lU United States
Environmental Protection
Agency
Office of Research and
Development
Washington DC 20460
EPA/620/R-95/003
June 1995
Mid-Atlantic Landscape
Indicators Project Plan
Environmental Monitoring and
Assessment Program
-------
-------
EPA620/R-95/003
June 1995
MID-ATLANTIC LANDSCAPE INDICATORS PROJECT PLAN
ENVIRONMENTAL MONITORING AND ASSESSMENT PROGRAM
by
William G. Kepner3
K. Bruce Jones*
Deborah J. Chaloud"
James D. Wickhamb
Kurt H. Riitters0
Robert V. O'Neilld
" U.S. EPA, Characterization Research Division, Las Vegas, NV
b Desert Research Institute, Reno, NV
c Tennessee Valley Authority, Norris, TN
d U.S. DOE, Oak Ridge National Laboratory, Oak Ridge, TN
National Exposure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
714MSD94
Printed on Recycled Paper
-------
NOTICE
The United States Environmental Protection Agency (EPA), through its Office of Research and
Development (ORD), prescribed the research described here. It has been subjected to the Agency's peer
and administrative review, and it has been approved as an EPA publication.
Mention of trade names or commercial products does not constitute endorsement or recommend-
ation for use.
11
-------
CONTENTS
Page
Notice ,. ....." ii
Figures ... iv
Tables jv
Abbreviations and Acronyms v
Executive Summary vi
Acknowledgments viii
1.0 Introduction 1
1.1 Overview of EMAP '.'.'.'.'.'.'.'.'.'. I
1.2 EMAP Indicator Development Strategy 1
1.3 Overview of EMAP-Landscapes 2
1.4 EMAP-L Monitoring and Assessment Approach 3
1.5 Landscape Approaches to Integration 4
1.6 Challenges to EMAP-L 5
2.0 Study Objectives and General Approach '. 7
2.1 Project Objectives 7
2.2 General Approach ; 7
3.0 Study Site Description 10
3.1 Site Selection/Location 10
3.2 Stressor Context/Relationship to Other Geographic Initiatives 10
3.3 Existing Data Sources 14
4.0 Project Components ; 16
4.1 Sensitivity Related to Statistical Properties ; 16
4.2 Sensitivity Related to Ecological Condition ,! 17
4.3 Comparability of and Synergism Among Different Remote Sensing Imagery 21
4.4 Future Practical Applications 23
i
5.0 Assessment of Status and Trends , 24
5.1 Chesapeake Bay, FY95 L 24
5.2 Mid-Atlantic Project Area, FY96 25
5.3 Data Bases of Landscape Statistical Summaries 26
6.0 Information Management j 27
6.1 Data Acquisition and Documentation 27
6.2 Data Analysis 27
6.3 Reports 28
6.4 Technology Transfer 28
7.0 Quality Assurance : 29
7.1 Data Quality Objectives .. '. 29
7.2 Audits of Data Quality ....29
7.3 Data Quality Assessments 29
.
8.0 Project Outputs and Timeline 30
9.0 References 32
m
-------
FIGURES
Number
1. Three-step Landscape Monitoring Approach ....... .. ............. ........ . ................ ................... ..... 4
2. Mid-Atlantic Landscape Indicators Project Area ..... ............ . ........................ ............. ........... 8
3. EMAP Sampling within Major Drainage Basins in the Mid-Atlantic
Region [[[ ..................... ............... ...... : ..... 13
4. USGS 8-digit Mid-Appalachian Watersheds ............. . ....... . .............................................. - 19
5. Sources and Flow of Data to be Used by EMAP-L in Conducting Research
and Assessments in the Mid-Atlantic Region ...................... ........... ............... ....... ................ 28
TABLES
Number ' Page
1. Societal Values, Example Indicators, and Candidate Metrics '. - 9
2. Number of EMAP Sampling Points by State in the Mid-Atlantic Region,
1990-1994 -. 9
3. Example National Monitoring and Information Systems Containing Ancillary ,
Data Available to EMAP-Landscapes 15
4. Approximate Schedule of Land Cover Data Base Availability in the
Mid-Atlantic Region 27
5. List of EMAP-L Products Anticipated From the Mid-Atlantic Region Landscape
-------
ADQ
AVHRR
BBS
C-CAP
CCRS
CDF
DQA
DQO
ECDMS
EMAP
EMAP-L
EMAP-LC
EPA
EROS
FIA
GCRP
GIS
HUC
ffil
Landsat-MSS
Landsat-TM
LUCAS
MAHA
MAIA
MAIAGRD
MRLC
NALC
NASQAN
NAWQA
NBS
NCBP
NOAA
NRC
NRI
ORD
QA
QAMS
QAPP
QMP
RASA
REMAP
SAB
SAMAB
STORET
TIME
TVA
USGS
ABBREVIATIONS AND ACRONYMS
Audits of Data Quality
Advanced Very High Resolution Radiometer
North American Breeding Bird Survey
CoastWatch Change Analysis Program |
Canada Centre for Remote Sensing
Cumulative Distribution Function
Data Quality Assessments
Data Quality Objectives
Environmental Contaminants Data Management System
Environmental Monitoring and Assessment Program
Environmental Monitoring and Assessment Program - Landscapes
Environmental Monitoring and Assessment Program - Landscape Characterization
Environmental Protection Agency
Earth Resources Observation System
Forest Inventory and Analysis Program
Global Change Research Program
Geographic Information System
Hydrologic Unit Code
Index of Biotic Integrity
Landsat-Multi-spectral Scanner
Landsat-Thematic Mapper
Land-Use Change and Analysis System
Mid-Atlantic Highlands Assessment Project
Mid-Atlantic Integrated Assessment Project
Mid Atlantic Integrated Assessment Geographic Reference Data base
Multi-Resolution Land Characteristics Consortium
North American Landscape Characterization Project
National Stream Quality Accounting Network
National Water Quality Assessment Program
National Biological Service
National Contaminant Biomonitoring Program
National Oceanographic and Atmospheric Administration
National Research Council
Natural Resources Inventory
Office of Research and Development
Quality Assurance
Quality Assurance Management Staff
Quality Assurance Project Plan
Quality Management Plan
Regional Aquifer-Systems Analysis Program
Regional Environmental Monitoring and Assessment Program
Science Advisory Board
Southern Appalachian Man and the Biosphere Program
Storage and Retrieval System
Temporally Integrated Monitoring of Ecosystems Project
Tennessee Valley Authority
U.S. Geological Survey
-------
EXECUTIVE SUMMARY
INTRODUCTION
Landscapes are described by the spatial arrangements of ecological resources. The Environmental
Monitoring and Assessment Program Landscapes component (EMAP-L) was initiated in late 1992 and issued
its research strategy in 1994 (EPA/620/R-94/009). In the 1994 research plan, EMAP-L selected an approach
that examines landscape patterns relative to their affect on the flow of energy, water, nutrients, and biota. The
EMAP-L Mid-Atlantic Landscape Indicators Project Plan proposes a set of research initiatives to resolve key
technical and operational issues first identified in the 1994 research plan. The issues are especially focused
at identifying, testing, and evaluating landscape indicators. Secondly, EMAP-L will test its ability to generate
ecological assessments of landscape status and trends of selected societal values at multiple scales, i.e.
watershed and region. The selected societal values are biodiversity, watershed integrity, and landscape
resilience.
The Mid-Atlantic Landscape Indicators Project Plan has been divided into eight sections which describe:
(1) the conceptual basis of the landscape approach including societal values, framework for indicator
development, proposed technology using a three-step monitoring process, and relation to other EMAP
components and objectives; (2) objectives and approach for the Mid-Atlantic project; (3) study site
description, relation to other geographic initiatives and.programs, and existing data sources; (4) a series ot
studies to address research and development issues related to indicator sensitivity, synergism among remote
sensing data, and landscape pattern assessment; (5) the conceptual approach for determining landscape statas
using 1990 Landsat-TM data for the Chesapeake Bay Watershed and landscape status and trends for the entire
8-state Mid-Atlantic region using Landsat-TM and multi-date Landsat-MSS; (6) reporting formats,
technology transfer, and data acquisition, analysis, and documentation; (7) application of quality assurance
principles to remote sensing and GIS data analysis; and (8) anticipated products and approximate completion
dates.
The Mid-Atlantic Region Project provides EMAP-L with its first opportunity to test a number of technical
issues and assessment protocols. The Mid-Atlantic is currently the focus of a number of EPA (Mid-Atlantic
Highlands Assessment, Mid-Atlantic Integrated Assessment) and other agency regional initiatives (e.g.
Southern Appalachian Man and the Biosphere). Collectively, the location contains available spatial data sets
and provides both variable landscape characteristics and land use pattern gradients over which to test indicator
sensitivity.
The program proposes to implement its research simultaneously along two project lines, i.e. landscape
indicator development and landscape status and trend assessment. Landscape indicator development has been
further divided into project components that include indicator sensitivity and synergism among remote
sensing data.
Indicator sensitivity will examine the influence of data and formula attributes used to calculate indicator
metrics, i.e. statistical properties, and the relevance of indicator application to environmental condition over
multiple scales, i.e. ecological sensitivity.
VI
-------
Landscape Indicator Development Related to Statistical Properties
i-
Sample number, calculation window-size, land cover class number, statistical independence between
metrics, and indicator sensitivity, as a function of land cover misclassificatiom, will be examined under the
category of statistical properties. The plan emphasizes an approach which assesses the impact of sample size
(number of pccels or pixel aggregations) and the number of attributes, e.g. land cover classes by varying the
? u °J 0coverJdaSSeS *** analysis window-size "sing Landsat-TM data for the Chesapeake Bay
Watershed. Secondly, it proposes to evaluate correlations among landscape indicators at varying spatial scales
to evaluate the degree to which relationships are scale-dependent. Additionally, landscape pattern metrics will
be evaluated for their sensitivity relative to land cover misclassification versus actual changes in land cover
using a simulation model to generate an error matrix for the Chesapeake Bay Watershed.
Landscape Indicator Development Related to Ecological Sensitivity
Gradient analysis, association with ecological resource condition, and landscape classification will be
examined under the category of ecological sensitivity. The plan proposes .using an urban-to-mountain
anthropogenic land use gradient that is present along an east-west alignment in the Mid-Atlantic region to
test the statistical significance of landscape metric response to environmental change. EMAP-L under the
proposed project plan, will determine the relationship of landscape metric information relative to independent
assessments of ecological condition generated by other EMAP components that are performing colocated
research withm the region, i.e. streams, forests, estuaries. Lastly, the project plan proposes to evaluate
spatially-nested relationships of landscape pattern. The project will examine the degree to which certain
landscape pattern types and values vary by natural region and by other biophysical attributes, such as climate
soils, and topography, etc. '
Synergism Among Remote Sensors
A - T Spatial data derived from different retoote sensi"g Platforms, i.e.
AVHRR multi-date Landsat-MSS, and Landsat-TM, in terms of advantages and limitations in generating
ecologically meaningful landscape pattern metrics. Secondly, EMAP-L will compare landscape pattern
estimates derived from classified imagery with those derived from spectral-clustered and reflectance data to
determine their utility in evaluating landscape change. Both evaluations will be made on a set of watersheds
selected across toe Mid-Atlantic region. Landscape pattern metrics will be calculated from different imagery
for each watershed and with labeled and spectrally-clustered data sets.
i
Assessment of Landscape Status and Trends
EMAP-L proposes to implement landscape status and trends assessments within the Mid-Atlantic area
The program is challenged with the goal of assessing and interpreting landscape pattern data with regard to
specific societal values and assessment questions. The project plan emphasizes an incremental approach to
report protocols by first generating a landscape status assessment of the Chesapeake Bay Watershed and then
kter expanding the analysis to include status and a twenty-year trend analysis of landscapes for the entire
JVLiQ-rVtifliitic region.
Vll
-------
ACKNOWLEDGMENTS
We would like to thank members of the review panel, John Jensen (University of South Carolina) Robert
Gardner (University of Maryland), Dennis Grossman (The Nature Conservancy) and Gary McVicker (U.S.
Bureau of Land Management), for their review and constructive comments on the project plan.
We are also especially grateful for the internal EPA reviews provided by Walt Whitford, Iris Goodman,
Jim Andreasen, Marge Holland, Tony Olsen, Don Garofalo, and Arthur Spingarn.
The authors thank the Visual Information Services Staff of the Characterization Research Division-Las
Vegas forLr support in graphics, technical editing, desk-top publishing and word processuig. We also
acSwS the assistance of Ruth Christianson (EPA) for administrative and word processmg support
related to early versions of the plan.
vni
-------
1.0 INTRODUCTION
1.1 OVERVIEW OF EMAP
Throughout the greater part of this century, most
environmental management efforts focused on short-
term, local-scale problems such as pollutant abate-
ment. Environmental policy most often reflected a
reactive response to past environmental problems
rather than a more proactive, anticipatory process
(Kaufmann et al. 1994). The 1980s witnessed in-
creased interest in protecting whole ecosystems from
chronic environmental problems, but these were often
partitioned in relation to specific media, e.g. water, air,
or soil pollution (Franklin 1993). Environmental
management philosophy of the 1990s has evolved to
examine critical environmental problems over larger
spatial scales, such as regions or the Nation, and is
now assessing the cumulative risk of impairment as the
combined result of multiple stressors (Woodley et al.
1993, Noss and Cooperrider 1994). More recently,
concern over the condition of larger-scale biogeo-
graphic provinces, e.g. landscapes, watersheds, and
ecoregions, has received considerable attention and
entire science programs have been dedicated to acquir-
ing information which will relate to evaluating the
effectiveness of environmental management policies
and management actions, and allow for informed
adjustments to management practices which favor
sustainability of the biosphere, i.e. ecosystem manage-
ment (Baron and Galvin 1990, Holden 1988, World
Commission on Environment and Development 1987,
Woodley et al. 1993, Franklin 1993, Kaufmann et al.
1994, Noss and Cooperrider 1994).
In 1989, the U.S. Environmental Protection Agency
(EPA), in collaboration with other federal and state
agencies and research institutions, initiated the Envi-
ronmental Monitoring and Assessment Program
(EMAP). EMAP is a national monitoring and research
program designed to assess the condition of ecologi-
cal resources in the United States. The program was
initiated as a long-term project to periodically
evaluate ecological condition, develop innovative
methods for anticipating emerging problems before
they reach crisis proportions, and contribute mean-
ingful information to decisions on environmental
protection and management (Kutz and Linthurst
1990, Messer et al. 1991, Thornton et al. 1993).
EMAP efforts are focused on linking existing envi-
ronmental data and monitoring programs, where
possible, and collecting new information as needed
to achieve program objectives.
1.2 EMAP INDICATOR DEVELOPMENT
STRATEGY
Selection, evaluation, and implementation of
indicators are critical in establishing a monitoring
program to determine ecosystem health. EMAP
describes indicators as measurable characteristics of
the environment, both abiotic and biotic, that can
provide quantitative information on ecological
resources (EPA 1993). Specifically, an indicator is
any expression of the environment that quantita-
tively estimates the condition of ecological re-
sources, the magnitude of stress, the exposure of
biological components to stress, or the amount of
change in condition (Barber 1994). Indicators may
be either a function of a single measurement, a
multivariate statistic, or an index based on multiple
measurements. Two categories of indicators are
defined by EMAP, i.e. condition and stressor (EPA
1993). Condition indicators are characteristics of the
environment that provide quantitative estimates of
ecological resource condition, whereas stressor
indicators represent characteristics of the environ-
ment that act as sovirces of both human-induced and
1
-------
natural stress, e.g. land use and weather. Selection
criteria for indicators require that the measurement, or
index of measurements, must be responsive to changes
in condition of a resource throughout the habitat and
region for which it is employed, have clear quantita-
tive signal with minimal variance or noise, can be
simply quantified and cost effective, and produce
minimal environmental impact during sampling
(Hunsaker and Carpenter 1990, Barber 1994).
The EMAP indicator strategy has been evolving
since its inception and now includes four basic phases
for development, i.e. indicator formulation, evaluation,
implementation, and reevaluation (Barber 1994). This
framework has been employed by EMAP for monitor-
ing within all major resource classes throughout the
United States and has practical application relative to
ecological risk assessment (Hunsaker et al. 1990, Suter
1990, Suter 1993). The indicator framework has been
applied internationally, including sustainable agricul-
ture in Australia (Hamblin 1992) and other locations.
In general, the indicator development approach
involves a series of steps involving definition of the
resource population of concern or resource class;
identification and selection of societal values with
biological relevance; formulation of assessment
questions relative to the targeted resource and regional
stressors likely impacting those resources; establish-
ment of conceptual models which relate ecological
function, structure, and composition; and selection,
evaluation, and implementation of indicators and
associated sample designs. Indicators are evaluated
through simulation and pilot studies. These studies are
intended to evaluate the potential of selected indicators
to determine status and trends of ecological condition
and are the basis for selecting indicators for long-term
implementation. This process is continually open for
reevaluation as new or improved versions of indicators
are developed.
Societal values are generally considered to be those
items that serve as goods or services to sustain human
development and have ecological importance (Ehrlich
and Ehrlich 1991, McNeely et al. 1990, Regier 1993).
Societal values are the basis from which assessment
questions related to status, extent of resource, change
or trend in status and extent, and association to re-
gional stressors are crafted. Conceptual models that
describe a resource's ecological components and how
they function are developed following the assign-
ment of societal values and assessment questions.
Conceptual models are identified either from pub-
lished accounts or derivations (including combina-
tion of models) of published accounts and should
consider the temporal and spatial dynamics or
pathways of the resource at multiple scales (Wiens
1989). Conceptual models are further utilized to
serve as reference points both for the identification
and selection of indicators needed to assess the
condition of ecological resources and for guiding
data analyses or construction of multivariate indices
(Barber 1994). Indicators and indices are evaluated
for both their sensitivity relative to statistical proper-
ties and ecological meaning. Sensitivity relative to
statistical properties include indicator variability
over a sampling index period, and sensitivity to
either the number of samples and configuration of
the sample design. Determination of indicators
which most accurately define the range of ecological
conditions is generally ascertained by evaluating the
sensitivity of an indicator or a selected set of indica-
tors to gradients of predefined environmental condi-
tion or stress (Ludwig and Reynolds 1988). Prede-
termined condition is established by convention
among consensus groups, usually governmental land
management or environmental regulation agencies,
and indicator performance is evaluated for correla-
tion against existing data or a consensus opinion of
condition (see Section 4.2).
1.3 OVERVIEW OF EMAP-LANDSCAPES
EMAP-Landscapes (EMAP-L) was created in late
1992 as a distinct project within EMAP to assess
status and trends in indicators of landscapes, primar-
ily landscape pattern metrics, and to relate its results
to ecological goods and services valued by society
(i.e., societal values). This program was originally
part of the EMAP-Landscape Characterization
(EMAP-LC) program. EMAP-LC's primary focus is
on the generation of a nationally consistent land
cover data base through the Multi-Resolution Land
Characteristics Consortium (MRLC, see later discus-
sion). Additionally, EMAP-LC is the primary
source of spatial data for other EMAP projects,
including EMAP-L. Therefore, EMAP-L has a
-------
collaborative arrangement with EMAP-LC on spatial
data acquisition, land cover classification, and quality
assurance (QA) issues on all of its projects.
EMAP-L proposes to estimate status and trends in
landscape composition and pattern. Landscapes are
described by the spatial arrangement of ecological
resources. EMAP-L intends to evaluate two general
types of landscapes, i.e., those that are defined by
heterogeneous land areas composed of clusters of
interacting ecosystems in repeated pattern (Forman
and Godron 1986, Wickham and Norton 1994) and
watersheds.
Hierarchy theory provides the context for integrat-
ing multiple scales of information to determine
whether landscape patterns are sufficient to allow
ecological processes, such as the flow of energy,
water, nutrients, and biota, to operate at the necessary
scales (O'Neill et al. 1986). In essence, it states that
landscapes are organized into patterns within a hierar-
chy of spatial and temporal scales. Numerous ecologi-
cal and human-induced disturbances maintain land-
scape patterns or elicit phase transition into new
patterns. These disturbance events occur across a
range of spatial and temporal scales (Jensen and
Everett 1994). This type of framework then allows the
investigation of changes in the distribution, domi-
nance, and connectivity of ecosystem components and
the effects that might occur on ecological resources.
Secondly, this approach also permits comparisons of
condition across mixed landscapes and across different
sources of stress to cumulatively assess ecological risk.
EMAP-L has completed a research plan to guide
resolution of key technical and operational issues
associated with identifying, testing, and evaluating
landscape indicators (EPA 1994). In particular, the
program is focused on developing indicators of land-
scape pattern that relate to societal values derived from
biodiversity, watershed integrity, and landscape
resilience. Workshops and peer reviews have helped
identify these societal values related to landscape
ecological patterns and processes; however, EMAP-L
is still refining landscape values and developing
additional assessment questions, e.g. Sharpe et al.
1993, Rapport et al. 1995. Biodiversity is defined as
the variety of life and its processes. Watershed integ-
rity is defined as the capability to collect, retain, store,
and purify water. Landscape resilience represents a
landscape or watersheds ability to sustain its inherent
richness of ecological goods and services in the face
of natural and anthropogenic stress.
The program is emphasizing the use of remote
sensing and geographic information system (GIS)
technology rather than ground sample-based meth-
ods. This approach allows: (1) simplicity and cost
effectiveness of data acquisition via remote observa-
tion platforms (including use of existing data), (2)
ability to integrate synoptic measurements of land-
scape pattern across all natural resource classes and
provide interpretive enhancement to individual
EMAP projects and other research or monitoring
projects, and (3) ability to develop assessments at
multiple temporal aiad spatial scales. For example,
use of Landsat Multi-spectral Scanner (Landsat-
MSS) satellite data would permit trend analysis of
landscape pattern metrics over a period of more than
20 years. In contrast, data derived from the Ad-
vanced Very High Resolution Radiometer (AVHRR)
could provide similai: assessments over larger spatial
scales and shorter temporal domains.
1.4 EMAP-L MONITORING AND
ASSESSMENT APPROACH
EMAP-L has developed a monitoring protocol
and proposed a series of landscape pattern metrics
that allow for landscape-level assessments of ecolog-
ical condition (EPA 1994). EMAP-L assumes that
Landsat satellite or equivalent imagery for regions of
interest will be available on approximately 10-year
cycles and hence, these data will form the basis for
status and trends assessments.
The approach for assessment has been proposed
as a three-step process in which baseline condition is
established from selected landscape metrics (Step 1).
Condition status for landscapes is regenerated 10
years after baseline from new land cover data (Step
2) and the datasets are combined to determine
change in status and extent of resources (Figure 1).
Both the type and magnitude of change in Step 2
determine if more detailed assessments are to be
carried out in Step 3. Step 3 involves more in-depth
analyses of associatioitis between observed landscape
-------
Regional
Scale
Stepl
Baseline
landscape
condition
Between 10 year
change detection
using AVHRR and
other data
Step 2
Ti-Tn
Satellite-based
land cover change
detection
Areas meeting
change criteria
StepS
In depth monitoring and
association with stressors
0677msd94.1
Figure 1. Three-step Landscape Monitoring Approach.
status and trends with environmental stressors and can
be focused to address specific subsets of the broader
societal values. The program recognizes that signifi-
cant landscape changes can occur more frequently than
on 10-year increments and hence, frequency of image
acquisition, processing, and evaluation is a major issue
within EMAP-L. Other data sources derived from
remote observation platforms, such as AVHRR, are
being considered to improve temporal resolution.
Although AVHRR spatial resolution is much coarser
than Landsat-MSS or Landsat Thematic Mapper
(Landsat-TM), its ability to detect major land cover
changes resulting from human or natural disturbances
may be sufficient to determine if an EMAP-L Step 3
response is warranted. Step 3 will require the use of
ancillary data such as stressor information or higher
spatial resolution imagery, such as the U.S. Geological
Survey (USGS) National Aerial Photography Program
1:40,000 color infrared metric photography. The
outcome of a Step 3 analysis is a detailed landscape
assessment related to a specific geographic area and
selected sources of environmental stress.
1.5 LANDSCAPE APPROACHES TO
INTEGRATION
The National Research Council (NRC 1995), EPA
Science Advisory Board (EPA 1995a), and EMAP-
Landscape Peer Review reports (Sharpe et al. 1993)
all state that landscape analyses hold great promise
for conducting integrated assessments across EMAP.
The following is a discussion of a few of these
potential assessments.
It may be possible to relate conditions in individ-
ual ecological resources to one another over broad
areas through a landscape approach. The procedure
involves relating condition of individual resources
(e.g., productivity of forests, biotic conditions of
streams, breeding bird diversity, to name a few) to
landscape composition and pattern at one to several
scales (see discussion in Section 4). In a sense,
landscape composition and pattern becomes the
common denominator by which the condition of one
resource type (e.g., streams) can be compared to
another (e.g., agricultural lands). Discovery of
-------
strong correlation and confidence between landscape
indicators and individual ecological resources is critical
to this process. If correlations between landscape
indicators and individual ecological resources are
relatively weak and insignificant then landscape data
cannot be used to understand the relationships between
conditions in individual ecological resources. Although
numerous studies suggest strong relationships between
landscape composition and pattern and conditions in
individual ecological resources (see discussion in
Section 4), these relationships will likely vary by region;
therefore, EMAP-L proposes to work with other EMAP
projects in the Mid-Atlantic study area to determine
these relationships and develop a strategy to test a
landscape integration concept.
Where strong relationships exist between landscape
composition and pattern and conditions of ecological
resources, including streams, forests, agricultural
lands, and estuaries, then it should be possible to use
landscape assessments as a "coarse filter" of ecological
condition over an entire region. This coarse filter is
made even more effective when landscape data cover
an entire region; this gives complete spatial coverage
and allows for condition estimates in areas where point
data are not available. Such an approach could be
used to prioritize areas for more detailed evaluation.
Landscape pattern metrics, combined with knowl-
edge of how landscape pattern influences ecological
resources, can be used to evaluate risk or vulnerability
of ecological resources in a given region. For exam-
ple, an area with decreasing forest patch size and
connectivity (increasing fragmentation) is more likely
to lose certain ulterior forest birds than areas with
increasing forest patch size and connectivity.
Finally, landscape assessments give environmental
managers a bigger, more holistic view of an area's
ecology. It allows for conditions derived from analy-
sis of individual samples of ecological resources to be
put into a broader context. For example, how uncom-
mon is a particular land cover type within a region? Is
forest fragmentation occurring over an entire region,
or is it limited to specific areas? What cumulative
impacts are discovered at the landscape-level that are
not evident at a local or site level? For example, by
converting a specific forest patch hi an area into a
shopping center, are we eliminating a critical "stepping
stone" for raptor migration across a human domi-
nated landscape?
1.6 CHALLENGES TO EMAP-L
Although landscape ecology has contributed
greatly to our understanding of ecology and ecosys-
tem management (Jensen and Everett 1994), the field
is still in its early stages of development. This is
especially true of our understanding of how land-
scape pattern relates to conditions in ecological
resources. Therefore, one of EMAP-L's greatest
challenges will be to relate empirically-derived
landscape patterns to conditions of ecological re-
sources, including individual ecological resources
embedded within landscapes (e.g., forests, streams,
wetlands), as well as ecological resources that
interact with entire kmdscapes (e.g., breeding birds).
One of the primary objectives of this project is to
determine the ecological relevance of landscape
metrics so that they can be applied as indicators of
ecological condition over regional scales. A number
of research activities are presented in this plan that
deal with this broad issue.
The utility of landscape metrics as indicators of
ecological resource condition is also influenced by
attributes of data and formula used to calculate
metrics. Several research projects have been pro-
posed to deal with these types of issues (see Section 4).
EMAP-L is also challenged with using existing
remote sensing data to analyze and assess changes in
landscape pattern. A number of potential sources of
data exist, each witti specific benefits and limita-
tions. EMAP-L will evaluate the range of existing
data and determine which provides the best estimates
of landscape pattern.
Another important challenge is to identify statisti-
cally valid procedures for testing hypotheses of
landscape change over time. Parametric statistical
procedures require kidependent and similar experi-
mental units, but these are unlikely to be obtained in
regional-scale investigations (Hurlbert 1984,
Hargrove and Pickering 1992). Neutral models
(Gardner et al. 1987, Gardner and O'Neill 1991,
O'Neill et al. 1992) may be used to specify null
-------
hypotheses as benchmarks to test observed changes,
but there is no a priori neutral model which will be
appropriate for all tests that could be made.
Finally, EMAP-L is challenged with assessing and
interpreting landscape pattern data with regard to
specific societal values. EMAP-L proposes to
evaluate and develop assessment and interpretation
protocols for landscape pattern date within the Mid-
Atlantic area and provide interpretive reports relating
landscape indicator information to the social con-
cerns identified for the region.
-------
2.0 STUDY OBJECTIVES AND GENERAL APPROACH
2.1 PROJECT OBJECTIVES
EMAP-L proposes a set of research initiatives to
address key issues relative to the development and
implementation of landscape indicators. Many of
these issues were highlighted in the Landscape
Monitoring and Assessment Research Plan (EPA
1994) which provided a conceptual basis and approach
for the program. A primary objective of this project is
to evaluate the sensitivity of landscape pattern metrics
to statistical properties and ecological condition.
Another key objective is to evaluate the ability of
different remote sensing imagery (e.g, AVHRR,
Landsat-MSS, and Landsat-TM) to derive landscape
pattern metrics and indicators.
Finally, EMAP-L will produce an assessment of the
status and change in landscape indicators, including
indicators for the entire Mid-Atlantic region.
Indicators are aggregations of metrics which
provide information to address assessment questions,
which in turn are based on societal values. In the
process of developing indicators, EMAP-L is
exploring the use of numerous landscape metrics. The
objectives of indicator development are to: 1) optimize
the number of indicators needed to effectively address
assessment questions, and 2) optimize indicators to the
fewest number of statistically independent metrics
required to fully describe indicator conditions. Table 1
lists indicators presently under development and
metrics which are candidates for inclusion in each
indicator. Additional indicators and metrics may be
developed.
2.2 GENERAL AiPPROACH
The Mid-Atlantic Region Landscape Indicators
Project provides EMAP-L with its first opportunity
to test many technical issues. The program will
implement its research simultaneously along two
project lines: (1) a series of studies to address
research and development issues related to landscape
indicators, and (2) determination of landscape status
and trends assessment within the project area. In
FY95, EMAP-L proposes to evaluate the research
and development issues within the Chesapeake Bay
Watershed and later expand the research and analysis
into the entire Mid-Atlantic region (Figure 2). The
Chesapeake Bay Watershed covers approximately
half of the Mid-Athuitic region, and EMAP-L will
utilize a number of existing remote sensing data
bases to begin analysis of the monitoring research
issues (Dobson et al« 1995).
I
Multi-date Landsat-MSS (early 1970s, 1980s, and
1990s), Landsat-TM, and AVHRR data coverages
will be acquired for the project area. In addition to
the spectrally-clustered and classified spatial data,
other ancillary datasets, e.g. stressor information or
other monitoring date (Table 2), will be utilized to
associate landscape condition with environmental
disturbance (Jensen 1993, Dobson etal. 1995). Suc-
cess in resolving key technical issues will determine
the ability of EMAP-L to provide meaningful
regional and watershed assessments of condition or
trend. When landscape indicators pass a series of
sensitivity analyses (see Section 4.1), they will be
incorporated into assessments (Hunsaker and
Carpenter 1990, Hunsaker et al. 1990, Barber 1994).
7
-------
Figure 2. Mid-Atlantic Landscape Indicators Project Area.
-------
Table 1. Societal Values, Example Indicators, and Candidate Metrics
Societal Value Indicator
Candidate Metrics
Biodiversity
Watershed Integrity
Wildlife Habitat Suitability
Stream Biological
Condition
Forest Plant Species
Richness
Landscape Sustainability
Water Quality
Vulnerability to Flooding
Landscape Resilience Landscape Sustainabiiity
patch statistics (number, total area, average sizo, largest size, distance between, ratio perimeter
to area, shape, fractal dimension, square pixel model), fragmentation, contagion, zone
fragmentation index, patch per unit area index, dominance, adjacency of land cover types,
Shannon diversity, biophysical attribute patterns!.
diversity, square pixel model, dominance, fragmentation, zone fragmentation index, patch per unit
area index, adjacency of land cover types, slopo, elevation, diffusion rates, percolation threshold,
erosion index, texture, biophysical attribute patterns, geochemical attributes
diversity, dominance, fragmentation, zone fragmentation index, patch per unit area index, slope,
erosion index, texture, patch statistics, square pixel model, biophysical attribute patterns
patch statistics, contagion, zone fragmentation index, patch per unit area index, fragmentation,
texture, dominance, fractal dimension, square pixel model, biophysical attribute patterns
patch statistics, erosion index, hydrologic modification, adjacency of land cover types,
dominance, contagion, zone fragmentation index, patch per unit area index, fractal dimension,
square pixel model, elevation, slope, biophysical attribute patterns, geochemical attributes
patch statistics, adjacency of land cover types, erosion index, dominance, contagion, zone
fragmentation index, patch per unit area index, fractal dimension, square pixel model, hydrologic
modifications, elevation, slope, texture, biophysical attribute patterns
patch statistics, contagion, zone fragmentation index, patch per unit area index, fragmentation,
texture, dominance, fractal dimension, square pixel model, biophysical attribute patterns
Table 2. Number of EMAP Sampling Points by State in the Mid-Atlantic Region, 1990-1994
EMAP Project
Agricultural Lands 1992
Estuaries
1990
1991
1992
1993
1994
Forests
pre-1994
1994
Lakes
pre-1994
1994
Streams
1993
1994
NY
6
5
2
5
PA
95
91
NJ
10
8
8
8
8
5
2
WV
21
55
43
MD
23*
29*
29
34*
16
7
7
DE
i-
6
8
' 6
6
[-
2
'
1
VA
30
28
31
28
1
104
55
55
NC
19
40
17
Includes one station identified as DC.
-------
3.0 STUDY SITE DESCRIPTION
3.1 SITE SELECTION/LOCATION
The following criteria resulted in the selection of
the Chesapeake Bay Watershed/ Mid-Atlantic Region
study location: (1) spatial extent of region sufficient to
address multi-state, multi-ecoregion, and multi-scale
watershed issues; (2) availability of remote sensing
and GIS data bases; (3) availability of variable land-
scape characteristics and land use to provide ecologi-
cal gradients to test indicator sensitivity; (4) regional
management focus and landscape emphasis with a
number of participating state and federal organizations
(see Section 3.2); (5) significant potential for collabo-
ration with other EMAP projects and other federal and
state monitoring programs; and (6) site of one of
EMAP's proposed geographic initiatives (Section 3.2).
The boundaries of the Mid-Atlantic region are
provided in Figure 2. The region extends from south-
ern New York state into northeastern North Carolina
and includes 17 major drainage basins, i.e. USGS
Water Resource Subregions identified by 8-digit
Hydrologic Unit Codes (HUC), and 11 ecoregions
(Omernik 1995). The proposed study region includes
all states within EPA Region m (PA, WV, MD, DE,
VA) plus the Susquehanna River and Allegheny River
Basins that extend into New York (EPA Region n),
Delaware River Basin that extends into New Jersey
(EPA Region IT), and the Chowan-Roanoke and
Neuse-Pamlico River Basins that extend into North
Carolina (EPA Region IV). Two prominent geo-
graphic features within the Mid-Atlantic region are the
Appalachian Mountains, dominated by deciduous
forests (oak-hickory and maple-beech), and the
Chesapeake Bay.
3.2 STRESSOR CONTEXT/RELATIONSHIP
TO OTHER GEOGRAPHIC INITIATIVES
The region has many large urban population
centers and extensive agricultural use. As a result,
the area is exposed to many stresses which poten-
tially could threaten the environmental and aesthetic
values of the region. As an example, this region is
estimated to receive the highest rates of atmospheric
acid deposition in the United States but has little
buffering capacity to counter subsequent impacts of
stream acidification. Other environmental threats
have been characterized in many forms: urban
development (land cover type conversion and frag-
mentation), industrial and municipal effluent, agri-
culture (habitat conversion and nonpoint runoff/
infiltration), mining and logging (resource extrac-
tion, erosion, siltation, off-site release of mining
leachates), and infrastructure development, e.g.
roads, pipelines, utility corridors. The consequences
of these stressor sources include interruption of the
flow and cycling of natural processes related to
energy, water, nutrients, and biota. The ultimate
outcome is the impairment of desired environmental
"goods and services" such as diminished biological
diversity, declines in recreational and commercial
fisheries, reduction of forest and agricultural prod-
ucts, and loss of recreational opportunities.
Because of the geographic scope of the project and
the complexity of stressor effects, EMAP-L will
collaborate with numerous geographic projects that
are currently in progress. These include several
interstate water management programs such as the
National Estuary Programs and River Basin Com-
missions. For example, the Chesapeake Bay Pro-
gram was initiated in 1983 as a joint project that
includes the states of Maryland, Virginia, and
Pennsylvania, EPA, the District of Columbia, and
the Chesapeake Bay Commission. Estuary Programs
10
-------
and River Basin Commissions are generally inter-
agency organizations responsible for developing and
implementing water resource management plans. They
routinely monitor water resource conditions and
provide important sources of information relative to
watershed integrity. EMAP-L intends to forge strong
partnerships with federal and state land managers and
environmental regulators with existing monitoring
programs or proposed regional initiatives. We expect
to take advantage of existing data and incorporate the
information into assessments at regional, watershed,
and landscape scales.
A brief summary of five major geographic moni-
toring initiatives is described below. These are exam-
ples of existing programs or projects where we expect
strong interaction and collaboration.
1. National Water Quality Assessment
(NAWQA) Program
In 1991, following a series of pilot surface and
groundwater projects, the USGS began a 4-year
transition into national implementation for NAWQA.
The goals of NAWQA are to describe the status and
trends in the quality of a large, representative selection
of our Nation's surface and groundwater resources and
to provide a scientific understanding of the primary
natural and human factors that affect the quality of
these resources. The program consists of two major
components, i.e. study-unit investigations and national
assessment activities. The principal building blocks of
NAWQA are the study-unit investigations of hydro-
logic systems that include parts of most major river
basins and aquifer systems. The program is accom-
plished by conducting intensive assessment activities
on 60 study areas distributed throughout the Nation
and that incorporate about 60 to 70 percent of the
Nation's water use served by public water supply. The
study-units are first intensively sampled to establish
baseline condition and then later sampled on a rota-
tional basis to determine trends in condition. It is
anticipated that the first cycle of intensive investiga-
tions covering all 60 study-units will be completed by
FY 2002. Eight NAWQA study-units are located
within the Mid-Atlantic region. They include: Dela-
ware River Basin, Lower Susquehanna River Basin,
Delmarva Peninsula, Potomac River Basin, Allegheny
and Monongahela River Basins, Kanawha River Basin,
Upper Tennessee River Basin, and the Albemarle-
Pamlico Drainage (Leahy et al. 1990).
2. Mid-Atlantic Highlands Assessment
(MAHA)
MAHA is a project initiated by the Environmental
Services Division of EPA Region in and is based
on a long history of Region ffl involvement in
comprehensive environmental monitoring projects,
e.g. Chesapeake Bay Program. MAHA is designed
to improve the effectiveness of local, state, and
federal environmental protection efforts by conduct-
ing a comprehensive environmental assessment of
more than half the Land in Region m, approximately
64,000 square miles. The assessment will employ
information developed with the assistance of EMAP
and is intended to aid strategic environmental plan-
ning and decision-making in six Omernik ecoregions
(Western Allegheny Plateau, Northern Appalachian
Plateau and Uplands, North Central Appalachians,
Central Appalachian Plateau, Central Appalachian
Ridges and Valleys, and the Blue Ridge Mountains;
Omernik 1987, JOmernik 1995). Specifically, the
objectives of MAHA include the following: (1)
assess the current ecological condition of the Mid-
Atlantic Highlands, its component ecoregions, and
states; (2) locate sensitive areas in need of special
protective action, either for remediation or preserva-
tion; and (3) prioritize the need for further research
into the causes and consequences of pollution in the
Mid-Atlantic Highlands. The highlands project area
was selected for special study by Region HI because
of its strategic ecological importance and current
exposure to multiple-stressors that threaten areas of
high environmental value, e.g. habitats for many
unique and critical species.
MAHA is applying two basic features from the
EMAP monitoring framework, i.e. measurement of
a standard suite of biological attributes to assess
ecological quality (indicators) and collection of
environmental information using a probability-based
sample design to ensure that indicator results can be
characterized with known confidence. Concurrent
efforts within MAHA will monitor 246 sites for a
several-year period. The sampling locations include
65 EMAP-Surface Water sites, 31 Regional Refer-
ence Sites, 46 Regional Environmental Monitoring
and Assessment Program (REMAP) sites, and
11
-------
104 Temporally Integrated Monitoring of Ecosystems
CTIME) acid deposition sites. Primarily aquatic ecologi-
cal attributes will be measured annually during a mid-
April through June index period.
3. Mid-Atlantic Integrated Assessment
(MAIA)
MAIA is one of EMAP's proposed geographic
initiatives. It is designed as a collaborative initiative
between EPA Office of Research and Development
(ORD) and EPA Region HI. The MAIA project area
is identical to the site proposed for the EMAP Mid-
Atlantic Region Landscape Indicators Project (see
Figure 2) and completely subsumes the MAHA project
area.
The purpose of the MAIA project is to develop
information and methods that will be useful to both
regulatory and resource management agencies at all
levels of decision-making. MAIA will employ the
EMAP sample design, indicator, QA, and data collec-
tion activities and is particularly focused at developing
and integrating assessment methods utilizing EMAP
data and other ancillary stressor data to create a large-
scale, multi-resource ecological assessment MAIA
defines integrated ecological assessment as a process
by which information is brought together within a
management context to discern patterns, relationships,
and associations that provide a scientific basis for
making decisions related to ecological resource
problems and to the protection of critical ecological
resources. Figure 3 shows the distribution of EMAP
sampling points within the MAIA project area for
surface waters, forests, and estuaries by USGS Water
Resource Subregions. Table 2 shows the number of
EMAP sampling points by state and date of collection
that will be used for the MAIA integrated assessment;
ancillary stressor data sets will be obtained from other
sources, e.g. weather data.
4. Southern Appalachian Man and the Bio-
sphere (SAMAB)
SAMAB is a partnership of ten federal agencies
(EPA, National Biological Service [NBS], National
Park Service, Tennessee Valley Authority [TVA], U.S.
Army Corps of Engineers, U.S. Fish and Wildlife
Service, U.S. Forest Service, USGS, Economic Devel-
opment Administration, and Department of Energy)
and three states (Georgia, North Carolina, and
Tennessee). The SAMAB Program includes the
west-central portion of the Mid-Atlantic region." It
is designed to promote and conduct ecosystem
management in pursuit of sustainable development
in the Southern Appalachian region. SAMAB
partners, along with several private organizations,
began a regional assessment of ecological status and
trends and their relationships to environmental
stressors in 1994. EMAP-L is participating in
SAMAB and will benefit from the protocols devel-
oped within the project. Objectives of the SAMAB
assessment are similar to the MAHA and MAIA
projects and include: (1) develop an integrated,
spatially explicit data base reflecting current ecologi-
cal quality or condition, thereby setting up issue-
specific assessments; (2) locate sensitive areas or
communities at risk, suggesting priorities or recom-
mendations for remediation or preservation; (3)
examine associations between resource impairment
to environmental stressors such as land development,
air pollution, non-point source pollution, and exotic
species, and; (4) develop and analyze policy options
for implementation within SAMAB.
A landscape status report, or atlas, based on
remotely-sensed imagery (primarily Landsat-TM)
will be produced in 1995, and associated maps and
summary statistics will be available as part of the
integrated data base.
5. CoastWatch Change Analysis Program
(C-CAP)
C-CAP is a program initiated by the National
Oceanic and Atmospheric Administration (NOAA)
to develop a comprehensive, nationally standardized
information system for monitoring land cover and
habitat change in the coastal regions of the United
States. Its purpose is to improve understanding of
coastal uplands, wetlands, and sea grass beds and
their linkages with the distribution, abundance, and
health of living marine resources. C-CAP utilizes
both satellite and aircraft-based sensors to map
emergent wetlands and surrounding uplands as well
as submerged aquatic vegetation (Dobson et al. 1995,
Klemas et al. 1993). The goal of the program is to
monitor coastal areas every 1 to 5 years depending on
the rate and magnitude of change in each region. The
prototype project for C-CAP is the Chesapeake Bay
12
-------
U.S.G.S. Water
Resources Subregions
BB Delaware
Susquehanna
^3 Upper Chesapeafce
HI Potomac
SH Lower Chesapeake
Chowan Roanoke
Neuse -Pamlico
Pee Dee
BB Eastern L. Erie L. Erie
BIB Southwestern L. Ontario
Allegheny
Monongahela
Ell Upper Ohio
IS! Kanawha
G3 Big Sandy -Guyandotte
BB Middle Ohio
Tennessee
Rirests
Surface Waters
Estuaries
November 1994
medlO&teGOll
Figure 3. EMAP Sampling within Major Drainage Basins in the Mid-Atlantic Region.
13
-------
Land Cover Classification Data Set which documents
changes in land cover over the 5-year interval from
1984 to 1989. This data set constitutes one of the
largest change detection efforts ever attempted, cover-
ing an area of approximately 30,000 square miles with
a source data resolution of 30 meters by 30 meters, i.e.
Landsat-TM.
3.3 EXISTING DATA SOURCES
As previously mentioned, EMAP-L will be drawing
data from a number of sources to develop analyses of
landscape indicator sensitivity relative to ecological
significance. Data from four EMAP projects, i.e.
Forests, Agricultural Lands, Estuaries, and Surface
Waters (Streams and Lakes), in addition to REMAP
data have been identified for this purpose (Table 2).
Other national monitoring networks will be accessed
for their information. Some examples of existing
monitoring and data base networks are included in
TableS.
The spatial data used for calculation of landscape
metrics will be derived primarily from two sources, i.e.
North American Landscape Characterization Project
(NALC) and the MRLC. A brief summary of each
data source is described below.
1. North American Landscape Character-
ization (NALC) Project
The NALC project has been developed in support of
the U.S. Global Change Research Program (GCRP)
and is designed to take advantage of historical and
current Landsat satellite remote sensor measurements.
NALC is focused at characterizing land cover types
and evaluating their change over time using satellite
sensors. The goal of the project is to produce stan-
dardized data sets for the majority of the North Ameri-
can continent. The data sets are three-date (July 1972
through September 1992), georeferenced Landsat-
MSS land cover characterizations which allow retro-
spective evaluations of change detection. The project
is being conducted in collaboration with the National
Aeronautics and Space Administration, USGS Earth
Resources Observation System (EROS) Data Center,
and the Canada Centre for Remote Sensing (CCRS).
These data have been archived in digital form and
represent the only existing remote sensor system that
has a digital archive with a long-term record of
acquisitions over a major portion of the Earth. The
Chesapeake Bay Watershed was selected by NALC
as the first pilot study in which to test standard land
cover characterization and change detection proce-
dures. The Chesapeake Bay pilot was conducted
over the 64,000-square mile watershed using six
land cover/land use classes. NALC is also interested
in developing a prospective evaluation methodology
based on the use of Landsat-TM data to facilitate
more detailed spectral and spatial analysis of ecosys-
tems and detection of changes in land cover.
2. Multi-Resolution Land Characteristics
Consortium (MRLC)
MRLC is being developed to provide the capabil-
ity for broad-based research on current and future
conditions of physical and biological resources of
the United States. MRLC began in April 1993 and
is sponsored by five federal environmental monitor-
ing programs having similar remote sensing and
research needs, i.e. NAWQA/USGS, EMAP/EPA,
NALC/EPA, C-CAP/NOAA, and GAP Analysis
Program/NBS, in partnership with the EROS Data
Center/USGS. The goal of the MRLC is to generate
a land cover dataset for the United States based on
Landsat-TM data. Collectively, advantages under
the consortium arrangement include substantial
savings of time, effort, and money related to scene
acquisition, processing, QA, and application and
management of the land cover data.
A six-class, land cover map has been produced
over the entire Chesapeake Bay Watershed; these
data are available through EMAP-LC. A ten-to-
twelve class land cover data base is being developed
by the MRLC; these classifications are being done
from 1992 to 1993 Landsat-TM digital data.
A three-date (early 1970s, mid-1980s, early
1990s) Landsat-MSS data base for the entire MAIA
area is under development; these data should be
available by July 1995. AVHRR data are available
from EMAP-LC via the USGS EROS Data Center.
14
-------
Table 3. Example National Monitoring and Information Systems Containing Ancillary Data
Available to EMAP-Landscaoes
U.S. Geological Survey
National Stream Quality Accounting Network (NASQAN)
National Water Quality Assessment Program (NAWQA)
National Digital Cartographic Data Base
National Water Information System
National Water-Use Information Program
Federal National Trends Network
Toxic Substances Hydrology Program
Regional Aquifer-Systems Analysis Program (RASA) '
U.S. Environmental Protection Agency
Storage and Retrieval System (STORET)
Industrial Facility Discharge File
River-Reach File
Environmental Monitoring and Assessment Program (EMAP)
North American Landscape Characterization Project (NALC)
I
U.S. Fish and Wildlife Service
Environmental Contaminants Data Management System (ECDMS)
National Wetland Inventory (NWI)
i
U.S. Forest Service
Forest Inventory and Analysis Program (FIA)
U.S. National Biological Service i
North American Breeding Bird Survey (BBS) !
National Contaminant Biomonitoring Program (NCBP)
GAP Analysis Program
U.S. Soil Conservation Service
National Resources Inventory (NRI)
U.S. National Agricultural Statistics Service
U.S. Census of Agriculture
U.S. Census Bureau
U.S. Census of Population
U.S. State Department
Land Use Change and Analysis System fl-UCAS} Man and thp Rirxsnhoro Pn
ogram
15
-------
4.0 PROJECT COMPONENTS
The EMAP-L program consists of multiple projects
related to indicator sensitivity, comparability among
remote sensing data, and landscape pattern assessment
in the Mid-Atlantic region.
EMAP-L will evaluate several aspects of landscape
metric sensitivity. For the purpose of discussion,
EMAP-L has divided landscape sensitivity into two
general categories: those issues related to statistical
properties and those related to ecological condition.
4.1 SENSITIVITY RELATED TO
STATISTICAL PROPERTIES
EMAP-L will evaluate a number of metrics with
regard to their utility as indicators of change relevant
to societal values. Particular metrics selected for use
in the Mid-Atlantic region are listed in Table 1.
Additional candidate metrics are discussed in EPA
(1994) and equations can be found in Riitters et al.
(1995).
4.1.1 Number of Samples, Calculation of
Window Size, and Number of Land
Cover Classes
Landscape metrics with a reasonable ecological
basis may still be unusable because they are overly
sensitive to measurement errors in land cover or are
insensitive to landscape change. For example, diver-
sity indices are overly sensitive to the number of
attributes, i.e. too many attributes result in values that
are insensitive to actual change. Fractal dimension
requires a relatively large number of patches, because
dimension is estimated from a regression of patch
perimeter on patch area. Furthermore, each patch must
contain at least four pixels; the shape of patches
smaller than four pixels is constrained, thus including
them would yield biased estimates. These and other
estimation issues are discussed by Milne (1988,1991).
EMAP-L will combine some additional studies
with existing studies to determine the extent of
landscape indicator sensitivity to sample size and to
the number of attributes. The impact of sample size
(number of pixels or pixel aggregations) and the
number of attributes (e.g., land cover classes) on
landscape pattern metrics will be determined by
varying the number of land cover classes and calcu-
lation window size. EMAP-L will change the
number of land cover classes using the Chesapeake
Bay Watershed Landsat-TM data base to simulate
the influence of the number of land cover classes on
landscape metric sensitivity to actual landscape
change. EMAP-L will vary the size of calculation
windows for each metric to evaluate the influence of
scale on indicator sensitivity.
4.1.2 Statistical Independence of Land-
scape Metrics
Some 50 existing landscape metrics have been
proposed for use in development of EMAP-L indica-
tors. It is the intent of EMAP-L to optimize metrics
and indicators, i.e., to employ only those metrics
which are sensitive to landscape change in a predict-
able and defined manner. As a first step in the
optimization process, the statistical independence of
the metrics is being investigated. Groups of statisti-
cally related metrics can be represented by one or
two critical metrics which are most responsive to
change or which reflect change with a defined
response. EMAP-L will evaluate correlations among
landscape indicators and determine the number and
nature of orthogonal axes that explain variation in
landscape pattern.
Window-based analyses (e.g., Potnick et al. 1993)
have been used to study effects of changing scale on
indicator values. EMAP-L is also interested in
evaluating the degree to which relationships are
scale-dependent. To accomplish this, EMAP-L will
16
-------
use a form of spatial filtering. Spatial filtering is an
approach that has traditionally been used in remote
sensing. It involves passing a window that is typically
an odd multiple of the pixel size itself through an
image and changing the pixel classification according
to one of several possible rules. When applied to a
land cover map, the result is often an aggregation of
pixels into identical classes that has the effect of
changing the scale of the image. The change in scale
is an important property of the landscape (and land-
scape change) that EMAP-L will investigate. EMAP-
L will vary the calculation window size to determine
the influence of the number of pixels on indicator
correlations.
This evaluation will be conducted over the Chesa-
peake Bay Watershed using the existing Landsat-TM
data base, and subsequently, over the entire Mid-
Atlantic region as remote sensing data become
available.
4.1.3 Indicator Sensitivity to Land Cover Mis-
classification
Landscape pattern metrics are generated from
remote sensing data classified into land cover maps or
digital data bases. These data are converted to ASCII
files and interfaced with landscape metric statistical
software to produce landscape statistics. It is impor-
tant to know the relative accuracy of assigning land
cover types and it is necessary to test the sensitivity of
the metrics to classification error. If landscape pattern
metrics are more sensitive to random errors in land
cover classification than to actual land cover change,
their use in long-term monitoring would be compro-
mised.
EMAP-L proposes two activities: (1) an extensive
review of existing studies and literature, and (2) a
simulation study. The objective of the literature
review is to evaluate different sources of error that
contribute to land cover misclassification (e.g., from
the satellite to the generation of landscape pattern
statistics). This error progration study will build upon
work conducted by Lunetta et al. (1991) and Jensen
(1993).
A Monte Carlo simulation model, written in
ARC/DSfFO GRID language, will be used to test the
sensitivity of landscape metrics to misclassification.
The simulation will be based on: (1) the
misclassification rates that are calculated from an
error matrix, and (2) spatial autocorrelation in land
cover classification error (Congalton 1988). The
error matrix (Story and Congalton 1986) is the
standard medium for reporting land cover classifica-
tion accuracy (Congalton and Green 1993). An error
matrix is constructed as a square contingency table
where the columns represent reference data and the
rows represent classified data. The main diagonal of
the error matrix contains the number of correctly
classified pixels aid the off-diagonal elements are
incorrect classifications. Either aerial photography,
site visits, or both are typically used to compile
reference data. Congalton (1988) showed that much
of the error in land cover classifications is at the
edge between two land cover classes. That is, land
cover misclassification is spatially autocorrelated.
Error is likely to be higher at the edge separating two
classes than in the interior of a patch. Spatial
autocorrelation in land cover classification error is at
least partly due to mixed pixel (edge effect) phenom-
enon (Lillesand and Kiefer 1987). Mixed pixels
occur when two of more distinct earth surface
features (e.g., soil and water) contribute to the signal
recorded for that pixel.
4.2 SENSITIVITY RELATED TO
ECOLOGICAL CONDITION
If landscape metrics are to be used as indicators,
they must be sensitive to change. Sufficient evi-
dence exists to support the general hypothesis that
changes in landscape pattern alter ecological pro-
cesses which in turn affect the condition of ecologi-
cal resources (Turner 1989). For example, habitat
fragmentation is known to increase extinction
probabilities of certain fauna and flora (Whitcomb et
al. 1981). Similarly, water quality is influenced by
landscape composition and pattern (Hunsaker et al.
1992, Walker etatl. 1993).
j
The primary objective of this activity is, to deter-
mine the sensitivity of landscape metrics to ecologi-
cal status and cluinge. This activity has been sepa-
rated into three components: (1) gradient analysis,
(2) association with ecological resource condition,
and (3) landscape classification.
17
-------
4.2.1 Gradient Analysis
Considerable variation in the degree of human
influence exists within the Mid-Atlantic region. Urban
areas have modified landscapes along the eastern
coast, whereas landscapes in the western mountainous
areas have undergone little modification. Landscapes
with anthropogenic influences (e.g., agriculture and
urban development) will have different landscape
patterns, i.e., landscapes with high anthropogenic
influence are likely to have: (1) smaller average patch
sizes of natural land cover (e.g., forests), (2) higher
degrees of fragmentation, and (3) more simplified
shapes than landscapes with relatively low anthropo-
genic influence. Biophysical attributes of areas within
the Mid-Atlantic region will likely influence the
response of landscapes to anthropogenic and natural
stress. We will take advantage of this existing east-
west gradient to test landscape metric sensitivity, i.e.,
how far along this gradient must one move to detect a
statistically significant change in the metric. Once the
statistical sensitivity of metrics to actual change is
known, it will then be possible to identify those
hypotheses which can be tested in practice. In other
words, statistical significance is different than practical
relevance. Knowledge of statistical properties is a
prerequisite for knowing which questions of practical
significance can be asked of the data.
EMAP-L will establish an urban-to-mountain
gradient by ranking the watersheds depicted in Figure
4 (USGS 8-digit HUCs) by U-index values. This
index (Krummel et al. 1987) is the ratio of the area
devoted to anthropogenic land use to total area. Low
U-values reflect a forest-dominated watershed, and
high values indicate dominance by urban, suburban, or
agricultural land uses. Use of the U-index between 8-
digit HUCs represents a space-for-time approach to
determining the feasibility of metrics for detecting
changes through time. Space-for-time is an accepted
methodological approach in ecological studies (Pickett
1989).
In addition to determining landscape pattern varia-
tion along this gradient, EMAP-L will evaluate the
influence of biophysical attributes, including topogra-
phy, landform, geology, and soils on observed re-
sponses of watersheds.
4.2.2 Association with Ecological
Resource Condition
First, this activity will determine the degree to
which landscape metrics can be used to supplement
assessments of status and trends hi individual
ecological resources at regional scales. Second,
individual EMAP projects will benefit from an
understanding of relationships between landscape
pattern and conditions of their ecological resources
of focus. EMAP-L proposes to work collaboratively
with other EMAP projects in conducting this re-
search. Finally, these analyses will emphasize
assessments for the Mid-Atlantic region. EMAP-L
will emphasize ecological risk assessment and
ecosystem management. For example, findings on
relationships between landscape condition and
stream water quality can be used to formulate risk
models relating landscape change to the risk of
impact on stream biota.
It is EMAP-L's central hypothesis that changes in
landscape pattern influence and constrain the condi-
tion of all ecological resources within the landscape
(EPA 1994). Hierarchy theory points out that larger-
scaled systems, such as landscapes and watersheds,
constrain the dynamic behavior of subsystems, such
as patches of forests or wetlands. Thus, monitoring
land cover patterns is necessary to assess the present
and changing conditions of ecological resources.
An outcome of this hierarchical relationship is
that landscape change may signal resource change in
advance of its onset. In other words, there may be a
tune-lag and landscape pattern may indicate a poor
or declining condition before such a condition is
apparent from analysis of field-collected data. The
Pacific Northwest region provides an example of
such a relationship. Increasing fragmentation of the
forest in this region suggests reduced forest
connectivity, increased distances between patches,
and increased simplification of forest patch edges
(Kaufmann et al. 1994). Many have argued that
such changes signal a decline in forest condition that
is not yet evident at the plot or stand levels. An
analysis of landscape pattern data could have indi-
cated the increased risk to spotted owl populations
from fragmentation of forest patches before local
population extirpation was observed and before
18
-------
Appalachian Plateau
, U SUSCHACK
£ ' Kidge and Valley
Coastal Plain
Figure 4. USGS 8-digit Mid-Appalachian Watersheds.
19
-------
landscape-level disturbance processes were widespread
enough to cause population decline in the entire
metapopulation of spotted owls. Therefore, monitoring
landscape condition, in close coordination with individ-
ual resource monitoring, is critical in assessing risks to
ecological resources within a region.
The sections which follow describe proposed
activities to be undertaken by EMAP-L in collabora-
tion with particular ecological resource monitoring
programs. In this initial phase of activities in the Mid-
Atlantic region, most efforts will be conducted in
conjunction with other EMAP projects participating in
the MAIA program.
(A) STREAMS
Stream biotic (e.g., fish and benthos) and abiotic
(e.g., water quality and physical habitat) conditions
will be compared to landscape pattern metrics at
several scales. This research will be collaborative with
the EMAP-Surface Waters project. EMAP-L and
EMAP-Surface Waters propose to evaluate a number
of associations in the Mid-Atlantic region, including
the following:
(1) Different stream organisms integrate conditions
over different scales of watersheds. Fish can migrate
considerable distances during a life span and should
show poorer correlation to land cover in the immediate
watershed where they are sampled. Relatively immo-
bile benthic invertebrates should show strong correla-
tion with the immediate watershed.
(2) Condition in low-order streams show highest
correlation with small, immediately adjacent land
cover. As stream order increases, the correlation
decreases as more of the adjacent watershed is in-
cluded in the analysis; individual additions of stream
order never explain as much variation as low-order
streams.
(3) Watershed land cover influences stream biota
through the physical and chemical characteristics of
the water. Therefore, there should be stronger correla-
tions between landscape characteristics and water
physicochemical data than with stream biota.
(4) Fragmentation affects diversity of stream biota,
i.e., isolation of streams from other streams reduces
the probability of recolonization. This condition
increases the probability of local extinction (given
natural and anthropogenic disturbance) and therefore
constitutes a risk to stream biotic diversity.
(5) Undisturbed low-order streams that are hi
different ecoregions or biophysical settings should
have higher variation in species composition.
Streams that are closer together with similar land
cover hould be more uniform in species
composition.
EMAP-L and EMAP-Surface Waters propose to
calculate landscape pattern metrics on various sized
watersheds that contain stream sampling sites.
Landscape metrics will be calculated for: (1) the
riparian zone, as defined by proximity of vegetation
to water drainage, (2) the entire drainage area, and
(3) land cover patches weighted by distance from the
stream. In selected areas, a combined area model
will be used. These analyses will clarify which areas
of the watershed have significant influence on
stream biota.
EMAP-L will select a gradient of watersheds
based on biotic and abiotic stream conditions, and
calculate landscape pattern metrics to determine an
initial sensitivity of stream condition to landscape
metrics. EMAP-L will rank watersheds based on
landscape metric values known to correlate with
stream condition for the entire area.
In addition to stream biotic condition, EMAP-L
will evaluate the relationship between landscape
pattern and water quality and discharge. EMAP-L
will use USGS and EPA data on water quality and
discharge and correlate these data to landscape
metrics and other biophysical attributes.
(B) BREEDING BIRDS
EMAP-L will evaluate the association of land-
scape pattern with presence and abundance of breed-
ing birds. The objective of this project is to deter-
mine concordance between status and changes in
landscape attern and bird species richness and
abundance. Additionally, results of this work should
allow EMAP-L to produce an assessment of status
and changes hi bird habitat suitability for the Mid-
Atlantic region.
20
-------
Breeding bird data collected from the North Ameri-
can Breeding Bird Survey (BBS), cover a 28-year
period within the Mid-Atlantic region starting in 1966.
EMAP-L will calculate landscape metrics by using
variable-scale windows around breeding bird sites.
Birds will be lumped into habitat quanta using
Rolling's (1992) allometric analysis of bird home
range sizes. Various bird home range requirements
will be used to select window size for landscape metric
comparisons. In addition to conducting a space-for-
time analysis by comparing early 1990 bird and land
cover data, EMAP-L will analyze correlations between
changes in landscape pattern (via three-date Landsat-
MSS data) with changes in breeding bird presence/
absence, abundance, and species richness. Data from
the BBS will be provided by the NBS. EMAP-Center
(Research Triangle Park, NC) will provide additional
data obtained from a song-bird census method devel-
opment study being conducted within the MAIA area.
(C) FORESTS
EMAP-L proposes to evaluate associations between
landscape pattern and indicators of forest condition,
including plant species diversity, presence/absence of
exotic species, and vertical vegetation. EMAP-L will
collaborate with EMAP-Forests in determining the
sensitivity of landscape metrics to forest condition.
Forest indicator data will be derived from the Forest
Health Monitoring program plots within the Mid-
Atlantic region. Landscape metrics will be calculated
on variable-scale windows surrounding each forest
sample. We will evaluate window sizes ranging from
a few hectares up to several hundred hectares to deter-
mine scaling relationships between landscape pattern
metrics and forest indicators.
(D) ESTUARIES
A proposed EMAP-L collaboration witti the EMAP-
Estuaries program to develop research relating land-
scape metrics to condition of estuarine environments
will commence in 1995. Estuarine conditions may be
a function of their spatial distribution, connectivity,
and shape. Therefore, we will evaluate relationships
between spatial patterns of estuaries and estuarine
condition. EMAP-L will evaluate the correlation
between landscape pattern on watersheds and condi-
tions of estuaries. A similar sample design to that
proposed for surface waters will be used to evaluate
which landscape metrics and which scales are
correlated to estuarine biotic and abiotic condition.
EMAP-L will evaluate the number of paired water-
sheds and estuaries necessary to correlate landscape
pattern and estuarime condition.
4.2.3 Landscape Classification
EMAP-L will evaluate the degree to which values
of landscape pattern and their association to condi-
tions in ecological resources (e.g., forests and
streams) are nested within or constrained by coarser-
scale biophysical attributes and processes (this is
often termed pattern recognition). For example,
EMAP-L will evaluate the degree to which land-
scape pattern varies by ecoregion (e.g., Omernik
1995), and by other biophysical attributes, including
climate, geology, soils, and topography. Additionally,
EMAP-L will evaluate the degree to which finer-scale
patterns of landscapes are constrained or determined
by coarser-scale landscape pattern. The landscape
pattern type analysis of Wickham and Norton (1994)
is one method for evaluating nested relationships of
landscape pattern.
This analysis will be performed in an ARC/INFO
GIS environment by evaluating the degree to which
certain landscape pattern types and values are
spatially-nested within digital coverages of biophysi-
cal attributes and natural region boundaries. Nested
hierarchies would suggest that landscape patterns are
constrained by higher-level ecological processes,
including climate, geology, and topography. Initial
analyses of these relationships will use data for the
Chesapeake Bay Watershed; this study will be
expanded to the entire Mid-Atlantic region as data
become available.
4.3 COMPARABILITY OF AND SYNERGISM
AMONG DIFFERENT REMOTE SENSING
IMAGERY
If change could be determined accurately from
unlabeled data (spectral reflectance and spectrally-
clustered data), then costs of certain landscape
assessments might be significantly reduced, and
EMAP-L could use this approach in Step 2 of its
proposed three-step monitoring and assessment
process (Section 114). EMAP-L will evaluate differ-
ent remote sensing data (e.g., spatial data derived
from AVHRR, Landsat-MSS, and Landsat-TM
21
-------
imagery) in terms of advantages and limitations in
derivation of ecologically meaningful landscape
pattern metrics. This study has been limited to the
three remote sensing data platforms listed above
because of their continuous coverage, spatial scale,
and easy accessibility. Landsat data have been shown
to be useful at the scales selected by EMAP-L in all of
the major ecosystem regions, including coastal areas
(Browder et al. 1989). Additionally, EMAP-L will
evaluate the potential of using spectral reflectance and
spectrally-clustered data to determine landscape
change. Labeling of remote sensing imagery into
digital land cover data bases is an expensive process.
EMAP-L will evaluate concordance in landscape
pattern metrics derived from different imagery
(AVHRR, Landsat-MSS, Landsat-TM) by selecting a
subset of watersheds (USGS 8-digit HUCs) across the
Mid-Atlantic region. On each watershed, landscape
pattern metrics will be calculated from the different
imagery. Some landscape metrics have been shown to
remain nearly constant over a range of 4 to 80 meters
(Wickham and Riitters, in press). However, the values
of particular landscape metrics do not remain constant
at a threshold of 1 ha and larger pixel sizes (Wickham
and Riitters, in press; Turner et al. 1989). Therefore,
we hypothesize that AVHRR-derived landscape
metrics will depart significantly from Landsat-MSS
and Landsat-TM because of its relatively large pixel
sizes (1.1 km on a side versus 30m and 80m for
Landsat-TM and Landsat-MSS, respectively). This
will be especially true for those landscape metrics
requiring several thousand pixels (e.g., diversity
metrics). We hypothesize that Landsat-TM and
Landsat-MSS will generate relatively similar land-
scape pattern results over 8-digit HUC watersheds
because pixels are closer in size and because at least
some spectral bands of the two sensors collect data in
similar portions of the electromagnetic spectrum (e.g.,
Landsat-MSS band 4 matches Landsat-TM band 2,
and Landsat-MSS bands 6 and 7 match Landsat-TM
bands 3 and 4).
Relating landscape pattern derived from Landsat-
MSS and Landsat-TM has some important ramifica-
tions for the EMAP-L program. Landsat-MSS data
will be available over a large portion of the Nation and
cover time periods dating back to the earlier 1970s.
EMAP-L would like to use Landsat-MSS data to
produce its first assessments of landscape status and
change. However, the Landsat-MSS program has
been phased out, leaving only the Landsat-TM
sensor to produce data at a similar scale. Therefore,
the later assessments of landscape change would
involve comparison of landscape patterns derived
from Landsat-MSS and Landsat-TM. Dottavio and
Dottavio (1984) provide a framework for matching
Landsat-MSS and Landsat-TM.
EMAP-L will compare landscape pattern esti-
mates derived from land cover data with those
derived from spectrally-clustered and reflectance
data to determine their utility in evaluating landscape
change. The comparisons will also be made on the
same set of 8-digit HUC watersheds; a space-for-
time concept will be used to evaluate their potential
use in landscape change detection.
Interest in detection of land cover change using
remotely sensed data pre-dates the advent of Land-
sat, the oldest satellite sensor system (Shepard
1964). Since the advent of Landsat and other satel-
lite systems, their use in detection of land cover
change has been a principal research focus. Several
change detection methods have been developed and
evaluated, including differencing, ratioing, and
regressing multi-date imagery, principal components
and comparison of independent land cover classifica-
tions (post-classification comparison). Reviews of
these methods can be found in Jensen (1981, 1983),
Singh (1989) and Mouat et al. (1993). These meth-
ods can be grouped into two broad categories: image
enhancement and multi-date classification (Pilon et
al. 1988).
Detection of land cover change using multi-date
satellite data must control for the confounding
effects of different atmospheric conditions, different
sensors, and accurate registration of the temporal
image sets. Image enhancement methods are de-
signed to control for different atmospheric condi-
tions and different sensor characteristics. The image
enhancement techniques have been found to accu-
rately identify "change" versus "no change" (see
Stauffer and McKinney 1978, Nelson 1983, Fung
and LeDrew 1988, Fung 1990). Once "change"
versus "no change" has been determined for a multi-
date data set, additional image processing techniques
are required to determine the "from" and "to" cate-
gories of change (e.g., "from" forest "to" urban").
22
-------
The logic of post-classification comparison is better
suited for categorical determination of land cover
change (Jensen 1986, Jensen 1993). However, the
method is often hindered by its inability to distinguish
changes in land cover from differences in the environ-
ment (e.g., atmospheric difference). Post-classifica-
tion comparison can lead to improbable "from" and
"to" categories (e.g., "from" urban "to" forest). These
improbable changes can be the result of the multiplica-
tive effect of error in each of the classifications. For
example, land cover maps that each have an 80 percent
overall classification accuracy could result in a
"from'V'to" change map that has an overall accuracy
of 64 percent (0.80 x 0.80 = 0.64). The potentially
poor accuracy of post-classification comparison can
apparently be overcome if the individual classifica-
tions and geometric registrations are sufficiently
accurate (Jensen 1986, Jensen 1993).
The NALC Landsat-MSS triplicate data set for
path/row 15/33 (Washington, DC) is currently being
analyzed by EMAP-L. Independent classifications for
the 1973, 1987, and 1990 imagery were created,
requiring the implementation of the post-classification
comparison change detection method. Several pro-
cessing techniques will be used to compare these
independent classifications. The techniques are a
compilation of those found in Pilon et al. (1988),
Schott et al. (1988), and Jensen (1993).
4.4 FUTURE PRACTICAL APPLICATIONS
The three-step approach proposed by EMAP-L has
a number of important potential practical applications,
particularly in the area of environmental management.
A recent report by the Science Advisory Board (SAB;
EPA 1995b) stresses the need for future planning and
recommends formation of an Environmental Futures
Committee to forecast future environmental problems
and develop resolutions. Specific recommendations
relevant to the work being conducted in EMAP-L
include providing as much attention to avoiding future
environmental problems as is given to controlling
current ones, establishing an early-warning system,
paying particular attention to the sustainability of
terrestrial ecosystems and development and use of
non-traditional environmental stressors, improving
and integrating environment-related futures studies,
focusing attention on broad causes of environmental
change, and establishing a broad-based data system
for anticipating future environmental risks (EPA
1995b). The applications being developed by
EMAP-L, if successful, will provide some of the
tools critical to achieving the objectives of the
SAB's recommendations. The combined use of
existing historical data and new data in establishing
trends can potentially be used hi an early-warning
system and provide critical information on the
sustainability of terrestrial and riparian ecosystems.
The metrics and indicators being developed in
EMAP-L are direct or surrogate measures of
ecosystem-level stressors. Finally, the EMAP-L
assessments integrate a wide variety of data sets, i.e.,
integration of ground-based ecological studies with
remote sensing data collected by other agencies for
non-environmental uses.
While the development phase of the EMAP-L
three-step approach is being conducted at the land-
scape scale, it should be possible to adapt applica-
tions to a variety of scales. The use of multiple
scales has applications in delineating status and
trends of the habitats of small to large animal spe-
cies, investigating the effects of pollution point
sources, and detecting changes in particular land
cover types from low-order streams to large forests
and rangelands. Scale applications was one of the
points raised in a review of EMAP by the National
Research Council (NRC 1995). Another issue raised
by the NRC was EMAP's reliance upon statistical-
based sampling designs; the EMAP-L approach
permits full-scale coverage. The EMAP-L approach
makes extensive use of existing data, another issue
raised by the NRC review. Finally, the EMAP-L
approach provides integration of multiple data layers
in a GIS-based work environment.
23
-------
5.0 ASSESSMENT OF STATUS AND TRENDS
EMAP-L has selected the Mid-Atlantic region as
a research and assessment location based on a number
of criteria. The EMAP-L assessment strategy is to
conduct assessments in areas where extensive spatial
data exist and preferably are of known confidence.
Secondly, EMAP-L has targeted indicator and assess-
ment research in locations where other EMAP projects
are active. Hence, EMAP-L interpretations of status
and trends will add value to other EMAP research and
assessment activities. Further, EMAP-L will collabo-
rate with other research or monitoring programs
currently under development, e.g. NAWQA, SAMAB.
EMAP-L will produce three types of status and
trends products in the Mid-Atlantic study area by late
FY96: (1) a landscape statistical summary, (2) an
interpretive assessment, and (3) a series of journal
articles. Estimates of landscape indicators for 8-digit
HUC watersheds will be determined within the Chesa-
peake Bay Watershed in FY95; landscape status and
trends and their relevance to societal values will be
highlighted in an interpretive report for the entire Mid-
Atlantic region by late 1996.
5.1 CHESAPEAKE BAY, FY95
A statistical summary will be developed for the
Chesapeake Bay Watershed using 1990 Landsat-TM
data. The status of landscapes in the Chesapeake Bay
Watershed will be reported in two formats, i.e. cumu-
lative distribution functions (CDFs) and raster maps.
The CDF format will utilize USGS 8-digit HUCs as
landscape units, that is, the indicators will be calcu-
lated separately for each HUC and aggregated to a
regional status report by aggregating the values for
each HUC. The raster map format will utilize the
same map extent, grain size, and resolution as the
input land cover maps, and indicators will be calcu-
lated by using sliding windows of various sizes. Thus,
the output maps will be "surface maps" of indicator
values. These surface maps are primarily useful for
visualizing broad-scale patterns of indicator values.
It will be possible to stratify these maps (by water-
shed, ecoregion, etc.) for further status and trend
assessments and for drawing correlations with finer-
scale information from other EMAP projects.
We will use the Chesapeake Bay Watershed study
to explore different ways to communicate results to
data users. Viable alternatives include paper reports,
CD-ROM, and Internet (via an existing EMAP
Mosaic server). The best distribution channel will
depend upon the capabilities of the primary data
users. Maps will be made available in formats
suitable for access via ARC/INFO GIS software.
The following are examples of assessment ques-
tions that EMAP-L will address in the Chesapeake
Bay:
1) What is the status of wildlife habitat suitability
across the Chesapeake Bay Watershed based on
landscape composition and pattern at multiple
scales?
2) Which 8-digit watersheds have the highest
amount of suitable habitat for interior forest species?
For forest-edge species?
3) Which patches of forests have the biggest
potential influence on habitat suitability? Which
patches, if lost, would have the greatest negative
influence on forest connectivity?
4) What is the distribution of roads and other
human influences across the Chesapeake Bay and
what is their spatial relationship to wildlife habitat
suitability?
24
-------
5) What is the spatial distribution of landscape
pattern types over the Chesapeake Bay? To what
degree do biophysical attributes and man influence
these patterns?
5.2 MID-ATLANTIC PROJECT AREA, FY96
An interpretive assessment will be developed for the
entire Mid-Atlantic region by late FY96. The FY96
Mid-Atlantic Region project area will be handled as an
expansion of the FY95 Chesapeake Bay Watershed
study. New elements will include trend analysis and
optional reporting units in addition to watersheds.
Twenty-year trends in landscape ecological condi-
tion will be the focus of this report, but a current status
report will also be prepared. The Mid-Atlantic re-
gional land cover maps from the MRLC Landsat-TM
data and the NALC Landsat-MSS data will be used.
The MRLC maps have an advantage of finer scale,
while the NALC maps offer the unique opportunity to
quantify past trends in land cover over the entire
region.
Landscape condition changes will be related to
societal values, i.e. biodiversity, watershed integrity,
and landscape resilience, established for the region and
can be compared to other assessments derived inde-
pendently from other EMAP projects or other multi-
agency assessment programs. The interpretive assess-
ment will also use ancillary data sets to help establish
associations between ecological condition and stressor
types and intensities.
Landscape units, metrics, indicators, calculation
windows, software, reporting formats, and other
details will be similar to those used in the Chesapeake
Bay Watershed study, but modified according to what
has been learned in that exercise. It is expected that
additional biogeographic and political boundaries (that
is, not just watershed boundaries as in the Chesapeake
Bay example) will be employed.
The following are examples of assessment questions
that EMAP-L will address over the entire Mid-Atlantic
region:
1) What are the status and trends in wildlife habitat
suitability across the Mid-Atlantic region based on
landscape composition and pattern at multiple
scales? j
2) Which 8-digit watersheds are gaining or losing
habitat for interior forest, forest-edge, and edge
species? How do these gains and losses vary with
spatial scale?
3) What are the relationships between type, magni-
tude, and distribution of anthropogenic stress and
status and changes in wildlife habitat suitability?
What is the influence of biophysical attributes,
including geology, landform, topography, soils, on
these relationships?
4) What are the associations between landscape
composition and pattern and water quality over the
Mid-Atlantic region? What is the influence of
biophysical attributes on these relationships? Based
on these associations, and landscape composition
and pattern change over the Mid-Atlantic region
from 1972-1992, which watersheds are vulnerable to
declines in water quality?
5) What are the associations between landscape
composition and pattern and biotic condition of
streams in the Mid-Atlantic region? What is the
influence of biophysical attributes on these relation-
ships? Based on these associations, and landscape
composition and pattern change over the Mid-Atlan-
tic region from 1972-1992, which watersheds are
vulnerable to declines in stream biota?
6) What are the associations between landscape
composition and pattern and forest condition in the
Mid-Atlantic region? What is the influence of
biophysical attributes on these relationships? Based
on these associations, and landscape composition
and pattern change over the Mid-Atlantic region
from 1972-1992, which watersheds are vulnerable to
declines in forest condition?
7) What are the associations between landscape
composition and pattern and estuarine condition in
the Mid-Atlantic region? What is the influence of
biophysical attributes; on these relationships? Based
on these associations, and landscape composition
25
-------
and pattern change over the Mid-Atlantic region from
1972-1992, which estuaries are vulnerable to declines
in overall condition?
8) To what degree does breeding bird richness vary
with landscape composition and pattern across the
Mid-Atlantic region at multiple scales? What is the
influence of biophysical attributes on this relationship?
To what degree do breeding bird guilds vary with
landscape composition and pattern?
9) What are the relationships between changes hi
landscape composition and pattern from 1972 to 1992
and changes in breeding bird richness and guilds over
the entire Mid-Atlantic region? Do these relationships
vary among different spatial scales? What is the
influence of biophysical attributes on these relation-
ships? Based on these assessments, which areas are
vulnerable to declines in breeding bird richness?
10) How have flooding events varied by watershed
across the Mid-Atlantic region? What are the charac-
teristics of watersheds that have exhibited little or no
flooding, moderate flooding, severe flooding? What
percentage of this variation is explained by status and
changes in landscape composition and pattern? by
biophysical attributes? by hydrologic modification?
11) To what degree does the condition of ecological
resources in watersheds vary by landscape pattern type
across the Mid-Atlantic region?
12) What is the spatial distribution of landscape
composition and pattern change between 1972 and
1992 over the entire Mid-Atlantic region? To what
degree did the distribution and magnitude of change
vary with anthropogenic stress? with natural stress?
What proportion of the variation in landscape change
was explained by biophysical attributes? Which
watersheds appear to be most resilient to anthropo-
genic stress? natural stress? combinations of an-
thropogenic and natural stress?
13) Is watershed resiliency related to the condition
of ecological resources? to biophysical attributes? to
sustainabiliry of specific ecological resources?
EMAP-L anticipates that part of the results of
landscape assessments will be included in a larger
report for the MALA region. A separate research
plan highlighting the overall MALA assessment is
under development by EMAP-Center.
5.3 DATA BASES OF LANDSCAPE
STATISTICAL SUMMARIES
EMAP-L will provide a data base of landscape
statistics derived for both the statistical summary and
interpretive assessment reports. GIS data layer
coverages will be provided in addition to the tabu-
lated descriptive statistics. Data layers will include
the spatial convolution maps for the selected indica-
tors and may include ancillary datasets used to
determine the spatial correlation relative to stressor
sources. Data bases of landscape statistics will be
produced on CD-ROM and distributed to other
EMAP projects and other agency programs.
26
-------
6.0 INFORMATION MANAGEMENT
There will be five primary information manage-
ment activities within the Mid-Atlantic Region Land-
scape Indicators Project: (1) data acquisition and
documentation; (2) data analysis and creation of data
bases with results of research and assessments; (3)
reports, including journal articles, statistical summa-
ries, and interpretative assessments; (4) technology
transfer of EMAP-L computer programs and assess-
ment protocols; and (5) a QA audit of data and data
flow used to estimate landscape status and trends.
6.1 DATA ACQUISITION AND
DOCUMENTATION
The primary digital imaging processing system
used is ERDAS Imagine Version 8.2. Spectral and
land cover data bases will be acquired via EMAP-LC
and NALC. They include a 6-class Landsat-TM data
base of the Chesapeake Bay Watershed, a 10-class
Landsat-TM data base of the entire Mid-Atlantic
region, and an 8 to 11-class Landsat-MSS data base
covering three dates (the early 1970s, the mid-1980s,
and the early 1990s). EMAP-L anticipates receiving
partial coverages of the 10-class Landsat-TM and 8 to
11-class Landsat-MSS data bases until the entire area
is covered in FY95 (Table 4). The 6-class Chesapeake
Bay Watershed data base has already been received by
EMAP-L. All classifications will be simplified (i.e.,
aggregated) to the Anderson Level I classification
system (Anderson et al. 1976).
EMAP-L will acquire other spatial data bases from
EMAP-LC that will be used to evaluate landscape
indicator sensitivity, including multiple-scale spatial
patterns of landscapes (see Section 4). EMAP-LC has
developed an on-line geographic reference data base
for the MAIA area (the MAIA GRD). This data base
can be accessed via Internet.
Figure 5 summarizes sources and flow of data to
be used by EMAP in the Mid-Atlantic Region
Landscape Indicators Project. Each spatial data set
is accompanied by a metadata file containing the
"genealogy" or lineage of the data. This metadata
file includes all ancillary information relevant to the
digital data, land cover maps produced using the
data, and results of iury data verification activities.
Specifically, this documentation must include the
following, at a minimum: source of digital data, date
obtained, results of initial data checks and scans,
summary of land cover classification system used
and definitions of each land cover category, notation
of all "missing" data categories (e.g., cloud and
shadows), minimum map unit, methods of data
aggregation, methods of data "smoothing", and land
cover error matrices.
Table 4. Approximate! Schedule of Land Cover Data
Base Availability in the Mid-Atlantic Region
Chesapeake Bay Watershed Landsat TM Data Base Aug. 1994
Mid-Atlantic AVHRR Data Base Dec. 1994
Mid-Atlantic Landsat-MSS Data Base (three date) Jul. 1995
Mid-Atlantic Landsat TM Data Base Dec. 1995
6.2 DATA ANALYSIS
EMAP-L will use the LandStat/Spatial Convolu-
tion programs (developed by K. H. Riitters at TVA)
to analyze landscape pattern and commercial statisti-
cal software to conduct multivariate analysis of the
data sets. Spatial coverages will be exported in
GRID ASCII format so that they can be imported
into statistical programs that calculate landscape
metrics. Similarly, results of landscape analyses will
be imported back into ARC/INFO in the GRID
ASCH format.
27
-------
EROS
Data Center
and MRLC
i
r i
EMAP-LC and NALC
Data Nodes - RTP, NC
and Las Vegas, NV
i
ftmit > '" '
m-
llfi'^,,
i
A
/
Other
Agency
Data
l
Other
EMAP Project
Data Nodes
i
A
EMAP-L
n-*. 11- ji'^* '<*'*? -'-yM^wMM
UflTfl BMfiflC" ' \ * , '''" * ^ '**^'^*v^"^i^^^^KS
ferMlVC BV%^^*V, - , , Jf^j vWv'J < 'fjf'fj'yVjf ^'/J^MpffiffieiF
\f MiV" ' ? '^^fer*ffi^8'
Las vegas,
A
DOE/ORNL
Data Node
Oak Ridge, TN
>
i ,
TVAData
Node
Norris, TN
0677msd04J2
Figure 5. Sources and Flow of Data to be Used by EMAP-L in Conducting Research and Assessments in
the Mid-Atlantic Region.
Spatial comparisons and analyses in ARC/INFO
require that spatial data have similar projections (e.g.,
represent the same areas), are in similar format (e.g.,
vector or raster), and are of similar scale or resolution.
EMAP-L will ensure that each of these criteria are met
prior to conducting an analysis.
6.3 REPORTS
EMAP-L will produce a series of journal articles,
statistical summaries, and interpretative reports of its
results for the Mid-Atlantic Region Landscape Indica-
tors Project area. Statistical summaries will be
distributed as hard copy reports, as well as data bases.
Data bases will include metadata, tables of results,
and spatial displays (e.g., spatial convolution
applications). These data bases will be made
available on tapes, via Internet, and through the use
of EMAP-wide information system.
6.4 TECHNOLOGY TRANSFER
EMAP-L has and will continue to develop a series
of statistical tools related to calculation of landscape
pattern metrics, including spatial displays of
statistical results. It is EMAP-L's intent to make
these tools available to EPA Regions, EPA Program
Offices, other agencies, and the scientific community
as specific applications are completed and tested.
28
-------
7.0 QUALITY ASSURANCE
Quality assurance is an integral component of all
EMAP-L activities (Kirkland 1994). Application of
QA principles to remote sensing and GIS is a rela-
tively new endeavor, therefore, much research and
development is still being undertaken. The EMAP-L
QA program is designed to take advantage of ongoing
QA research, much of which is being conducted at
EPA's Characterization Research Division in Las
Vegas (CRD-LV). In addition, the EMAP-L QA
program contains a large research component to
investigate issues directly related to the EMAP-L
design and indicators. The primary areas of QA
research and development within EMAP-L include:
(1) data quality objectives (DQO), (2) audits of data
quality (ADQ), and (3) data quality assessments
(DQA). The QA program for EMAP-L is documented
in the Quality Assurance Project Plan (QAPP;
Chaloud, in prep.) which is updated as needed to
reflect changes and advances in development of the
QA program.
7.1 DATA QUALITY OBJECTIVES (DQO)
For many of the EMAP-L indicators, existing data
are not available on the sources and extent of
variability. The primary thrust of research in this area
is identification and quantification of sources of
variability and investigation of the impacts of
variability on particular metrics or indicators. The
indicator development projects, e.g., indicator sensitiv-
ity, error propagation and ecological significance
investigations (see Section 4), are the major contribu-
tors to this research. Once the major sources of
variability are identified, QA procedures can be
developed to control, minimize, and measure the
contributed uncertainty.
A program-level DQO for EMAP has been estab-
lished as the ability to detect 20 percent change over
10 years with an alpha of 0.2 and beta of 0.3 (power of
0.7) (Kirkland 1994). EMAP-L, through its research
on landscape indicator sensitivity (e.g., space-for-time
and 3-date Landsat-MSS evaluations), will evaluate
each indicator's ability to meet this DQO. Because
the two primary sources of data used by EMAP-L
are NALC and MRLC, these three groups will work
closely together in evaluating DQOs.
7.2 AUDITS OF DATA QUALITY (ADQ)
EMAP-L makes extensive use of data bases
generated by other programs. The primary data
sources for EMAP-L are the NALC and MRLC
programs. In addition, EMAP-L will use expanded
synoptic data sources, i.e. AVHRR, Landsat-TM,
Landsat-MSS, and ground sample probability-based
data (see Tables 2 and 3). In using data generated by
other sources, it is critical to ascertain the quality of
the data and associated metadata, establish the
traceability and integrity of the data, and define the
potential uses and limitations of the data set. The
ADQ is the assessment tool used for this purpose.
Development of processes for conducting ADQs has
been given high priority in the EMAP-L QA re-
search program because EMAP-L plans to utilize
data sets dating back to the 1970s. Initially, the
EMAP-L will develop the ADQ procedures in
conjunction with MRLC and NALC programs.
Once procedures have been fully developed and
tested, they will be used to conduct ADQs of histori-
cal data sets.
7.3 DATA QUALITY ASSESSMENTS (DQA)
Data quality assessments are necessary to evaluate
achieved data quality; against the DQOs established
for the program. The DQAs include calculation of
achieved precision, accuracy, completeness, and
detection limits or other acceptance limits as
appropriate for specific analyses and indicators.
These calculations will be summarized in statistical
tables for inclusion in statistical summary reports
and status reports. In addition, the DQA data are
available for inclusion in journal articles.
29
-------
8.0 PROJECT OUTPUTS AND TIMELINE
A number of products will result from the Mid-
Atlantic Region Landscape Indicators Project, includ-
ing: (1) journal articles on findings of research, (2)
assessments of landscape status and trends, and (3)
digital data bases with spatial coverages of statistical
analyses. Assessment reports will include both statisti-
cal summaries and interpretive assessments; these will
be in the form of EPA reports and journal articles.
Table 5 provides a list of anticipated products,
relative priority ranking, and approximate comple-
tion dates. Completion of Mid-Atlantic Region
Landscape Indicators Project reports will depend on
the completion of land cover data bases over the
entire area. Table 4 lists key land cover data bases
and delivery dates (to EMAP-L).
30
-------
Table 5. List of EMAP-L Products Anticipated From the Mid-Atlantic Region Landscape Indicators Project.
Each is Listed by Major Activity Type (see Sections 4 and 5). Priority: 1 = high; 2 = medium; and 3 = low.
Product
Priority
Anticipated Completion
Date
Sensitivity - Statistical Properties
Cumulative Error Propagation in Estimating Landscape Status and Trends - Journal article
Impact of Land Cover Misclassification on Landscape Pattern Metrics - Journal article
Orthogonality of Landscape Metrics at Multiple-scales - Journal article
Ecological Sensitivity of Landscape Pattern Metrics
Landscape Metric Sensitivity Along an Urban to Rural Environmental Gradient in the Chesapeake
Bay Watershed - Journal article
Relationships between Landscape Pattern and Stream Condition in the Mid-Atlantic Region -
Journal article
Relationships between Landscape Pattern and Forest Condition in the Mid-Atlantic Region -
Journal article
Relationship between Landscape Pattern and Water Quality in the Chesapeake Bay Watershed -
Journal article
Watershed Vulnerability to Severe Hooding - a Landscape-Level Approach - Journal article
Status and Trends in Breeding Bird Habitats - a Landscape-level Assessment - Journal article
A Landscape-level Analysis of Wildlife Habitat - an Example from the Chesapeake Bay Watershed
- Journal article
Influence of Multiple-scale Biophysical Attributes on Landscape Pattern - Journal article
Sensor Synergism
Comparison of Satellite Sensors and Their Ability to Detect Landscape Pattern Status and
Change -Journal article
Detecting Landscape Pattern Change from Reflectance and Spectral Cluster Data - Journal article
Assessments
Landscape Status in the Chesapeake Bay Watershed - Statistical Summary
Landscape Status and Change in the Mid-Atlantic Region - Interpretive Report Relative to
Landscape Values - Journal article
Other
Findings of Landscape Indicator Research in Eastern U.S. Landscapes - Journal article
EMAP-L Quality Assurance Project Plan - EPA Report
2
1
2
1
1
2
1
3
1
1
2
1
3
' 1
1
2
1
September 1995
January 1995
June 1995
March 1995
May 1996
May 1996
October 1995
September 1996
June 1996
June 1995
July 1996
July 1996
July 1996
August 1995
September 1996
October 1996
August 1995
31
-------
9.0 REFERENCES
Anderson, J. R., E. E. Hardy, J. T. Roach, and R. E. Witmer. 1976. A land use and land cover
classification system for use with remote sensor data. U.S. Geological Survey, Professional
Paper 964, Washington, D.C.
Baker, W.L. and Y. Cai. 1992. The r.le programs for multi-scale analysis of landscape structure using the
GRASS geographical information system. Landscape Ecology, 7(4):291-302.
Barber, M.C. (ed.) 1994. Environmental Monitoring and Assessment Program: Indicator Development
' Strategy. EPA/620/R-94/022, Office of Research and Development, Washington, D.C.
Baron, J. and K. A. Galvin. 1990. Future directions of ecosystem science. Bioscience, 40(9):640-642.
Browder, J. A., L. N. May, A. Rosenthal, J. G. Gosselink, and R. H. Bauman. 1989. Modeling future
trends in wetland loss and brown shrimp production in Louisiana using Thematic Mapper
imagery. Remote Sensing of Environment, 28:45-59.
Chaloud,D.J. 1995. Quality Assurance Project Plan for EMAP-Landscapes. U.S. Environmental
Protection Agency, Las Vegas, Nevada (in preparation).
Congalton, R. S. 1988. Using spatial autocorrelation analysis to explore the errors in maps generated
from remotely sensed data. Photogrammetric Engineering and Remote Sensing, 54(5):587-592.
Congalton, R. S. and Green, K. 1993. A practical look at the sources of confusion in error matrix
generation. Photogrammetric Engineering and Remote Sensing, 59(5):641-644.
Dobson, J., R. L. Ferguson, D. W. Field, L. L. Wood, K. D. Haddad, H. Airedale, J. R. Jensen, V. V.
Klemas, R. J. Orth, and J. P. Thomas. 1995. NOAA CoastWatch Change Analysis Project:
Guidance for Regional Implementation. National Oceanic and Atmospheric Administration,
Washington, D.C. (In press). 128pp.
Dottavio, C. L. and F. D. Dottavio. 1984. Potential benefits of new satellite sensor for wetland mapping.
Photogrammetric Engineering and Remote Sensing, 50(5):599-606.
Ehrlich, P. R. and A. H. Ehrlich. 1991. Healing Planet: Strategies for Resolving the Environmental
Crisis. Addison-Wesley Publishing Company, Inc., Reading, Massachusetts. 366pp.
Environmental Protection Agency (EPA). 1993. Environmental Monitoring and Assessment Program:
Master Glossary. EPA/620/R-93/013, Office of Research and Development, Washington, D.C.
Environmental Protection Agency (EPA). 1994. Landscape Monitoring and Assessment Research Plan.
EPA 620/R-94/009, Office of Research and Development, Washington, D.C.
32
-------
Environmental Protection Agency (EPA). 1995a. Review of the Environmental Monitoring and Assessment
Program Landscape Component. U.S. Environmental Protection Agency, Science Advisory Board
(EPA-SAB-EPEC-LTR-95-002). Washington, D.C.
Environmental Protection Agency (EPA). 1995b. Beyond the Horizon: Using Foresight to Protect the
Environmental Future. U.S. Environmental Protection Agency, Science Advisory Board,
Environmental Futures Committee, Washington, D.C. 34 pp.
Forman, R. T. T., and M. Godron. 1986. Landscape Ecology. John Wiley arid Sons, New York.
Franklin, J. F. 1993. Preserving biodiversity: species, ecosystems, or landscapes. Ecol. Appl., 3:202-
205.
Fung, T. and E. LeDrew. 1988. The determination of optimal threshold levels for change detection using
various accuracy indices. Photogrammetric Engineering and Remote Sensing, 54(10):1449-
1454.
I
I.
Fung,T. 1990. An assessment of TM imagery for land-cover change detection. DEEE Transactions on
GeoScience and Remote Sensing, 28(4):681-684. j
i
Gardner, R. H. and R. V. O'Neill. 1991. Pattern, process and predictability: The use of neutral models
for landscape analysis, pp 289-307. In: Turner, M. G. and R. H. Gardner (eds.), Quantitative
Methods in Landscape Ecology. The Analysis and Interpretation of Landscape Heterogeneity.
Ecological Studies Series, Springer-Verlag, New York.
i
Gardner R. H., B. T. Milne, M. G. Turner, and R. V. O'Neill 1987. Neutral models for the analysis of
broad-scale landscape pattern. Landscape Ecology, 1: 19-28.
Hamblin,A. 1992. Environmental Indicators for Sustainable Agriculture. Department of Primary
Industries and Energy, Bureau of Rural Resources. Canberra, Australia. 96 pp.
Hargrove, W. W., and J. Pickering. 1992. Pseudoreplication: a sine qua nonfor regional ecology.
Landscape Ecology, 6: 251-258.
Holden, C. 1988. The ecosystem and human behavior. Science, 242: 663.
Rolling, C. S. 1992. Cross-scale morphology, geometry, and dynamics of ecosystems. Ecological
Monographs, 62(4):447-502.
Hunsaker, C. T., R. L. Graham, G. W. Suter H, R. V. O'Neill, L. W. Barnthouse and R. H. Gardner. 1990.
Assessing ecological risk on a regional scale. Environmental Management, 14:325-332.
Hunsaker, C. T., and D. E. Carpenter, eds. 1990. Environmental Monitoring and Assessment Program:
Ecological Indicators. EPA/600/3-90/060. U.S. Environmental Protection Agency, Office of
Research and Development, Washington, D.C.
33
-------
Hunsaker, C. T., D. A. Levine, S. P. Timmins, B. L. Jackson, and R. V. O'Neill. 1992. Landscape
characterization for assessing regional water quality. Pp. 997-1008, In:. D.H. McKenzle, D.E.
Hyatt, and V.J. McDonald (eds.), Ecological Indicators. Elsevier Appl. Sci., New York.
Hurlbert, S. H. 1984. Pseudoreplication and the design of ecological field experiments. Ecological
Monographs, 54:187-211.
Jensen, J. R. 1981. Urban change detection mapping using Landsat digital data. The American
Cartographer, 8(2):127-147.
Jensen, J. R. (ed.) 1983. Urban/suburban Land Use Analysis. Manual of Remote Sensing, Volume H,
2nd. Ed. pp. 1571-1666. American Society of Remote Sensing, Bethesda, Maryland.
Jensen, J. R. 1986. Introductory Digital Image Processing: A Remote Sensing Perspective. Prentice-
Hall, Englewood Cliffs, New Jersey.
Jensen, J. R., D. J. Cowen, J. D. Althausen, S. Narumalani, and O. Weatherbee. 1993. An evaluation of
the CoastWatch change detection protocol in South Carolina. Photogrammetric Engineering and
Remote Sensing, 59(6):1039-1046.
Jensen, M. E., and R. Everett. 1994. An Overview of Ecosystem Management Principles. Pp. 7-16, In:
Jensen, M. E., and P. S. Bourgeron (eds.), Eastside Forest Ecosystem Health Assessment, Vol H,
Ecosystem Management: Principles and Applications. USDA Forest Service. PNW-GTR-318.
Kaufmann, M. R., R. T. Graham, D. A. Boyce Jr., W. H. Moir, L. Perry, R. T. Reynolds, R. L. Bassett, P.
Mehlhop, C. B. Edminster, W. M. Block, and P. S. Corn. 1994. An Ecological Basis for
Ecosystem Management. USDA Forest Service. General Technical Report RM-246.
Kirkland, L. L. 1994. Environmental Monitoring and Assessment Program Quality Management Plan.
U.S. Environmental Protection Agency, Office of Research and Development, Washington, D.C.
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.
Krummel, J. R., R. H. Gardner, G. Sugihara, R.V. O'Neill, and P. R. Coleman. 1987. Landscape patterns
in a disturbed environment. Oikos, 48: 321-324.
Kutz, F. W., and R. A. Linthurst. 1990. A systems-level approach to environmental assessment.
Toxicological and Environmental Chemistry, 28 (2-3):105-l 14.
Leahy, P. P., J. S. Rosenshein, and D. S. Knopman. 1990. Implementation Plan for the National Water-
Quality Assessment Program. Open-File Report 90-174. U.S. Geological Survey, Reston,
Virginia. 10pp.
Lillesand, T. M. and R. W. Kiefer. 1987. Remote Sensing and Image Interpretation. 2nd. Ed. John
Wiley and Sons, New York.
Ludwig,J. A., and J.F.Reynolds. 1988. Statistical Ecology. John Wiley and Sons, New York.
34
-------
Lunette, R. S., R. G. Congalton, L. K. Fenstermaker, J. R. Jensen, K. C. McGwire, and L. R. Tinney.
1991. Remote sensing and geographic information system data integration: error sources and
research issues. Photogrammetric Engineering & Remote Sensing, 57(6):677-687.
|
McNeely, J. A., K. R. Miller, W. Reid, R. Mittermeier, and T. Werner. 1990. Conserving the World's
Biological Diversity. IUCN, World Resources Institute, World Bank. WWF-US and
Conservation International, Washington, D.C.
Messer, J. J., R. A. Linthurst, and W.S. Overton. 1991. An EPA program for monitoring ecological
status and trends. Environmental Management, 17:67-78.
.
Milne, B. T. 1988. Measuring the fractal geometry of landscapes. Applications in Mathematics and
Computation, 27:67-79.
Milne, B. T. 1991. Heterogeneity as a multi-scale characteristic of landscapes. Pp. 69-84, In: J. Kolasa
and W. E. Waters (eds.), Ecological Heterogeneity. Springer-Verlag, New York.
I
Mouat, D. A., G. G. Mahin, and J. M. Lancaster. 1993. Remote sensing techniques in the analysis of
change detection. Geocarto International, 8(2):39-50. j
.
National Research Council. 1995. Review of EPA's Environmental Monitoring and Assessment
Program: Overall Evaluation. National Academy Press, Washington, D.C.
Nelson, R. F. 1983. Detecting forest canopy change due to insect activity using Landsat-MSS.
Photogrammetric Engineering and Remote Sensing, 49(9):1303-1314.
Noss, R. F. and A. Y. Cooperrider. 1994. Saving Nature's Legacy: Protecting and Restoring
Biodiversity. Island Press, Washington, D.C.
O'Neill, R. V., D. L. DeAngelis, J. B. Waide, and T.F.H. Allen. 1986. A Hierarchical Concept of
Ecosystems. Princeton University Press, Princeton, New Jersey.
O'Neill, R. V., R. H. Gardner and M. G. Turner. 1992. A hierarchical neutral model for landscape
analysis. Landscape Ecology, 7: 55-61. ;
Omernik, James M. 1987. Ecoregions of the United States. Map at a Scale of 1:7,500,000. Supplement
of the Annals of the Association of American Geographers, 77(1).
Omernik, James M. 1995. Ecoregions: A spatial framework for environmental management. In: W.S.
Davis and T.P. Simon (eds.), Biological Assessment for Criteria: Tools for Water Resource
Planning and Decision Making. Lewis Publishers, Chelsea, Michigan. (In press).
Pickett, S.T.A. 1989. Space-for-time substitutions as an alternative to long-term studies. Pp. 110-135,
In: Likens, G.E. (ed.), Long-Term Studies in Ecology: Approaches and Alternatives. Springer-
Verlag, New York.
Pilon, P. G., P. J. Howarth, R. A. Bullock, and P. O. Adeniyi. 1988. An enhanced classification
approach to change detection in semi-arid environments. Photogrammetric Engineering and
Remote Sensing, 54(12):1709-1716.
35
-------
Plotnick, R. K, R. H. Gardner, and R. V. O'Neill. 1993. Lacunarity indices as measures of landscape
texture. Landscape Ecology, 8: 201-211.
Rapport, D. J., C. Gaudet, J. Karr, B. Norton, and W. Jackson. 1995. Guiding principles for evaluating
landscape health. (Bioscience submitted).
«
Regier, H. A. 1993. The notion of natural and cultural integrity. Pp. 3-18, In: S. Woodley, J. Kay, and
G. Francis (eds.), Ecological Integrity and the Management of Ecosystems. St. Lucie Press,
DelRay Beach, Florida.
Riitters, K. H.-, R. V. O'Neill, C. T. Hunsaker, J. D. Wickham, D. H. Yankee, S. P. Timmins, K. B. Jones,
and B. L. Jackson. 1995. A factor analysis of landscape pattern and structure metrics.
Landscape Ecology, 10:23-39.
Schott, J. R., C., Salvaggio, and W. J. Volchok. 1988. Radiometric scene normalization using
pseudoinvariant features. Remote Sensing of Environment, 26:1-16.
Sharpe, D., J. Estes, and K. A. Herman. 1993. Proceedings of a Workshop: Environmental Monitoring
and Assessment Program - Landscape Ecology. U.S. Environmental Protection Agency, Office
of Research and Development, Research Triangle Park, North Carolina.
Shepard, J.R. 1964. A concept of change detection. Photogrammetric Engineering, 30(4):648-651.
Singh, A. 1989. Digital change detection techniques using remotely-sensed data. International Journal
of Remote Sensing, 10(6):989-1003.
Stauffer, M. L. and R. L. McKinney. 1978. Landsat image differencing as an automated land cover
change detection technique (interim report). Computer Science Corporation, report to National
Aeronautics and Space Administration (NASA), CSC/TM-78/6215, Greenbelt, Maryland.
Story, M. and R. S. Congalton. 1986. Accuracy assessment: a user's perspective. Photogrammetric
Engineering and Remote Sensing, 52(3):397-399.
Suter, G.W., U. 1990. Endpoints for regional ecological risk assessment. Environmental Management,
14:9-23.
Suter, G.W., n. 1993. Ecological Risk Assessment. Lewis Publishers, Chelsea, Michigan. 538pp.
Thornton, K.W., D. E. Hyatt, and C. B. Chapman. 1993. Environmental Monitoring and Assessment
Program Guide. EPA/620/R-93/012, U.S. Environmental Protection Agency, Office of
Research and Development, Washington, D.C.
Turner, M. G. 1989. Landscape ecology: the effect of pattern on process. Ann. Rev. Ecol. Syst,
20:171-197.
Turner, M. G., R. V. O'Neill, R. H. Gardner, and B. T. Mime. 1989. Effects of changing spatial scale on
the analysis of landscape pattern. Landscape Ecology, 3(3/4):13-162.
36
-------
Walker, J., F. Bullen, and B. G. Williams. 1993. Ecohydrological changes in the Murray-Darling Basin.
Part I. The number of trees cleared over two centuries. J. Appl. Ecol., 30:265-273.
Whitcomb, R. F., J. F. Lynch, M. K. Klimkiewicz, C.S. Robbins, B. L. Whitcomb, and D. Bystrak. 1981.
Effects of forest fragmentation on avifauna of the eastern deciduous forest. Pp. 125-205, In: R.L.
Burgess, and D.M. Sharpe (eds.), Forest Island Dynamics in Man-dominated Landscapes.
Springer-Verlag, New York.
Wickham, J. D. and K. H. Riitters. Sensitivity of landscape metrics to pixel size. International Journal of
Remote Sensing. (In press).
-
!
Wickham, J. D., and D. J. Norton. 1994. Mapping and analyzing landscape patterns. Landscape
Ecology, 9(l):7-23.
I
Wiens, J. 1989. Spatial scaling in ecology. Functional Ecology, 3:385-387.
Woodley, S., J. Kay, and G. Francis. 1993. Ecological Integrity and the Management of Ecosystems. St.
Lucie Press, Ottawa, Canada.
World Commission on Environment and Development (WCED). 1987. Our
University Press, New York.
Common Future. Oxford
37 -&U.S. GOVERNMENT HUNTING OFFICE: 199S - 650406/22060
-------
-------
-------
-------
-------
m
i
g
1
n%° c
o§3 5
o to Q.
fs-
. DO
o
""* CO
"a 5'
3- CD
< CO
BJ.CO
CD
CO
CD
o o m c
1111
MSB
> .
TO 3 CD
00 CD o
T3 "***
CD (O
CO CD
CD 13
(B O
8-=
o
o
m
9
CO _
Ol J
J>
------- |