United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
P.O. Box 93478
Las Vegas NV 89193-3478
May 1993
EPA/600/S-93/ 005
vvEPA
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-
raphy and human features.
Typically, changes in land
cover occur when agricul-
ture/pasture is converted
to urban, or forest to
agriculture/pasture. The
results as to type of land
cover and change in land
covers will be valuable as
input to U.S. Global
Change Research Pro-
gram (GCRP) measure-
ment 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, quan-
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.
The NALC Project is focused on accomplishing
several objectives, including: ,
1) Collection of 1992 Landsat iMiiltispectral Scanner (MSS) Data to
develop an archive of recent observations;
2) Assembly; of three-date geqreferenced data sets, called triplicates, for
GCRP analysis and fefrosp'ective evaluations of change;
3) Development of methods to complete land cover categorizations and
detection of change in land icover types overtime; and
4) Analysis of NALC products [for global change research studies by
establishing a series of agreements.
Research Products include: ,]
a) Standard MSS data triplicates for the majority of the North American
Continent; j
b) Development of standard land cover and land cover change
data processing methods;
; c) Border-to-border 1991 +/- o'ne year land cover products for priority
locations in North America; [and
d) Land cover change products for the 1970's, 1980's, and 1990's.
-------
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 1cis
an image product called
a Reduced Cloud Cover
Composite (RCCC) from
3/03/92 and 4/04/92. In
Figure 1a the arrow
points to a coastal man-
grove area and the pink
color of healthy, vigor-
ously growing vegetation
in this False Color Com-
posite (FCC). In Figures
1b and 1c the pink color
is mostly absent and is
Indicative of a sustained,
decrease in plant vigor
identified by local scien-
tists (arrow on Fig. 1c).
The presence of forest
fire burn scars (top arrow)
and a new reservoir
(bottom arrow) are
marked in Figure 1b.
Figure 1b.
Figure 1c.
-------
BACKGROUND
Global change results from
alteration 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
influence of human contri-
butions 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
land cover conditions and
their change will be re-
quired. To meet these
requirements the U.S.
Environmental Protection
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 they
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
The NALC global change
research project at the Las
Vegas Environmental
Monitoring Systems
Laboratory (EMSL) is a
component of the Office of
Research and
Development's (ORD)
national program on global
change research. The
effort is being conducted
as part of the National
Aeronautics and Space
Administration (NASA)
Landsat Pathfinder Pro-
gram of pilot studies. The
goals 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 (EDO)
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
on the data products to be
generated and organized
into data sets for use in
GCRP activities. The
specific research and
development tasks in-
clude: a) acquiring
Landsat MSS images with
less than 30% cloud cover
during 1992, b) assem-
bling the individual scenes
from 1973,1986, and
1991, plus or minus one
year, to be used for gener-
ating coregistered "tripli-
cate" 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). Procedures
have been developed to
create high quality
georegistered images in
which systematic correc-
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.
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.
-------
DATA 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
methods for Landsat data
analysis 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
North American continent.
These standard ap-
proaches also address
important Agency issues
related to data Quality
Assurance/Quality Control
(QA/QC), such as data
validation.
Pilot studies will test and
determine the standard
land cover (LC) categori-
zation procedures for the
project. Methods develop-
ment projects will evaluate
approaches using study
areas in forested, agricul-
ture/pasture, and cloud-
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 1993 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
Laboratory - Las Vegas,
and 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
EMSL LAS VEGAS
, EyR|)S, D AT A t| C^.N T E.R.BBH^
-------
REDUCED CLOUD COVER 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
3a and 3b.
Figure 3c.
-------
f *
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.
Figure 4a,
Figure 4b.
Printed on Recycled Paper
•tr US GPO: 1993 — 780-520
------- |