Assurance/ Quality Control, etc.
Protocol development projects have included
- Accuracy Assessment
- Tidal Effects
- Palustrine Forest Classification
- CIS Integration with National Wetlands Inventory and Other Data,
- Special Change Analysis
- Linkage with process and transport models.
Major project has been Chesapeake Bay, a 4-scene area comparing 1988-89 data with 1984 data.
-------
Completed and for sale in Washington Data Center in Washington, D.C. $200.00. Classifications
for 88-89 and 1984 and the changed products with two complete maps with loss by class and
areas that did not change. Information in Appendix E.
o Other projects:
1. Maine and Canada 1985-1992
2. Columbia River Estuary Task Force 1989-1992 with change detections.
3. YakJtat Bay, Alaska
4. Seagrass work in Core Inlet, NC
NAWQA - Gifl Tndin (Overheads in Appendix F)
o NAWQA objective is to describe the quality of the nations'* water resources in a nationally
consistent manner. Began in 1991.
— study »"frt - detailed long-term atfurnrrr* of the most important individual hydrologic
systems covering 45% of land in the U.S.
— national and regional analyses of highest priority water-quality issues. VOC's have just
begun.
- using existing multi-Male data and collecring new data
o 60 study units being hnplmmnmi in 20 study mitt in moving 3-year windows: 3 years of
intensive data collection tallowed by 3 years of monitoring
o Hierarchy of land use levels
1. National Design and SU piy»»»^g
2. SU Design and National Analysis
3. Component design and analysis
4. Site characterization and analysis
NAWQA - Gary Fisher (Potomac River study unit)
o In region 3 there are 8 study areas. Currently there are 4 active study areas. In 1994 three new
studies, NC,VA,WV. 1997 wfll start the second cyde for the Debnarva region.
o NAWQA cannot afford data collection on a short-term basis and MRLC is critical for success
of program.
o Potomac River study unit is 15,000 sq. mfles, in four states and DC; therefore, there is much
interest Study area divided into 5 subsections, tumganmltHliscfrlinary approach. We use land
cover to interpret what we see. InMD/VA we have existing land cover data sets. We are going
to do comparisons of the two t
Since NAWQA field crews are out in die field anyway, and can do data collection for other
purposes such as QA/QC of database.
EMAP - Denise Shaw (Overhead in Appendix G)
-------
o EMAP objectives:
1 to estimate the current status, treads, and changes in selected indicators of the condition
of the nation's ecological resources on a regional basis with known confidence.
2. Bfcriiprtf. the geographic coverage and extent of the nations's ecological resource with
known confidence.
3. Seek associations between selected indicators of natural and anthropogenic stresses.
EMAP Mid Atlantic Highlands Assessment (MAHA) - Tom DeMoss (Overheads in Appendix H)
o Reliance on Data Goal • Objectives.
1. Annually analyze data to fgfNfr11 regional and programmatic environmental and
iit goals and implement a system to measure progress towards meeting those
goals.
2. Integrate indicators into goals, priorities and success measure of regional surface water
program as a prototype for full information in all programs.
EMAP sampling within the integrated assessment region, shown with ecoregkms. We identify
threats to the region by looking at historic data, and are working on ecoregions and organizing
principles, within defined areas like Chesapeake Bay, MAHA.
MAHA is on 4-5 year project with 250 sites each year. Each site has reference sites - what is
me baseline, what are least impacted areas? REMAP program collecting data from water quality,
fish fafa""ntion, benthics, *jph tissue samples, water quality, *nHtift OTJ what association do we
see between indicators an* water quality nti htfrffft
Sampling approach is multi-agency.
Products' of MAHA -Condition of stream reioim» inipaired biota and habitat WA based upon
reference conditions by biocriteria development, pollution prevention,
Assessment Information Sources/Uses — Surface Water, Forest, Estuaries, Landscape
Characterization, Coastal Initiates, GIS, Historical Data, Bird Survey, Wetlands, Groundwater,
Human Health, Agroecosystons.
- Assessment Information Sources/Uses integrate forest, estuaries, and surface water.
Landscape characterization against estuaries will be great to compare against each other.
NALC • Dorsey Wormy (Overheads in Appendix D
o LandiHtf Pathfinder uses MSS, combining 1970's and 1980's data in a co-registered dataset
NALC is producing triplicate images - georeferenced and reregistered, Four Band MSS Images,
1970', 1980's, with 1990's TM data from MRLC; pixels are dated with identified datasets,
resampled to 60 meters to align with TM.
o Current study regions: Southern Mexico, Oregon and Cues. Bay Watershed. Will complete all
of Norm American and Mexico this next year.
1970 and 1980 are photo interpreted, hope to have ground interpretation inthe90's. Some areas
-------
can be validated, but a lot of tbe areas cannot be.
NALC interested in taking advantage of die fact that a lot of people are out in die field already
collecting data.
VS. Forest Service - Frank Koenig
o Three years ago USFS became formally adopted ecosystem management approach. A multi-
agency couortiam, consisting of federal agencies and state has been formed, wim meetings twice
a year. The frMfr is on ecological mapping, education, landscape planning, etc. A copy of
the 1993 Annual Report on the uteragency Cooperation on Ecosystems Management was
provided and is included in Appendix J. The group has many similarities to MRLC approach,
and the two consortia should communicate with each other.
U.SJorest Service - Rachel Hetshey (NE Region FIA)
o We would like ID coinbme our efforts with MRLC, and partichtttem landscape characterization.
Oar data oar point samples, and we are looking at interpolation to continuous coverages. We
would like to have 100* landcover by satellite mngery. We have 14 states in NE. Data are
digital; bom photo plots and field plots are digitized, wim aapfMrtmg ah-photography.
o We would like to make oar plot data availably to omen. Ootistr«*"ft'
i. P*ot jftBuiinmtiofl
2. oonfirtanriility agreements limit access on private land, though an public land is
Soil Conservation Service - Bob Smith
o SCS is a varied group; inventory practices were established in 1972. In 1982 we gained the i
inventory. Currently contracted wim Iowa State Univ. There are 300 sample units from 1982
to 1992 witb field operation field data forms. All data processed on computers. Also use of high
altitude photos and color slides.
o Three LC/LU data layers wim seven components. LC/LU is similar to Anderson concept SCS
allocates all land mat can be seen, crop, forest, range, etc. which encompass all land in U.S.
We historically have a problem wim inconsistent estimates compared to USFS on how land is
specified.
o We have a data base compiling information on earth cover use. We don't actually do mapping
of the areas. Attribute data will be available in April. The point will be valid to certain size
units, but precise location will not be provided. We have agreements with other agencies to give
out tbe data points.
-------
EROS Data Center (EDC) - Joy Hood
Joy discussed the EDC operations concerning MRLC. The overheads used in her presentation are
included in Appendix K.
o EDC is worlmu; on devdopment and application of a radti-r^
system.
o EDC was commissioned to purchase one-time coverage for 1991-1992. EOSAT also allowed
an additional 100 scenes, plus ISO scenes previously purchased by GAP. 1993 data was
subsequently allowed, because there was inadequate good data in 1991-92 window.
Scene selection was based on ^laameiitg between consortium members particularly those of
GAP, NAWQA AND C-CAP. Optimum date for each path/row was determined based on land
cover. Initial selections based on optimum date and availability of high quality 1991-92 data.
Smiritatm 1990-1 scene, 1991 -77 scenes, 1992 -170 scenes. 1993 -199 scenes. 126
are multi-temporal, 321 are one-time coverage. Of the 63 multi-temporal pairs ordered,
52 an same year, 11 are different yean.
Preprocessing
initial steps
1. Debanding
2. Detector Noise
3. Qiuck prim dbeckftn-additkmal ouality assurance
- Tyeof preprocessed image is copied and stored in EDC archives, with p-code copy
returned to EDC.
— EDC is considering adding a browse capability so users can check quality of image
Once preprocessing is done we do geometric processing and let each agency pick their own
parameters. We decided to have a set default data set
- 20 control points selected for each scene; The control point files wfll be written to a
special Cm Archive director. The appropriate control point tafonnation wfll be passed
to Archive Management Section for fachmop in the complex data base.
- Multitemporal coverages - the most recent image wfll be registered to a map base and
used as the reference image for subsequent registrations.
- Images wfll be verified against 1^4,000-ccale topographic maps. All images failing to
meet the requirement of geometric accuracy of between +/- 1 pixel wfll be rejected and
new control selected.
- The registered image and corresponding DEM data wfll be archived to tape The MRLC
database wfll be updated to contain the registered unage archive location and appropriate
- EDC is developing a project plan that wfll detail full processing.
Cluster registered image wfll be a standard product
- where single date data are available the six rectified TM images will be input to the
clustering algorithm, if multitemporal data exists over the past/row the 12 resampled
preprocessed TM bands wfll be input to the clustering algorithm.
— Clustered unage wfll be archived to tape with asswiatffd statistics files. The MRLC
-------
base wfll be updated to contain the clustered image archive location and appropriate
— Upon successful registration pertinent processing information will be archived on digital
tape.
— Processing wfll not be dependent on clustering algorithm.
o Derivative data sets developed by MRLC programs wfll be returned to EDC and archived.
o Metadata wfll be complete for image, facMfag how ft is preprocessed and geocoded, and what
is applied in mis process. It is all interconnected and you should be able to backtrack. We are
still in to development stage, but mis is what we are working towards.
o Currently, 14 path/rows nave been geocoded. We want to be able to provide information for
consortium meinbersOT what is available tiiroughmterneL We are working dosdy to prioritize
to various states. Regions 2/3 have top priority. Not everyone's data wfll be done tomorrow.
Each scene taking 3-4 days. We have 3 analysts working and hope to have 3 scenes per day
done each day.
Procurement is ifanfrfa^ processing. We are currently Atnumg out of storage space. We have
ID DDPCDSSfi lufi uSIulVaVB DV
We are working with consortium to determine what level of data and production, and cost of
reproduction through a standardfred MRLC order form. We also need someone in the
consortium ID sign off on the data each person or agency needs. Costs nave not been set as of
yet but we wfll have mem determined shortly.
Susan Benjamin (USGS. NASA Ames Research Center)
Susan discussed to spectral clustering of MRLC images. Overheads are included in Appendix L.
o Hyperdustering program to duster and classify mnhisnectral digital images into « high nmnher
of dusters. Product of Las Alamos National Lab.
— Advantages over more «**nimn«i dustering methods:
1. able to buud more dusters
2. dusters more likely to be representative of to full spectral variation m tonnage
3. duster statistics are attached to to output file
4. fast and robust.
- dustered dataset can be handled as if it were stiU multi-spectral image.
— wiiuen to run under the Khoros image processing system.
o Clustff*n£ routine constructs a 'codebook" of duster statistics from the multispectral image.
Monte Carlo sampling of multispectral data to develop duster statistics. At each iteration, a new
random sample is drawn, in current implementation can build 240 dusters in 12 iterations. We
may be able to increase to more man 4000 dusters.
- routine assigns each pixd in to multispectral image to one of to dusters described in
to codebook; when done, adjusts to codebook to reflect to spectral signature of to
pixels actually assigned to each duster value; adjusted codebook (duster means and
covariance matrix) is inserted into to header of to duster file.
-------
Results:
— data compression of n input bands reduced to a single clustered data band
— cluster statistics are tight - deviation from cluster mean tends to bless than 5 DN's,
function of high number of clusters.
- clusters represent the spectral variation of the image (e.g. will not get lots of water
classes).
- clustered image can be viewed as thought it was still a multispectral image using the
known per-band cluster means:
!• different OCDK oflniPHUffons C8D DC wisplAyou
2. functions of bands can be computed and displayed
3. spectral reflectance corves for individual clusters can be displayed
— interpretation of the clustered df+wf nun land cover ™fr« is done using Spectrum.
Hyperclustering programs now run nmif Kboros on Sun. EDC is implementing the algorithms
to run as a pan of MRL£ production system. Los Alamos cluster algorithm will be interfaced
with LAS image processing system used at EDC. A Khoros-independent Spectrum is being
developed at University of New Mexico.
- EDC inmlementationwiU teased to test perfo^
First MRLC TM dataset has been clustered at Los Alamos: Path/Row 15/32 over central PA,
multhemporal 6/17/93 and 10/20/92.
— Clustered 3 ways:
1. 12 bands cluster, bom scenes bands 1-5 and 7
2. 6 bands duster, June scene
3. 6 bands duster, October scene
- These will be inteipieted and evaluated at the first MRLC users workshop by field
personnel of the cooperating agencies.
Uncertainty about abflity to pick up rare cover types.
Atmospheric correction will not be perioiuied - since bdieved to cause more problems by
correcting it than ignoring it
NALC Status - Dorsey Wormy
o There are 830 scenes and we have gone through imagery for Norm America. We have some
areas where 10% cloud free is all we have gotten for the past 20 years.
o We are working in southern Mexico, we will be doing Cuba, Honduras. We will begin the U.S.
in 1995. The U.S. was taken off-line as a priority, but we are trying to get h back on-line as a
priority.
o We are hoping to do die Caribbean in 1996 and Cuba and Mexico next year. If access is needed
to this data, we can bump h up in order, but wfll require additional funding.
The Ches. Bay Watershed Pilot Project covers imagery from 1988-1991. Began with invalidated
spectral dasses, reduced it to six categories and had 80% accuracy. Air photography followed
up with ground points for assessment. Six levels: water, woody, herbaceous, barren, developed-
-------
high, developed-low.
Thursday, March 4, 1994 - Morning Session
CLASSIFICATION STANDARDS
Mike Jennings discussed die GAP vegetation community classification standards.
o GAP Analysis has evolved as a state level approach to date, and each state use their own system.
A taxonomy of vegetation cover types has developed with GAP, the Nature Conservancy, and
die Natural Heritage Program.
- A vegetation commontty sub-committee has been formed in the Ecological Society of
America. This will be a forum where non-agency, scientific personnel can discuss and
refine a vegetation community taxonomy. This effort will provide long term stability.
o GAP vegetation riamtlficatkin is described in GAP Technical Bulletin 2. We are undertaking a
revision of mis document.
baaed on UNESCO world vegetation classification
— physiognomic and floristic approach
- floristic tier wfll describe existing vegetation rather man potential vegetation.
— sot categories: Class, subclass, groupylufiiuTlhiB, cover types and ***mn|""|^y type
(community type wfll be named alliance in new version)
- for states that did not use mis approach, wfll need to go back and break mem out m tie
future
o We are interested in mapping vegetation cover with remote sensing. There is confusion with
wetland classification system, but we hope to incorpMateweOattlcla»i«wim the GAP approach.
— aquatic cover types wfll be included
- developed areas wfll be stratified out as developed categories (ie., urban, agricultural).
» Experience with states mat have started from scratch with this approach has been good. Arkansas
is the best example. They developed list of cover sites few men- state and agreeing on cover would
be, oiough some dlsagraeniem on attributes of cover types. They have been m the field and using
imagery.
- No illusions mere are not going to be holes where we won't be able to bring
classification down to cover types
- Similar cover types and non-contrasting landscapes (such as in NY and PA) create
problems
» Unclear how FGDC vegetation subcommittee wfll fit in with GAP. A lot of FGDC wfll not work
because it does not represent ecosystem-building blocks. FGDC should be flexible enough to
adopt other programs.
- VicKlemas: We looked at FGDC system and mey don*t break om enough information.
We can not adopt their system totally.
The dilemma is there wfll never be a system mat everyone can buy into. We all have different
needs for our own purposes. More realistically, we have to come up with tiered and layered
categories to meet many needs.
-------
Standards taking a data base approach with different variables and display in map format
and statistical calculations. The expanding technologies of data base Tnanagement are
expanding our utility to do this and will be a great equalizer.
Bailey has in print a revised ecoregion map. We aggregate to it in data base format, and
it is a back end issue for analysis.
Standards are important, but being flexible m usuig dirTerent standards is also important.
GAP is interested in cross-walking from one approach to another, we need consistency
in labelling and describing features from one agency to another, but we are not limited
to one description for a piece of ground.
Thaddeus: There has been much talk of thematic consistency, but mere also needs to be
consideration for scale consistency. The scale of observation/measurement has a
effect on bow the feature is described. Simple example - a woodland at one
icalewoiu^l be a mosaic of woody patches ma herbaceous matrix at a finer sca^ Mike:
We should be able to extract map at any scale but we have to agree on data base
standards to do mis.
Vic: C-CAP is developing classes needed for C-CAP, but other classes are possible if
states need them for other purposes. The final test is whether we can cross-match
Minimum PMIP MM* 100 hectares,
- the data bases retain more detailed information, mat can be analyzed and displayed.
ares for viable assemblage systems to contain lull complement of plants and
— Map polygon in Bart are «h«wn «g mirt^rgtj pi California Qftahfx? fhnpat gfhigh
primary features which occupy total polygram — modeler approach.
This is debated at every GAP meeting.
- KathyrUSGS land use progcam piper map and digital and done at 4 hectares. Everything
done is kept in data base not just printed product
- Vic: C-CAP determined data base could maintain 90 x 90 meters for minimum detection
unit There are users for whom 2^ acres aren't good enough That is ok if they can't
aggregate into database.
- MY GAP expressed reservations about retaining information on smaller features with
spatial aggregation.
GAP analysis underwent review by Academy of Sciences; criticized mat mere was not enough
emphasis on accuracy assessment. Now need to provide performance rating for map products.
A workshop was held in Santa Barbara to develop accuracy awKsmftnt approach for GAP.
- Map doesn't need to meet standards to be accepted. The map just needs to have a
reported accuracy. Results reported with analysis to inform client what they are dealing
with.
- Three options based on funding:
1. high access all locations to cover type level have been assessed,
2. medium access accuracy with standard protocol with level 4 collect as much info
at mat level reasonable echoer to 5th level,
3. least cost use air photos only.
- Sampling units will be regular shapes, 14m size unit. Actual shape would be regular
but don't know what shape would be. Flexibility to change sample shape because of
advantages to varied shape or different orientations. We will improve accuracy by
stratifying in ecoregions to ensure adequate sampling.
— » Decision to buffer borders, and sampling only units mat fall complete within polygon
(interior sampling). Use random distribution who increased sampling for rare types.
-------
Getting into field to sample miy be difficult due to legal issues and terrain accessibility.
There was agreement mat we needed a field based approach. Include what percent
inaccessible and what fell on boundary buffer areas. Compare map polygon attributes
infield and with air photos.
GAP accuracy flgg*^ifni*nt wfll be on mapped products that are derived from the database.
Accuracy assessment wfll not be used to validate the database. The ongoing field work
by cooperating programs may have value to database validation.
METADATA STANDARDS
Chris Cogan provided a presentation on the GAP metadata content standards. Overheads used in his
afp faclwfftd far Approfe M A copy of the GAP metadata content standard is included in
Appendix N.
GAP standards on derived from FGDC standards, with nine categories.
- FGDC guidelines have 250 guidelines. GAP has added some and dropped some, but
essentially homogenous with FGDC guideline*
Purposes of metadata: needs to be carried with database, able to inform the user about data
fitness, the responslbflrry of the data provider, developed using a shared terminology.
issues relating to metadata format and management:
metato digiod text fflecimernal to the GIS data tyen,av^ Trying to
keep this internal to the data set so fc cant be lost,
- USGS gofagmtoAixdnfo INFO ffleniethod of storing data 0- But users without Arclnfo
wfll need to be able to view metadata.
— them are a lot of «nft«Mne dutf em handl* tn«it«y» tm*«d«fr mA frfr fr p^ uHfattS JP
o Future directions: user interface hnprovementx, buflt in documentation, automated MD
processing, Migration of previous MD to new schema.
o Conclusions: current and future standards are necessary, cost effective and efficient data sharing,
responsible data documentation produced by provider.
o Need proactive approach to antiripatf what future users wfll require regarding
o There has been sharing of *?T*tt^ifti across the various agencies. This *g»«tfcnphjp should be
ffnhinr.fld in the future.
DISCUSSION OF LAND COVER
Much of the remainder of the session revolved around an open-ended discussion of landcover types, in
particular agricultural classes - the ability to extract from remotely sensed data, and the difficulties of
incorporating land use in land cover systems.
-------
o We are trying to track change at die ecosystem level, without consistency we cannot track any
change. As time goes on we warn to tie our data together statistically. As long as we can layer
data we can characterize landscapes but we need to show attributes specifically.
o Neither C-CAP nor GAP are limited to satellite data only.
o GAP will not typically provide detailed agricultural information, but NY is interested in cultivated
lands as habitat
o Critical thing is for subgroups to nest into larger groups.
o Some wfll do watershed modeling and wfll care what type of agriculture is around watershed.
From a water quality perspective it is very important
o USGS H«.f}firTrinn is dose to C-CAP. We are attempting to breakout grassland and other
developed agricultural areas. SCS breaks crop land into four categories, and the kind of crop
into about 20 categories, corn and use of corn. SCS Vf«!tffift cropping system mat is mere to
- Bob - Crop land would not be put under cultivated cropland for us. FGDC starts out
with cultivated cropland as mere first category.
- GAP has yet to resolve issue of tree plantations as either forest or agricultural land.
o C-CAP has found «""ti«r janes m separating wffg**1* wetlands *"d estuarine woody wetlands.
There is a need to separate out land use from land cover - example, mapping land'cover by actual
cover in urban land me vet.
Need to determine if (eared to cover and don't add another layer.
- Developed land as a cover is not appropriate - ft is usually trees, shrubs, etc. mis has
to be sorted om to get a dear classification. There is a whole ecosystem mat is ignored.
Loveland: row and field crops, forage and pasture can probably be distinguished and mapped.
Mapping can combine forage and pasture and have attributes reflecting nuMgamunt practices.
Orchards, vineyards woody crops- can we do this with unsupervised TM7 But can bring in SCS
and local and regional information. On a state by state effort can probably do a quite bit on a
national level more difficult I would like more detail at mat level. Currently have subclass
under row and field, forage crops and pastures.
Rather men identify crop type, irrigated component of land is important to SCS. The irrigation
systems wfll have impact on some areas.
What we map today wfll not determine what we do five years from now.
FUTURE DIRECTIONS
o Denice: Where do we go from here to insure consistency and explore opportunities for new
agencies and partnerships, etc.
LCD Mangiracina (EPA Region 3,215-597-6666) wfll serve as facilitator. He wfll be in contact
-------
with everyone and can schedule ""»**h«f«. etc.
- Leo: I would like to compile who is doing what so that we can't coordinate our efforts.
I am looking for opportunities to get the ball rolling for everyone's needs.
- Denice: Coordination wife the states wfll be very valuable. We would like to have photo
data available and be able to use field activities to increase the database. We want to get
you involved with people that have not been involved before.
- Rachel: Maybe we could have a few tests on data transfer and see how it works. This
would be a small step to get us used to this scenario. We have quite a bh of data that
could be available.
What are me forces that hold MRLC together? What wfll be able to do to keep mis spirit alive
in the administration,
- Leo would be one
- having meetings would be another (meeting 3 times a year might be a good tiling)
- Ponding has been the major hurdle. The easy part is deciding who gets to do what with
the data.
- Tnis is an huerageocy type of approach that wu^brfcu;p
— MR1£ as a process aiid regions wim similar problems aiid to ine^
What wfll attract people ID come to these meetings and stay involved: data, derivative data,
ground trnoX research needs. BobSnrim: But other programs and agencies need to know exactly
what benefits wfll accrue so they en sell the concept to their management
- There is no way individual progrra co^
Bat this can't be a one shot thing, we need to demonstrate the long term need.
TomDeMbn: Ifhnikioivtemeoitiniiya>pen^
to date, backed up by agreements with agendas. This will secure your longevity and feel
you should do this as a larger group. There is a need to define a client - when this is
done, then mere wffl be greater success in lobbying for coatnmed support and funding.
- Altfaoiigh the MRl£d<>esliaveagTeementt, they i^
to be done specifically, in order to ensure the life of MRLC.
-------
APPENDIX A
-------
-------
r£B 25 '94 li:59flM CP« O-MP P.3/6
MRLC REGIONAL MEETING
March 2-3 1994
Annapolis, VA.
2 March
Wednesday
10:00 INTRODUCTION (SHAW)
MRLC
PARTICIPANTS (EMAP.NAWQA, C-CAP, GAP, NALC,
EDO
10:30 MEETING OBJECTIVES (JENNINGS)
11:00 MRLC GOALS AND OBJECTIVES (LOVELAND)
LUNCH
1:15 PARTICIPATION
GAP MID-ATLANTIC STATES (JENNINGS)
C-CAP (FIELD)
NAWQA (?)
EMAP RESOURCE GROUPS, MAHA PROJECT (SHAW)
NALC (WORTHY)
USGS LANDUSE/COVER (LINS)
POTENTIAL COOPERATORS (SHAW)
MRLC REGIONAL COORDINATION (SHAW)
3:15 BREAK
3:30 DATA STATUS AND PROCESSING
DATA STATUS, PROCESSING (HOOD)
TM SCENES (Selection, Status) (HOOD)
MSS TRIPLICATES (WORTHY)
CLUSTERED DATA (BENJAMIN)
-------
25 '94 12:00PM EPA O-ttP P.4'6
5:00 ADJOURN
3 March
Thursday
8:00 STANDARDS
CLASSIFICATION SYSTEM (JENNINGS)
FIELD VALIDATION (JENNINGS)
METADATA (COGIN)
Others (COGIN)
9:30 BREAK
9:45 SPECTRUM PRESENTATION (BENJAMIN)
10:45 IMPLEMENTATION DISUCSSION (SHAW)
12:00 MEETING SUMMARY
12:30 ADJOURN
-------
APPENDIX B
-------
-------
MRLC REGIONAL MEETING
LIST OF INVITEES AND ATTENDEES
Note: Attendee generally mt'"***** by pretence of phone number tad email mddress, in addition to
Peter Avers
201 14m Street, S.W.
Auditor's fiaOding
3rd Floor Sooth
Waatington, DC 20250
MarkAyers
Long bland and New Jeney
Coa«al Plain Study Unit
810 Bear Tavern Road, Suite 206
We* Trenton, NJ 08628
Thaddeos J. Baa
2 Triangle Drive
P.O. Box 12313
Research Triangle Fade, NC 27709
none: (919) 541-2755
Fax: (919)541-4958
tfaBnOerdgsvj4nc.ept.gov
US Geological Survey
NASA Ame* Reeatrch
MS-242-4
Moffctt Field, CA 90435
(415) 604-3914
JohnBrakebOI
Potomac River Baain Study Unit
208 Carroll Building
8600 LaSalle Road
Towion, MD 21286
Phone: (410)828-1535
Fax: (410) 828-1538
jwbnkebOusgs.gov
Kevin Breen
Lower Susqnebann* River Baain Study Unit
840 Market Street
Lemoyne. PA 17043-1586
Dr. Grace Brush
at of Geography
; Hopkins University
Baltimore, MD 21218
-------
Mafk H. Cleveland
EPA, R HI (3WM11)
841 Chestnut St
Fhilaridphii. PA 19107
Phone: (215)597-8226
Fax: (215) 597-3359
Cbnctopber Cogia
Dipt of FM A. Wildlife
Umvauty of Idaho
Moscow, ID 83844-1136
Phase: (208)885-5788
Fax: (208) 8854080
Stopbeo D. DeGlom
SCAS/CFE/CLEARS
158EmenonEUl
OooMO Univemty
Ufa**, NY 14853
Voice: (607)255-6328
F«: (607) 255-6143
TomDeMoH
US pA
Rjfiao 3, Powv T«cb
201 DataK HjflnMir, Roate4SO
MD 21401
Pnl
USDA-ARS
Bwewwt Bafldtng 7 KM
BetevOk,MD 20705
DnFanow
Strategic Environ. Ass. Dinctor
1305 Etft/WMt Highwiy
SSMC4, 9tb Floor
Silver Springs, MD 20910
Donald W. Field
National Marine Fuoeries Service
NOAABesnftxtLeb
101 Piven Island Rd.
Beaufort, NC 28516
Pboae: (919)728-8764
Fax: (919) 728-8784
-------
Gary Fisher
Potomac River Bum Study Unit
208 Curoll Building
8600 LaSalle Road
Towno, MD 21286
Pbooe: (410)828-1535
Fax: (410) 828-1538
Internet: gfBdMfOutg8.gov
Ms. Bess Gillelan, Cbitf
NOAA/OeaspeakB Bay Office
410 Severn Avenue. Suite 107A
Annapolis, MD 21403
H. Grossman
lue Nature Conservancy
1815 North Lynn Street
Arlington, VA 22209
USD A Forest Service
Forest Inventory ft Analysis
5 Radnor Corporate Center
PO Box 6775
Radnor, PA 19087
(610) 975-4047
'Fax: (610) 975-4200
Internet: fa^s-r.beiriiey/ow«s244MisJttiiisiLcain
Joy J. Hood
Hughes SIX Corporation
South Dakota Operations
EROS Data Center
Sioux Falls, SD 57198
Michael D.
?<**T|" Cooperative Fish and
Wildlife Research Unit
University of Idaho
Moscow, ID 83843
Phone: (208)885-6336
Internet- jmningsflEiridiho.edu
Dr. Vic Kfanas
University of Delaware
Center for Remote Sensing
College of Marine Studies
Newark, Delaware 19716
Phone: (302)831-8256
Fax: (302) 831-6838
-------
Frank Koenig
U.S. Fore* Service
(Eattam Region)
180 Canfield St
Morgntown, WV 26505
Phone; (304)285-1536
Fax: (304) 285-1505
514 USGS NaTl Canter
Raton. VA 22092
Phone: (703)648-4535
Fax: (703(648-5585
UiBC.iiais.fav
iLoveland
U.S. Geological Survey
EROS Data Ceotar
Sioux FaDt. SD 57198
Phone: (605)594-6066
Fax: (605)594-6589
Iovdcodbedecflwl9.cr.iucs.sov
Leonard Mangiaraeiai
US EPA
RafiaoUI
841 ChaatoBt Stnat
PUhdalplua, PA 19107
(215)597-6666
Steve McAnley
i Study Unit
810 Bear Tavern Road, Suite 206
We* Trenton, NJ O8628
Dick M^^rfrd
Delaware Eftuary Project
US Fiab and Wildlife Service
RD#1, Box 146A
Smyrna, DE 19977
Roetlmd Moore
US EPA
Region IV
345 Courtiand St., N.E.
Atlanta, CA 30365
Dr. John Morgan
Dept of Geography &. Environ. Planning
TowBon State University
Baltimore, MD 21204-7097
-------
Dr. Wayne Myers
124 Land & Water Reaeuch Bid*.
Panuylvania Slate Univenity
Univeniry Paric, PA 16802
Phone: (814)863-0002
Fax: (814) 865-3378
Internet wfanOpaovm.piu.edu
MatkOlaan
US EPA
MD 3405*
401 M Street S.W.
Washington, DC 20460
TomPheifier
U.S. EPA
4th & M Sto. S.W.
Washington, DC 20460
Phone: (202) 2604723
Fax: (202) 260-2159
AmRasbeny
MD Dept of National Resources
Tawe* Stale Office Bldg.
Annapolia. MD 21401
Voice: (410)974-3195
(410) 974-3587
Mflo
New Yo* Cooperative Fiah & Wildlife RaMatdi Unit
Dept. of Natmal Reaooroec
Cornell Univeraity
Ithaca, NY 14853
Phone: (607)255-2151
Fax: (607) 255-1895
Deaice Shaw
US EPA
EMAP Center
MD-75
Raaearch Triangle P«rk, NC 27711
Harvey Simon
26 Federal Plan
Room 900
New York. NY 10278
Phone: (212) 264-1361
Fax: (212)264-9695
-------
diaries Smith
NYCFWRU
FeraowHall
Coneli Umvenrity
Ithaca, NY 14853-3001
Voice: (607)255-3219
Fax: (607) 255-1895
Intenet: cnOcomeU.edu
Robert E. Snath, Jr.
USDA
Soil CauKrvitian Service
Reeouraei Inventtny & GIS Div.
P.O. Box 2890
Washington, DC 20013
Phone: (2Q2)72£M452
PCX: (202) 690-2019
Intenet: meftmhq trt fnv.
Tunothy D. SpnriD
Alb«Mrie-P«nIico Study Unit
3916 Saoaet Ridfe ROM)
27607
NAWQAPtafiam
U.S. Geolofiei
CA 95825
(916)978-4645
Rebecca Wajd*
VA Dept of Game & Inland Fufaerie*
4010W. BnMdSL
Richmond, VA 23230
Phone: (804)3674351
Fax: (804) 367-2427
JeffWaldeo
Fiab *pA Wildlife Info F^"***>f*
Vu^inuTech
2206 S. Main Street, Suite B
Bbcksbmy, VA 24060
DcveWert
US EPA
841 Chestnut Building
3PM53
,;. PA 19107
Phone: (215)597-1198
-------
L. Dorsey Worthy
US EPA
EMSL-L«s Vegts
P.O. Box 93478
LuVegu. NV 89193-3478
Phooc: (702) 798-2274 or 2200
Fax: (702) 798-2692
Internet: amdldwOvegasl.la*.efNUfov
-------
MRLC Consortium
Documentation Notebook
May 1995
SECTION 15
MRLC DATA RECIPIENTS
This section contains information on the organizations and projects which have
received data through the MRLC Consortium. There are two subsections:
15.1 Summary listing of MRLC data recipients
15.2 Details of MRLC data use by organization and project
-------
MRLC Consortium
Documentation Notebook
May 1995
15.1 Summary Listing of MRLC Data Recipients
Recipients of MRLC Consortium Data
as of March 1995
Federal Agencies and Programs 35
State Agencies
Academic Institutions 20
MRLC Consortium Data Recipients:
Federal Agencies Represented
U.S. Environmental Protection Agency
U.S. Geological Survey
U.S. Fish and Wildlife Service
Natural Resources Conservation Service
National Marine Fisheries Service
Tennessee Valley Authority
State Agencies Represented
Illinois Natural History Survey
Kansas Biological Survey
Kentucky Dept. of Fish and Wildlife
Maryland Dept. of Natural Resources
Missouri Geographic Resources Center
Oregon Dept. of Fish and Wildlife
Tennessee Wildlife Resources Agency
-------
MRLC Consortium
Documentation Notebook
May 1995
Academic Institutions Represented
Cornell University/New York CFWRU
Florida CFWRU
New Mexico CFWRU
Oklahoma State University CFWRU
Pennsylvania State University
Texas A&M University
University of Arizona
University of Arkansas
University of California, Santa Barbara
University of Idaho CFWRU
University of Massachusetts
University of Montana CFWRU
University of Nebraska
University of Texas, Austin
University of Vermont CFWRU
University of Washington CFWRU
University of Wyoming
Utah State University
Virginia Tech
West Virginia University
CFWRU = Cooperative Fish and Wildlife Research Unit
-------
C .* V>
1 8»
C .0 —
1 - »
S o «
a^s
U .2
S^ 1
S |
3
u
O
Q
U
2
1 t*
a >^ o\
a > <*
0 C- ~
- 3 J=
a J2 a
•S 13 «a
1 gS
™ ,00
° ^ "
? 'i!
S a u w
4) f-
|1§ o
*O ••• r \ •—••
^ Q.U *-•
u a a 3
d ° « •M
2 ° Q '+=
2 8c« w
•s * ° c
° a c< ~~
J2 o W O
1 ll E
Q 5 s a)
«1 "S
— •*• (D
f-i .S£ 3 0
^2 £3 <
v~>
0\
o\
_c
"8
8
a.
X
4)
B.
>
0
0
•a
1
c
2
D
e/5
4-«
3
«
OJ
CC
•a
ii
*^ « **
CO 2
^J OT t/J
J3 ^ O
cfl ^! /^
- •= o ff,
8 -H^S
'e1^ 1 2
B.r JJ B .
Z O v)
> «
s M a § s
= § 1 s »
^* o ^"!? ^i
IB ^- »* ™ *^
1 1 S 2 §
£ 6-g^cj
Cornell University
New York CFWRU
S
Cti
.S
•i
1
K
4)
O.
^
g
U
•o
I
C
Q
in
4-1
3 m
i/i o\
OJ O\
cr »
8
I
—
'u>
>»
"e3
§
d.
CO
o
0)
(O
3
>•
n
E
i
Florida CFWRU
•
1
QQ
•g
^
1
^5
1
co
4^
3
CO
V
cc
8
'S1
o.
OT
'S
>%
o.
O
o\
oc S
8
I
.2
2.
>\
o.
5T
O
(U
3
t
n>
E
£
D
8
g
z
£
Ox
^^
.s
1
8,
s
&
s
s
•o
1
1
Q
CO
4-f
"3
CO
V
cc.
**
8
•g1
Q.
»
o.
S
. i ..*
1" 1
•sss.8
88 3 -S <8
. . ** 41 *"
CO -- ^ ft
n S ff SJ
3 s a <
s » a S
V 3>- W ^"
OC S^CU U]
o 2
S [jy
'e1 1 •§ w
O.^ ^ U4
.2 V 99 ^ S
M 'O JO ^ «
tol§!
0.3 M & B
5s 8g|
• • ^^ «*^ ^^ o
s •« 3 -s s
>i1 s 1
£• = g CL 3
2 5 •& < «>
E x: 3 5 .9
"C *^ S ^^ M
i o e w 3
^,
s
Q
ou
*4M
0
8
*a
Pennsylvania State Univer.
Office of Remote Sensing,
-------
Hi
11 -
2 O n
§ 2 S
0 c
u .2
y 3
Oi C
2 |
§
•a
J
«
2
Q
v>
S.
£si
S2
1 ^
0-5
1 1
1 S
r"
*•*
3 S
w >
£ 5
OC o
Primary use: Gap analysis program.
Other uses: Fire hazard mapping in east Texas:
cooperating with Texas Forest Service to
facilitate data verification for GAP.
£»
•ta^
1
•1§
» §
s S
*^>
< 2
<2 'S,
S o.
X ett
|2S
2="!
<, u
o-g5
v o.
3 s
to >
c 8
*K -Sr
° £
2* *o
00*°
21
& 3
a >
— a
en O
.11
-^ N
es 'S
.2 -c
!•§
£ e.
tfl
id's
"5 .2
C/l ^3
S «
en Z
Primary use: Gap analysis project.
Other uses: Urban development patterns of
Little Rock, Arkansas. Graduate research
project to improve the spatial reliability of
population density calculations from census data
State highway department visual basemap for
planning. State wetlands prioritization using
derived data integrated into GIS. Sample
selection for Breeding Bird Atlas project.
Prediction of boll weevil and gypsy moth
occurrence.
>.
eo
o
]
8
H
13
*a
O,
t/)
1
eg .>
la
1.2
< t
i a
0 C
>
'o
a
I
&
. .
(0
+^
w
a>
OC
.9
3
1
•a
• c
*8
l-a
•si
^ ?
ea g.
s ?
S-€
op
§ i
w „
> D * .
1 « 1
1 € i
CL O a
^
§
•Sfc
SS S
I **" "5
D u
^
si
1d
11
•a -a
*Jg
•M s 2
^ J *3
•§ 2 «fi
i §»a
ai-B
III.
sfif
Illi
•c S « o\
Q. ft, W —
>»
BA
ll
<*8
«f ?
« S
H •§
**" n.
D Q
-------
c .a >n
1 88
e -o —
o - ^
i£ C «
o Z S
U e
U -S
sii
S p
c
3
x
<§
1-4
2 cu
2 "S < «•
-8 §° <
« 1 c fr°
> § — e "**
w H 60 C S
-•2 j; w fe-a
x u ** \* > c
•Is -a E 8"
.mm
f Still
« " " C w § «
0.3 0 « -o ™
«gJi.?S
S&l |3*
Ill^fl
i i i M i
a: > S-z cZ £.
3
c£
*^ *^
§ £
§ U
0
^" C
0 s
3 >
in
ON
ON
"8
QJ
5
"E.
E
o
u
Cu
E
S
8
TJ
1
C
2
Q
CO
3
CO
cu
tr
0 ^
Primary use: Gap analysis project.
Other uses: State wildlife department is als
using data for watershed and hardwood fon
management.
D
ft/
M«
c ^
t^ ^>
£ S
ooU
ll
CQ feA
»w WJ
^ -S
<*. -^
^ S
D ^
>n
ON
ON
•9
u
(U
"Cu
E
o
0
Cu
ra
E
v«
U
O
u
•o
1
.
<5
E
cL
CO >N
11
i*
*> <*«
> 0
Ck« >^^
^ 8-
D Q
^.
C
Primary use: Forestry change detection.
Project for California Department of ForesI
Using four scenes from MRLC to compare
against 1990 imagery owned by the state.
2
1 &
•tt «J
A Q
CO -^
3 -a
OB 5
V3 S)
u 8
D 0
ra
. «
cS'?
2^
^J 2
CO CO
•p §
d
t5 —
6.0
&> —
S S.
u- O
« S
81
"S -s ^
J £:§
"™ co -S
c -c -3
e 8.5
Q §;g
J2 "S *
3 B£
CO « ft
«2 "g £^
oc J D
^
8
Primary use: Gap analysis program.
Other uses: Related projects with GAP
cooperators include classification of woodc
habitat and riparian zones.
iciences
£* —
•« e
S B
> •=
'a §
= '&
« •<
:i*
S) o
.2 rt
>.l
«. CO
j:s
^ Q
§
i
U)
U)
0)
"o
c
o
(D
0)
E
S
I
to
'in
8-2
O
re dj
.i I
J
S
1
p
ll
u
fa
s 1
s §
I a
z u
-------
£.* w-l
o o\
•s S <"
stz
S -5 -S
o 2 S
u S
U -2
•J 3
o£ c
2 i
a
8
Q
vd
2
>"\
.0
•s
i
X
W
in
+-•
"5
w
a)
tr
• tn 0
c « -n
.2 £ o 1?
v " « S
£ >>£ S
« -o ~; s
•*» 3 •£ _L
« <" '> H
a -2,, §
§ e H «
•5 J^.£
•3<1^
§ *g s s
« • e s
* « **- M
— 2 2 >*
2 8 S *
« ._ ^3 -O
§ -S ^ -2
u o ||
«£">i
3 _ .5 o
>- E is °
» a« 1 8
E g g g
£ £ u »
u
o
'E
u
CO
.8
s
j=
03
°£
4)
'« ^>
|J3
11
« S
Z CQ
o\
C7\
T3
C
•••
"8
i
x
s
1
£
«
S
Q
in
"5
(n
-,
o
^5
< u
4J
>-. >
U • S
=5 2
> 8.
II
II
§
CO
£
0
«
^n
8
"o
S
1
Cu
o3
in
3
>•
TO
.E
ol
<
o.
w
CO
D
Primary use: Presentations.
Photos of Washington, D.C.
2
Ul
3
Primary use: Photos of Wisconsin.
Cu Cu
w 53
coS
P u
£
«?
Primary use: Photos of coastal North C
V*
£
3
< °
0. CM
w 53
c« S
P U
.
"8
•c
1
u
^ii
0
a
•«
1
CO
+••
(/)
GC
tyj
s.
<£
'*•
^N
Primary use: Training,
Plan to experiment with classification <
types in Southern Appalachians.
2
X
r*
< °f
CU OH
w <
ss
P U
Primary use: Photos of south Florida,
Wisconsin.
. a
£Z
w <
CO 2
P w
-------
C .tt *O
Ija
1 1 >•
S .2 5
o 2 5
U c
u-S
* 1
2 |
CJ
5
U
•n
Primary use: Water quality monitoring.
Comparing water quality data with adjacent lam
cover classified from MRLC data. Study areas
in Gulf of Mexico and Pensacola, Florida.
<
a.
W j
CO &
D U)
•8
•c
3
V)
•fe*
t>
>,
"o
c
t^
1
(/}
W
D
V)
0)
OC
Primary use: Water quality monitoring.
EMAP Surface Waters Project. Correlating
surface water quality with adjacent land cover
classified by MRLC. Sampling sites in
Northeast, Mid-Atlantic, and Pacific Northwest
CO
~&
D (A
a
Primary use: Presentations.
Photos of Delaware.
Other uses: Digital data to be used for progran
including MAIA, MAHA, Delaware-Maryland
Coastal Bays Assessment, Chesapeake Bays.
< m
«•!
CO U
P C4
1
i
(4
'&
8
o
<4-r
O
I
oi
w>
>•
03
e
£
< "*
63.1
CO tf
P OS
-------
e ^ w-\
is o o\
.2 o ch
c •& ™~
1 S 5-
o 2 S
u S
U -2
_) 3
OS c
S |
1
a
1
<
<*.
o
1
o
a.
0)
CO
3
>•
fo
E
al
Q
Qi
O
UJ
as
d.
C/3
0
< *
&.I
CO u*
P OS
Primary use: Photos of southern Ohio.
< "Q
uj 2
,„ °°
CO ,
Q
UJ
< •*
^ S
U4 .S
co S1
P as
Primary use: Land cover accuracy assessment.
Photos of eastern Virginia. Cooperating with
Department of Defense.
^
g
co
U.
CO
< s
&«
co .3
P >
t~
£
.3
•o
u
K
CO
O.
fc
1
•a
i
-------
= 1 i
c -o —
§1$
C3 cS
U -2
tt 1
a
Q
1
CO
1
o
£
+-*
3
w
0)
(C
0.
o
T3
1.1
f 1
« es
3 U
Primary use: Soil mapping.
Plans to use Landsat imagery as vis
for soil mapping projects in North <
u
^*
'fc
u
CO
c
o
SJ
JJ
co '"5
P CO
— ex
C o
U E.
6 3
oo y
i U
E oo
il
II
§1
s —
« 1
a-g
3 -a
10 n
flj si
en -a
wi
C.
^ O
e o_
4) ••
tM ""^
to o **•
•3*0.
3 00 g
1 -1 3
I] ;
w « eo
TO O !
ct £ -S
s
•^
c
u
CO
1
U-
CO «
0 g
co 5
P Q
o
U5
u m u
^ s-s s
§1 s °
S5 S i 5 .
1> O O « C
Primary use: Land cover mapping r
Researching methodologies for regi
land cover mapping using TM and
sources. Designing a database of d
facilitate client-specific classiflcatio
—
c
U
2
Q
co co
0 0
CO &.
P UJ
•§ .
« to
§8
Primary use: Land cover mapping r
MRLC Monitoring and NALC proj
u
8
U
3
&
88
CO Pi
P W
Primary use: Presentations.
MRLC program administration.
u
8
u
«
Q
CO CO
0 0
CO pi
P U
&o ^
*00 00
J T •§.
-1H
•2 -c o -°
*sl2^
'g J g a '«
1 o S -i o
• w "^ S ^
8
u
00
.S
o.
Cu
»
8
.S
§
co U
Primary use: Land cover mapping.
For NAWQA.
jo
8
u
a
i
If
P CO
-------
ej*
o
o
T3
«
,3
•
CO
E
ol
V3
<
00
Z
c/;
O
T3
§ • -
,2 . tn --
i
H
Si
^
a
o
"3
o
en
s
"3
i
oo
2
S1
13
r
"a
u
2
00
1
CO
O
oo
3
-C
'5
8
JD •
;T"
•J«*
1
a
CO
a
CO
D
a
•3
s
0
o
•a
J
•a
CO
8
3
^O
ts
0
"en
en
CO
u
h*
CU
IS
NAWQA.
o
u_
B
O
2
'c
i
cj
Q
en
cu
en
D
CL
.C
4-<
O
-o
CO
00
c
1
o
u
5
CO
UJ
T3
a
1
00
c
rocess
D.
U
OO
s
5
8
PI
CO
o
u
"o
'•E
<
en
U
£
3
S
"ca
CJ
o
T3
i
c?
0
Q
s
H
00
'in
3
tn
"3
en
£
1
_4>
"«
1
o_
<
U
5
00
"5.
ca
S
1
o
U
•a
j
^^
"w
'5
_B"
CQ
"s"
CJ
a
t#*
^O
1
s
a,
1
oo
Q:
83
X
E
1
CO
CO
3
V)
0)
'o
c
(U
O)
(D
•»-•
(0
•4-"
W
i
's1
o.
en
'en
>>
•3
s-
o
Primary use:
B
_o
i
>> s
S «S
M
OO 0
2 2*
en 5)
2 8
_ oo
2 2
5 CQ
CO w
Z,o
.S2 v«
is
= a
00
: Land cover mapping project
rson Level II classification; ftmdi
EPA grant to two Kentucky
8-§,
J^
0
I
4)
1
1
=3 ob
M
.8j
Is
eo "3
•R|
If
o S
» 1
eu u
.8 "1
B|^
l>6l
1
CO
t»
>,
•5
a
•«
o
^
CO
+— '
"5
CO
cu
CC
"o «
E a 2
3 >* «
Q
-------
c j* v/i
E o o\
.2 o o*
a I
C* C
s i
1
1
o
*
s
Q
in
*-•
D
U)
(U
OC
. O "O
w in o a
r il
2 S3 g >
-^ 6 S 'g
i o s -3
u
CJ
0. "
< ^
O Sd
•c-g.
3 »
S 5>
^ S
2 u
c r-
cs O
— ON
I- « 0
4) > D
> ~ a.
O «a x
O "i5 4)
u
'C
oo
s-
« 3 e
3 g. S
"> c >
a; o o
OC u u
'§"
a. •=
n
i
§•
o ,
S fe •§
•I •£ 1
dl O oo
-Q
-2
'5
.o
i
.0
T3
I
£
O
ents
(D
O
(U
DC
"ro
c
o
2?c*
oo
------- |