United States
Environmental Protection
Agency
Office of Research and
Development
Washington, DC 20460
EPA/600/R-00/059
September 2000
www.epa.gov/ncerqa
Proceedings
2000 STAR Regional Scale Analysis
and Assessment Progress
Review Workshop
September 18-19, 2000
Las Vegas, Nevada
NATIONAL CENTER FOR ENVIRONMENTAL RESEARCH
-------
U.S. Environmental Protection
Region 5, Library (PL-12J)
77 West Jackson Boulevard, 12th Roof
Chicago. II 60604-3590
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Table of Contents
Introduction v
Section 1. Projects Initiated With Fiscal Year 1997 Support
Assessment of Forest Disturbance in the Mid-Atlantic Region: A Multiscale
Linkage Between Terrestrial and Aquatic Ecosystems 3
Keith N. Eshleman, Robert H. Gardner, Louis F. Pitelka, Steven W. Seagle,
Philip A. Townsend, William S. Currie, James N. Galloway,
James R. Webb, Alan T. Herlihy
Section 2. Projects Initiated With Fiscal Year 1998 Support
Multilevel Statistical Modeling for Generalizing From Case Studies 7
Richard A. Berk, JandeLeeuw, Richard Ambrose, Cindy Lin
Regional Scale Impacts of Phase I of the Clean Air Act Amendments: The Relation
Between Emissions and Concentrations, Both Wet and Dry 8
Gene E. Likens, Tom Butler
Regional Analysis of Variation in Adirondack Lake Ecosystems: Landscape Scale
Determinants of Dissolved Organic Carbon 9
Michael L. Pace, Charles D. Canham, Michael Papaik
Methodologies for Extrapolating From Local to Regional Ecosystem Scales: Scaling Functions
and Thresholds in Animal Responses to Landscape Pattern and Land Use 10
John A. Wiens, Beatrice Van Home
Section 3. Projects Initiated With Fiscal Year 1999 Support
Application of Remotely Sensed Data to Regional Analysis and Assessment
of Stream Temperature in the Pacific Northwest 15
Stephen Surges
Effects of N Deposition on Gaseous N Loss From Temperate Forest Ecosystems 17
Peter M. Groffman, Louis V. Verchot, Christopher Potter,
Mary Beth Adams, Ivan Fernandez, Lindsey Rustad
Regional Analysis of Net Ecosystem Productivity of Pacific Northwest Forests: Scaling Methods,
Validation, and Results Across Major Forest Types and Age Classes 19
Beverly E. Law, D. Turner, C. Daly, M.E. Harmon, S. Acker,
W. Cohen, M. Unsworth
Multiscale Effects of Forest Fragmentation and Landscape
Context on Population Health of Birds 20
Thomas E. Martin, R.L. Redmond, M.M. Han, M.E. Clark
The Office of Research and Development's National Center for Environmental Research iii
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Table of Contents (continued)
Regional Ecological Resource Assessment of the Rio Grande Riparian Corridor: A Multidisciplinary
Approach To Understanding Anthropogenic Effects on Riparian
Communities in Semi-Arid Environments 21
Jay Raney, Melba Crawford, Frank Judd, Gene Paull, Robert Lonard,
Amy Neuenschwander, Thomas Tremblay, William White
A Hierarchical Patch Dynamics Approach to Regional Modeling and Scaling 23
Jianguo Wu
Index of Authors 25
iv The Office of Research and Development's National Center for Environmental Research
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Introduction
The mission of the United States Environmental Protection Agency (EPA) is to protect public health and
safeguard and improve the natural environment—the air, water, and land upon which life depends. Achievement of
this mission requires the application of sound science to the assessment of environmental problems and to the evaluation
of possible solutions. The National Center for Environmental Research (NCER) at EPA is committed to providing
the best products in high-priority areas of scientific research through significant support for long-term research.
The Office of Research and Development's (ORD) Environmental Monitoring and Assessment Program
(EMAP) identifies integrating information across multiple scales as one of its largest challenges. This approach is
based on The Committee on Environment and Natural Resource's (CENR) publication entitled Integrating the Nation's
Environmental Monitoring and Research Networks and Programs, A Proposed Framework.
In support of this identified research need, NCER issued a Request for Applications (RFA) on Regional Scale
Analysis and Assessment in 1997. It was followed up with subsequent RFAs in 1998 and 1999. The purpose of these
solicitations was to support research that would lead to the development and demonstration of approaches to link site-
specific information with regional survey data and remote sensing imagery for conducting regional-level ecological
assessments. A total of 11 grants have been funded under this program. This research represents part of the extra-
mural component of ORD's Environmental Monitoring and Assessment Program (EMAP). You may find extensive
information about the EMAP Program at http://www.epa.gov/EMAP.
Annual program reviews such as this allow investigators to interact with one another and discuss progress and
findings with EPA and other interested parties. If you have any questions regarding the program, please contact the
Program Manager, Barbara Levinson, by telephone at 202-564-6911, or by e-mail at levinson.barbara@epa.gov.
The Office of Research and Development's National Center for Environmental Research
-------
Section 1.
Projects Initiated With Fiscal Year 1997 Support
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Assessment of Forest Disturbance in the Mid-Atlantic Region:
A Multiscale Linkage Between Terrestrial and Aquatic Ecosystems
Keith N. Eshleman1, Robert H. Gardner1, Louis F. Pitelka1, Steven W. Seagle1, Philip A. Townsend1,
William S. Currie1, James N. Galloway2, James R. Webb2, and Alan T. Herlihy3
'University of Maryland Center for Environmental Science, Appalachian Laboratory, Frostburg, MD; Department
of Environmental Sciences, University of Virginia, Charlottesville, VA; Department of Fisheries and Wildlife,
Oregon State University, Corvallis, OR
The objective of this project is to develop, test,
validate, and demonstrate an analytical framework for
assessing regional-scale forest disturbance in the Mid-
Atlantic region by establishing a multiscale linkage be-
tween forest disturbance and forest nitrogen export to
surface waters. It is hypothesized that excessive nitrogen
(N) leakage (export) from forested watersheds is a poten-
tially useful, integrative "indicator" of a negative change
in forest function that occurs in synchrony with changes
hi forest structure and species composition. This re-
search project focuses on forest disturbance associated
with historical defoliations by the gypsy moth larva at
spatial scales ranging from small watersheds to the entire
region.
The technical approach for establishing a multi-
scale linkage between forest disturbance (i.e., gypsy
moth defoliation) and N leakage to surface waters has
made use of extensive forest, forest disturbance, and
water quality data collected for the Mid-Atlantic region
at all spatial scales (intensive watershed sites, subregion-
al survey, regional survey, and remotely sensed data).
These data were supplemented with on-the-ground mea-
surements of forest species composition for a selected
sample of watersheds for which N export had been pre-
viously monitored.
In early 2000, an analysis was completed of
regional-scale forest disturbance and associated dissolved
N export from Shenandoah National Park (SNP), Vir-
ginia—a large contiguous area of forested land within the
Chesapeake Bay Watershed. Long-term watershed re-
search conducted in SNP indicates that annual export of
dissolved N from forested watersheds to surface waters
increases dramatically in response to vegetation distur-
bances. These results suggest that a parsimonious, em-
pirical unit N export response function (UNERF) model
can explain large percentages of the temporal variation in
annual N export from a group of small gaged forested
watersheds in the years following disturbance. The em-
pirical UNERF modeling approach is completely analo-
gous to the unit hydrograph technique for describing
storm runoff, with the model representing annual N ex-
port as a linear deterministic process both hi space and
time.
The purposes of this analysis were to: (1) test the
applicability of the UNERF model using quarterly
streamwater nitrate data from a group of ungaged water-
sheds hi SNP; (2) demonstrate a park-wide application of
a regional UNERF model that references the geographic
distributions of bedrock geology and the timing of gypsy
moth defoliation over the entire SNP area; and (3) visu-
alize the temporal and spatial patterns hi vegetation dis-
turbance and annual dissolved N export through the use
of PC-based animation software.
The results of this analysis showed that: (1) for-
ested ecosystems within SNP normally retain a very high
percentage of atmospherically deposited N, but forest
disturbances such as insect defoliation can dramatically
alter the input/output balance; (2) annual nitrogen export
from SNP forests began increasing hi 1987 from a base-
line rate of about 0.1 kg/ha, peaked in 1992 at an aver-
age rate of 1.68 kg/ha (more than a 1,500% increase),
and has since been steadily declining (see Figure 1);
(3) natural biogeochemical processes need to be con-
sidered in modeling N export dynamics of forests; and
(4) linear systems models can apparently provide parsi-
monious approximations of ecosystem complexity that
are useful for purposes of regionalization.
In the final year of the research project, this type
of integrative analysis will be extended to other forested
watersheds hi the Chesapeake Bay Watershed. The focus
will be on first through third order drainages hi the Mid-
Atlantic Highlands Area (MAHA) that comprise a target
subpopulation of the EMAP surface water survey in the
eastern United States. In completing this analysis, his-
torical remote-sensing data (AVHRR) will be used as a
means of better characterizing the spatial and temporal
variations hi forest defoliation within the region during
the last 15 years.
The Office of Research and Development's National Center for Environmental Research
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Modeled SNP-Wide N Export and "Heavy" Defoliation
Figure 1. Modeled SNP-wide N export to surface waters and heavy gypsy moth defoliation during the period 1980-1999.
The Office of Research and Development's National Center for Environmental Research
-------
Section 2.
Projects Initiated With Fiscal Year 1998 Support
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Multilevel Statistical Modeling
for Generalizing From Case Studies
Richard Berk, Jan de Leeuw, Richard Ambrose, Cindy Lin
Institute of the Environment, University of California at Los Angeles, Los Angeles, CA
Case studies, the foundation of environmental re-
earch, are too often faced with a very difficult "so
what?" criticism. Just because certain findings apply to
a given study site does not necessarily mean that they
apply to any others. Yet, the goal of such research com-
monly is to arrive at broadly relevant conclusions.
In this research project, a generalization of multi-
level statistical modeling is employed to consider the de-
gree to which credible conclusions across a set of case
study sites can be drawn. Multilevel modeling within a
regression analysis framework allows for the explicit
consideration of how statistical summaries vary system-
atically over a set of sites. As a result, the degree to
which findings can be applied beyond a single site is ad-
dressed directly.
The contribution of this research project is to pro-
vide a generalization of multilevel modeling and the nec-
essary software, so that the generalized linear model can
be worked with, including such features as spatial and
temporal dependence, inherently nonlinear relationships,
and latent variables. For example, outcomes can be ca-
tegorical, and sites closer to one another in space may be
treated systematically as more alike. The emphasis in
this research project is on technique, with the data used
as an illustration. The data come from a Regional En-
vironmental and Monitoring Assessment Program (R-
EMAP) project, U.S. Environmental Protection Agency
Region 9, undertaken in conjunction with the Environ-
mental Science and Engineering Program at the Univer-
sity of California at Los Angeles. The research involves
a spatially intensive stream bioassessment monitoring
study for Calleguas Creek Watershed, Ventura County,
CA.
This project includes sampling of riparian habitat
and streams for two seasons between 1999 and 2000.
Data are collected at 70 sites. Field assessment in the
watershed includes water quality, physical habitat assess-
ments (e.g., measures and/or visual estimates of channel
cross-sectional dimensions, substrate, fish cover, bank
characteristics, and riparian vegetation structure), and
benthic and fish community sampling. The project seeks
to assess the current ecological condition of coastal
Southern California streams and examine impacts of land
use on water quality and aquatic ecosystem integrity.
The Office of Research and Development's National Center for Environmental Research
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Regional Scale Impacts of Phase I of the Clean Air Act Amendments:
The Relation Between Emissions and Concentrations. Both Wet and Dry
Gene E. Likens and Tom Butler
Institute of Ecosystem Studies, Millbrook, NY
A comparison of data records in the 1990s both
before (1991-94) and after (1995-97) implementation of
Phase I of the Clean Air Act Amendments (CAAA) of
1990 for seven regions of the eastern United States
shows a significant reduction in S02 emissions for most
states, except for Texas, North Carolina, Illinois,
Florida, and Alabama. However, of the major NOX
emitting states, only two eastern states (New York and
Pennsylvania) show significant declines in NOX.
A pattern of large declines in S02 emissions
(> 20 %) after CAAA implementation, and large declines
in precipitation concentrations of SO4= and H+, as well
as air concentrations of S02 and S04= (components of
dry deposition) exists for most regions of the eastern
United States. The regions examined include: Northern
New England and the Adirondack Mountains (NE); New
York, excluding the Adirondack Mountains (NY); Pen-
nsylvania, northern West Virginia and eastern Ohio
(PA); Delaware, Maryland, and eastern Virginia (M-A);
Ohio, eastern Indiana, and northern Kentucky (OH);
Illinois and western Indiana (IL); and the southern Appa-
lachian Mountains and surrounding area (SA). In most
cases, the emission/concentration relations are close to
1:1 (a near equal percent decline in concentrations for a
given percent decline in emissions) when the source re-
gion based on 15-hour air mass back trajectories is used
for the New England region, and source regions based on
9-hour back trajectories are used for the other eastern
regions.
The southern Appalachian Mountain region is an
acid-sensitive area receiving high levels of acidic depo-
sition that has not seen an appreciable improvement in
precipitation acidity. This area also has had the least im-
provement in wet and dry sulfur concentrations of the
areas examined (see Table 1). Precipitation base cations
(Ca++ and Mg++) show a pattern of increasing or level
concentrations when comparing 1990-1994 to 1995-1998
data for six of the seven regions that were examined.
Ammonium concentrations generally have changed less
than 10 percent, except for the Illinois region and the
southern Appalachian Mountain region, both of which
increased more than 15 percent.
A future focus of this work will include an eval-
uation of how changing emission levels will affect New
England waters, using the Hubbard Brook Experimental
Forest as a test case.
Table 1. Changes in emissions of: (a) NOX, (b) SO2, and (c) combined emissions from source regions based on 9-hour back trajectories
(except for NE, which is based on 15-hour back trajectories), expressed as percentagess; and changes, also expressed as per-
centages, in: (d) precipitation SO4" concentrations, (e) SO2 and paniculate SO4" air concentrations (sources of dry S depo-
sition), and (f) precipitation acidity (H+ concentrations) for the study regions. The years 1991-1994 (pre-CAAA implemen-
tation) are compared with 1995-1997 (post-CAAA implementation). The percentage change in parentheses for NE and under
(b) represent changes in emissions when the Province of Ontario is included as part of the source region.
91-94
vs.
95-97
NE
NY
PA
M-A
OH
IL
SA
EMISSIONS
(a)
NOX
% change
-10
-12
-7
-3
+2
+2
+5
(b)
S02
% change
-20 (-19)
-24 (-22)
-23
-22
-23
-20
-19
(c)
2S+N (moles)
% change
-17
-20
-18
-15
-16
-13
-11
CONCENTRATIONS
(d)
SO4= wet
% change
-25
-24
-21
-24
-18
-12
-6
(e)
Sdry
% change
-29
-23
-21
-18
-23
-20
-10
(f)
H+wet
% change
-14
-22
-17
-19
-16
-17
<-l
The Office of Research and Development's National Center for Environmental Research
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Regional Analysis of Variation in Adirondack Lake
Ecosystems: Landscape Scale Determinants of Dissolved Organic Carbon
Michael L. Pace, Charles D. Canham, and Michael Papaik
Institute of Ecosystem Studies, Millbrook, NY
Lakes represent a critical ecological and economic
resource. The consequences of many types of air and
water pollution have been manifested and best under-
stood in the study of lakes, especially at regional scales,
as for example in studies of the consequences of eutro-
phication and acidification.
The goal of this research project is to develop an
approach to assessing variation in lake-watershed Inter-
actions at a regional scale. Specifically, it is sought to
predict variation in lake ecosystem properties as a func-
tion of landscape characteristics, especially in relation to
landscape and within lake processes that determine var-
iation in lake dissolved organic carbon (DOC).
Lake DOC is an important measure of overall lake
conditions and is one of the primary features that varies
among lakes in landscapes where lakes are not heavily
impacted by nutrient loading. The focus of this study is
on the Adirondack Park of New York State. This region
represents one of the largest areas of wilderness and
minimally developed land in the United States east of the
Mississippi River. Because of extensive prior studies,
there are large numbers of data sets on the watersheds
and lakes of the Adirondacks.
From this information, data on the water chemistry
and complementary watershed data have been assembled
for more than 600 lakes in the Oswegatchie Black and
Upper Hudson Watersheds of the Park. This data set is
a representative sample of the larger number of lakes in
the region. Data layers allow for visualization of the ele-
vation contours (from a new digital elevation map with
10 m resolution); all standing water bodies; uplands in
five categories of land use; wetlands in eight categories
derived from a more extensive classification of wetland
type; the extent of wetlands that fringe lakes at 50 m,
100 m, and 200 m intervals; streams; watershed bound-
aries; roads; and developed areas. A complete chemistry
data set for the study lakes has been assembled from
earlier water quality studies conducted by the Adirondack
Lake Survey. A series of watershed statistics have been
added to this data set using Geographic Information Sys-
tem (GIS) analysis.
Two analytical projects currently are underway.
The first is a multivariate statistical analysis of watershed
and lake characteristics with a goal of empirically eval-
uating controls on the variation in DOC observed among
lakes at a regional scale.
The second analysis is the development of a max-
imum likelihood mass balance model for lake DOC.
This model will describe the inputs, outputs, and within-
system processing of DOC in lakes. A key feature of
this model is the spatial analysis of inputs of DOC along
flow paths to a lake from various landscape source areas.
Flowpath data sets have been developed in the initial GIS
analysis of the watersheds to parameterize these input
functions. A simulated annealing algorithm will be used
to iteratively solve the basic model equation describing
inputs, processing, and outputs. This approach will
solve simultaneously for the parameter values that max-
imize the likelihood of the data set where mean lake
DOC is predicted.
The Office of Research and Development's National Center for Environmental Research
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Methodologies for Extrapolating From Local
to Regional Ecosystem Scales: Scaling Functions and
Thresholds in Animal Responses to Landscape Pattern and Land Use
John A. Wiens and Beatrice Van Home
Colorado State University, Fort Collins, CO
The objective of this research project is to develop
and test new concepts and methods for evaluating how
the structure and configuration of landscapes change with
changes in spatial scale, and how these changes in land-
scape scale relate to the scales at which various groups of
plants and animals respond to the environment, which
ultimately affects their distribution and abundance.
Specifically, the aims of this project are to: (1) de-
velop ways of modeling how the scaling of species'
responses to environmental variation is coupled to scale-
dependent changes in landscape composition and struc-
ture; (2) develop protocols for determining the scales at
which the responses of species and communities are
likely to exhibit thresholds, based on functional proper-
ties of the species; and (3) develop mathematical scaling
functions that can be coupled with Geographic Infor-
mation Systems (GISs) to make use of remote sensing.
The overall goal is to derive scaling functions and GIS-
based spatial models that can be used to assess how in-
formation gathered at fine scales in intensive studies can
be extrapolated to the broad scales of ecological mon-
itoring and environmental risk assessment.
There are three phases to this research. The first
phase is using existing data sets (faunal surveys, GIS lay-
ers, remote sensing) to explore how to link the spatial
structure of landscapes with the distribution of organisms
over multiple scales, and to determine the most appro-
priate statistical approaches to such scaling relationships.
The second phase is gathering information on the spatial
configuration of landscapes and the distribution of key
groups of organisms (vascular plants, beetles, butterflies,
and birds) at five study sites located on a broad-scale gra-
dient from the shortgrass steppe to the tallgrass prairie
(see Figure 1). This series spans a gradient of increas-
ing precipitation, productivity, and land-use intensity,
and will provide the foundation for detailed multiscale
analyses. The third phase is using the data to derive and
test mathematical, GIS, and spatial models that relate
changes in species occurrences, functional group com-
position, and biodiversity at multiple scales to the scaling
properties of landscapes.
Data sets have been used for grasshopper distri-
butions in eastern Wyoming and bird and vegetation
distributions in Idaho shrub steppe to explore several ap-
proaches to the analysis of spatially referenced data.
These studies have evaluated: (1) the differences be-
tween using land-cover classification data and quantified
or ordinal data in multivariate ordinations in GIS-based
analyses of the environmental correlations of animal
distributions; and (2) ways of determining the contri-
butions of geographical distance to assessments of com-
munity-environment relationships. The studies in phases
two and three are just beginning.
These studies show that multivariate ordinations
may reveal features of environmental associations of
organisms that are not apparent when using classified
(categorical) data, and that geostatistical modeling can
considerably improve the explanatory power of causal
models of variations in community composition.
Detailed information on the spatial properties of
landscapes and organism distributions at scales from
1 m2 to tens of km2 will be gathered through field sam-
pling in June-September 2000. Analysis of these sam-
ples will be conducted during the following 8 months,
along with generation of data sets derived from on-
ground surveys, published literature, and remote sensing
to develop accompanying GIS layers. Combined with
the results of the ongoing examination of the properties
of various statistical approaches to spatial pattern and
scaling, these will form the foundation for the initial
stages of model development. The resulting models will
be tested using field data gathered hi summer 2001,
followed by another phase of sample analysis and model
refinement.
10
The Office of Research and Development's National Center for Environmental Research
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
^•::^'^^^mm
***v.V :" **T * • * •«*" ^tar-S"" ' ' - ' -••,,: r*Hi. s; •? "•'• .'£w53P*
,^ *** . *" , MS&gf ' \ •* • , „ ' »V ' "^W**^; ^*-'<*&*£f -^ "-
•jgjag +' ^ ^ .^^ ™ » *- '_ ' * - v*t '* ? ^•.'•^^ss- *' '
JJP*./ , ^^^^' '''- *: ^ ^ ».- ~ *> t* - ' ^ \ "^ *^_, '^V ^^ -" "^ '
' -; • ^-
--* !*• T'*ti*
*N,:/; •*, ^PS**. ^fc^s^N -!*w-«*'~ KL^
'
Figure 1. The Western Great Plains, showing the region encompassed by the scaling transect. LTER = long-term ecological research site.
SGS = shortgrass steppe.
The Office of Research and Development's National Center for Environmental Research
11
-------
Section 3.
Projects Initiated With Fiscal Year 1999 Support
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Application of Remotely Sensed Data to Regional Analysis
and Assessment of Stream Temperature in the Pacific Northwest
Stephen Surges
Department of Civil and Environmental Engineering, University of Washington, Seattle, WA
The characteristics of water vary throughout a
stream network, presenting a major obstacle to regional
water quality assessments. Figure 1 illustrates the var-
iation in stream temperature observed on a single day in
one stream network (covering approximately 100 km2) hi
western Washington State. The goal of this project is to
develop efficient methods for regional assessments of
stream temperature using remotely sensed thermal infra-
red (TIR) images of stream corridors to augment in-
stream temperature monitoring. The TIR images are ob-
tained from the ASTER camera mounted on the Terra
Satellite that was launched on December 18, 1999, as
well as from a forward-looking TIR radiometer mounted
on an aircraft.
The research is focused on water temperature to
illustrate and explore methods of water quality assess-
ment because water temperature is biologically impor-
tant, it is affected by anthropogenic activities, and the
kinetic temperature of visible surfaces can be measured
over large areas using aircraft- and satellite-based instru-
ments . A hierarchical approach is used in which temper-
ature monitoring techniques are adapted to address die
different issues faced hi different parts of a stream
network. The range is from large streams that are clear-
ly visible from both remote platforms, but may not be
thoroughly mixed, to small streams that constitute only
a portion of any "pixel" in a remotely sensed image from
the Terra Satellite.
Although the specific methods for monitoring
stream temperature vary depending on location in a
stream network, this approach takes advantage of the
synergy between remotely-based and in-stream instru-
ments as illustrated by four examples. First, remotely
based instruments identify strong longitudinal (down-
stream) temperature gradients and the appropriate loca-
tion and "density" of in-stream monitoring stations.
Second, in-stream measurements quantify the spatial
temperature structure of streams that is represented by
single values (i.e., pixels) hi remote images. Third, in-
stream instruments provide a continuous time-series of
temperature for periods when remotely sensed images
are not available. Fourth, remotely based measurements
facilitate interpolation of water temperatures between in-
stream measurements.
Existing water quality monitoring programs provide
a spatially limited view of conditions in stream networks.
Synoptic surveys, such as those illustrated in Figure 1,
expand the purview of a water quality monitoring pro-
gram, but many gaps hi the coverage remain. The target
is to produce stream temperature maps that fill hi the
gaps and provide a more complete picture of water
quality in the region.
The Office of Research and Development's National Center for Environmental Research
15
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
10km
Base map: USGS 1:100,000, Tacoma, WA
Figure 1. Mid-day stream temperatures (C) on August 19, 1998, throughout the Soos Creek Network, King County, WA.
16
The Office of Research and Development's National Center for Environmental Research
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Effects of N Deposition on Gaseous
N Loss From Temperate Forest Ecosystems
Peter M. Groffman 1, Louis V. Verchot '*, Christopher Potter1, Mary Beth Adams3, Ivan Fernandez4,
and Lindsey Rustad5
'Institute of Ecosystem Studies, Millbrook, NY; 2NASA-Ames Research Center, Moffett Field, CA; 3USDA Forest
Service, Timber and Watershed Laboratory, Parsons, WV; 4University of Maine, Orono, ME; 5USDA Forest
Service, Northeastern Research Station, Pownal, ME (*Now at the International Centre for Research in
Agroforestry, Nairobi, Kenya)
Although much effort has gone into determining the
fate of atmospheric N in temperate forest ecosystems,
many uncertainties remain as to just where N is stored
and what processes and pathways influence N retention
and/or loss. One of the largest areas of uncertainty is
gaseous loss. This flux may be large and may be very
sensitive to N deposition.
The specific objectives of this research project are
to: (1) determine the importance of gaseous loss of N
from temperate forest ecosystems; (2) determine the im-
pacts of N deposition on gaseous loss of N from these
ecosystems; (3) test a mechanistic model that relates N
gas emissions to N availability and soil moisture content;
and (4) develop a new and more mechanistic version of
the daily NASA-CASA ecosystem model for N gas emis-
sions that can be applied at the regional level using
satellite remote sensing and other spatial data sets in a
Geographic Information System (GIS) format. This new
simulation model will be used to assess trends in N
cycling over gradients of N deposition in the northeast
United States and to project changes hi N gas fluxes with
changing air pollution.
Gas fluxes (NO, N20, N2) will be sampled on a
monthly basis at five sites along an N deposition gradient
hi the northeast United States: Fernow Experimental
Forest, WV; Catskills State Forest, NY; Hubbard Brook
Experimental Forest, NH; Harvard Forest, MA; and
Bear Brook Watershed, ME. Several additional mea-
surements of factors known to control flux rates (e.g., N
pool sizes and turnover rates, denitrification rates, soil
temperature, soil pH, and soil moisture) will be made.
Samples will be taken in both N fertilized and unfertil-
ized plots at each location. These data then will be used
to develop a new and more mechanistic version of the
daily NASA-CASA ecosystem model for N gas emis-
sions that can be applied across a 10-state region (ME,
NH, VT, MA, RI, CT, NY, NJ, PA, and WV) using
satellite remote sensing and other spatial data sets in a
GIS format. This new simulation model will be used to
assess trends in N cycling over gradients of N deposition
hi the northeast United States and to project changes in
N gas fluxes with changing air pollution.
Field measurements for this project have not begun
yet; however, one of the first tasks has been to establish
experimental designs to capture the main factors in-
fluencing N gas fluxes at the different sites (e.g., topo-
graphic position, N availability, species composition).
Preliminary data from one of the sites, the Hubbard
Brook Experimental Forest, suggest that tree species
composition may be a strong controller/indicator of N
gas fluxes at these sites (see Figure 1).
One of the least well-understood facets of the com-
plex N cycle is loss of N hi gaseous forms from the soil.
Quantifying this flux is important from the perspective of
understanding the fate of N deposited on these eco-
systems from the atmosphere and for determining the
source strength of northeastern forests hi regional and
global budgets of NO (an ozone precursor) and N2O (a
greenhouse gas). The modeling effort will permit pla-
cing constraints on the regional contribution of the soil
source of N gases and make projections of the impact of
increasing N saturation on the regional atmospheric
budget of these gases. Regional quantification of gas-
eous N loss and improved understanding of the spatial
variability of this process will be important for devel-
opment of critical load tolerances for N deposition.
Sites currently are being established, instruments
are being purchased and set up, and field and laboratory
protocols are being finalized. Flux measurements should
begin by early June 2000.
The Office of Research and Development's National Center for Environmental Research
17
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
so
40
o>
z
D)
30
_o
= 20
1
o>
O
10
10
15 20
Time (days)
25
30
35
Figure 1. N2O production in soils from sugar maple and yellow birch stands at the Hubbard Brook Experimental Forest, NH. Soils were
incubated in the laboratory for 33 days. All differences between points were statistically significant at p < 0.05.
18
The Office of Research and Development's National Center for Environmental Research
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Regional Analysis of Net Ecosystem
Productivity of Pacific Northwest Forests: Scaling Methods,
Validation, and Results Across Major Forest Types and Age Classes
Beverly E. Law, D. Turner, C. Daly, M.E. Harmon, S. Acker, W. Cohen, M. Unsworth
Departments of Forest Science, Geosciences and Atmospheric Sciences, Oregon State University, Corvallis, OR
This project started on July 1, 2000. The objec-
tives are to: (1) develop and test a regional scale ap-
proach that combines modeling, data from remote
sensing, sample surveys, and intensive research sites to
better estimate variation in the carbon balance of forest
ecosystems in the Pacific Northwest; and (2) apply this
strategy to investigate how processes controlling var-
iation in net ecosystem productivity are influenced by
forest development, disturbances, and contrasting climat-
ic conditions.
Net ecosystem productivity (NEP) is a critical char-
acteristic of terrestrial ecosystem response to environ-
ment. Processes controlling NEP operate on a variety of
temporal and spatial scales and are influenced by plant
physiology, forest age or developmental stage, climate,
and disturbance. This project will simulate productivity
and NEP in Oregon and Washington using a combination
of remote sensing, survey data, process-level measure-
ments in different age classes of forests, and process
models. Model outputs will be tested using detailed eco-
system studies at intensive sites, more basic ecological
measurements at other existing intensive sites, and sur-
vey data from Forest Health Monitoring (FHM) and
Forest Inventory and Analysis (FIA) plots.
For each 30 x 30 m unit of ground, forest produc-
tivity will be predicted and evaluated for an east-west
longitudinal swath along a steep climatic gradient
through central Oregon from the coast to the semi-arid
east side of the Cascade Mountains, and a north-south
latitudinal swath from the southern Oregon border to
southern Washington. BIOME-BGC, a physiologically
based process model, will generate current NEP, produc-
tivity, and "carbon stress index" estimates for the region
for a mean climate year, 1999, and 2000. STAND-
CARB, an ecosystem process model, will be used to esti-
mate current carbon pools by accounting for long-term
trends in NEP. BIOME-BGC will be initialized using
remote sensing (Thematic Mapper) estimates of forest
cover type, age class and the amount of leaf area present
for photosynthesis, and soil survey data. The model will
be driven by spatially distributed climate data based on
interpolations of weather station data by climate models.
Remotely sensed variables will be validated with data
from intensive sites and the survey sites, and new mea-
surements in underrepresented forests.
Model predictions of carbon budget components
will be validated with flux data, other intensive site data,
and survey data. Sensitivity of NEP to forest type, de-
velopmental stage, disturbance, and interannual variabil-
ity hi climate will be evaluated. A practical approach for
linking remote sensing, sample surveys, and intensive
site data via modeling at the regional scale will be de-
monstrated.
The data acquisition phase will start soon, so there
are no findings to report as of yet. In the first year of
this project, existing data will be assembled, remote sen-
sing imagery will be acquired, and the spatially derived
climate data sets will be produced. Working with stat-
isticians, a sample design for the field measurements will
be developed. The field season will begin in June 2001.
Modeling activities will include testing the models with
existing site data across the region.
The Office of Research and Development's National Center for Environmental Research
19
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Multiscale Effects of Forest Fragmentation and
Landscape Context on Population Health of Birds
Thomas E. Martin, R.L. Redmond, M.M. Hart, andM.E. Clark
Montana Cooperative Wildlife Research Unit, The University of Montana, Missoula, MT
Data from intensive studies of avian demography at
replicated sites and multiple spatial scales within and
among geographic regions will be used to assess the ef-
fects of spatial scale and landscape context on major fac-
tors (nest predation and Brown-headed Cowbird para-
sitism) that influence the health of bird populations.
Results from this work will provide new and more
general insight into the influence of spatial scale on frag-
mentation and landscape context in relation to nest pre-
dation and cowbird parasitism in birds. Moreover, re-
sults of this work then can be applied to land cover data
across North America and across time to examine and
predict the potential demographic consequences of future
land-use changes on bird populations.
A Geographic Information System (GIS) will be
used to characterize landscapes, particularly hi terms of
forest fragmentation, at roughly 30 sites across the
United States using bird demographic data compiled for
the Breeding Biology Research and Monitoring Database
(BBIRD) program. At these sites and the more than 250
plots within them, data on nesting success have been
collected from more than 30,000 nests of more than 200
species of birds.
The BBIRD and GIS data will be used to examine:
(1) the spatial scales that influence demographic pro-
cesses (i.e., predation and parasitism) within and among
geographic regions; (2) the role of cover type (e.g., for-
est, agriculture, human habitations, etc.) hi the landscape
(landscape context) on predation and parasitism relation-
ships; and (3) predation and parasitism relationships with
landscape characteristics among regions.
The information on predation and parasitism rela-
tionships obtained from these three sets of analyses will
be used to: (1) develop predictive models of bird demo-
graphic responses to forest fragmentation throughout
North America; (2) examine the demographic consequen-
ces among functional groups (i.e., nest types, habitat
requirements) to determine variation in population sensi-
tivity and identify high-risk species and species groups;
and (3) model these demographic relationships in terms
of population sustainability to attempt to identify land-
scape conditions that support source (self-sustaining)
populations.
At this time, efforts are focused on building the
necessary GIS databases for characterizing landscapes.
GIS data are being compiled on a site-by-site basis; each
site's landscape is defined by a circle with a 100 km
radius centered on that site. The National Land Cover
Data set (NLCD) provides a consistent layer for the con-
terminous United States at a spatial resolution of 30 m.
Other GIS layers include digital elevation data, hydro-
graphy, roads, and the locations of the BBIRD sites.
Once the databases are assembled, predation and para-
sitism will be examined, then modeling can begin.
20
The Office of Research and Development's National Center for Environmental Research
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Regional Ecological Resource Assessment of the Rio Grande
Riparian Corridor: A Multidisciplinary Approach To Understanding
Anthropogenic Effects on Riparian Communities in Semi-Arid Environments
Jay Raney1, Melba Crawford2, Frank Judd3, Gene Paull4, Robert Lonard3, Amy Neuenschwander2,
Thomas Tremblay', and William White'
'Bureau of Economic Geology and2 Center for Space Research, The University of Texas at Austin, Austin, TX;
3The University of Texas-Pan American, Edinburg, TX; 4The University of Texas at Brownsville, Brownsville, TX
Riparian ecosystems of the southwestern United
States are characterized by high species diversity and are
among the most productive ecosystems of North Ameri-
ca. The rapid decline of riparian ecosystems throughout
the United States has made riparian conservation a focal
issue for the public, federal and state governments, and
private organizations.
Among the objectives of this 3-year multidisciplin-
ary study of the riparian corridor are to: (1) acquire and
analyze high-resolution remotely sensed data from mul-
tiple sensors; (2) integrate existing and new field data
and remotely sensed data into a Geographic Information
System (GIS); (3) ascertain whether the native com-
munities are maintaining themselves and identify the
topographic, edaphic, and other ecological factors that
perpetuate these communities; (4) interpret spatial var-
iations in riparian habitats, including comparisons of the
north and south banks of the Rio Grande; (5) analyze
temporal changes at specific locations; and (6) develop
a foundation for future analysis of riparian floodplain
communities by linking local and remotely sensed region-
al data using GIS.
The study area includes the lower reach of the Rio
Grande from Falcon Dam to the mouth of the river (see
Figure 1). Detailed, local-scale, ecological transects of
dominant riparian vegetation will be correlated with
high-resolution videography and multispectral data to
delineate the spatial extent of the riparian community.
This correlation will provide ground truth for the clas-
sification output from high-resolution (4-7 m) hyper-
spectral and airborne synthetic aperture radar (SAR)
data. Classification output from high-resolution data will
in turn provide the class mixtures for medium-resolution
(20-30 m) Landsat™ and SPOT multispectral data that
cover the entire study area, on both sides of the Rio
Grande. Data on geology, soils, land use, water quality
and hydrology, and topography from Topographic Syn-
thetic Aperture Radar (TOPSAR), as well as airborne
laser altimetry data acquired for the study, will be in-
vestigated as additional inputs to the classification pro-
cess and will be used to help explain temporal and spatial
changes in ecological resources indicated in the remotely
sensed data. GIS-based spatial models and statistical
modeling results will be used to predict the expected
future effects of landscape change on plant distribution,
biodiversity, and functional organization at multiple
scales of resolution. The developed methodologies will
help to guide future assessments of riparian regions.
Only very preliminary analyses of data have been
completed as of April 2000. Initial integration of data
into a GIS and preliminary analysis reveal the utility of
certain data sets. For example, U.S. soil maps include
parameters such as conductivity or soil salinity, which
allow construction of specific GIS layers that can be used
to examine relationships with riparian vegetation. One
difficulty has been correlating Mexican soil maps with
U.S. soil maps because of differences in scale and ter-
minology. Preliminary analysis of two airborne hyper-
spectral data-imaging systems, which have similar spatial
resolution but different spectral coverage, shows that
riparian vegetation composition is better defined by the
sensor that includes longer wavelength infrared bands.
The next steps are to continue acquiring needed
data, classifying remotely sensed data, ground truthing,
and to begin preliminary analysis and testing of the meth-
odology.
The Office of Research and Development's National Center for Environmental Research
21
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Figure 1. Map of Lower Rio Grande valley showing approximate location of existing ecological transects and example of riparian vegetation
in the Santa Ana National Wildlife Refuge.
22
The Office of Research and Development's National Center for Environmental Research
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
A Hierarchical Patch Dynamics
Approach to Regional Modeling and Scaling
Jianguo Wu
Department of Life Sciences, Arizona State University West, Phoenix, AZ
Scaling is imperative for understanding and pre-
dicting broad-scale dynamics based on fine-scale obser-
vations. To scale up or down, it is necessary to con-
struct multiscale models or to link individual models at
different scales, which together capture the spatial het-
erogeneity of pattern and process. Hence, to develop
cross-scale models, there are both theoretical and techni-
cal challenges.
To meet these challenges, a hierarchical patch dy-
namics modeling (HPDM) approach is proposed (see
Figure 1). HPDM involves three basic steps. First,
based on empirical data, the patch hierarchies relevant to
the phenomena and objectives of study using the prin-
ciples of loose vertical and horizontal coupling and near-
decomposability need to be identified. The second step
is to develop and validate unit hierarchical models in
which the next higher level provides boundary condi-
tions and other constraints, while dynamics at the next
lower level are described mechanistically. The third step
is to link unit hierarchical models across scales through
an input-output chain.
The HPDM approach will be discussed and illus-
trated through an example of modeling the pattern and
process of urbanization in the Phoenix metropolitan area.
Increase both grain
and extent
Figure 1. Hierarchical scaling or extrapolating information along a hierarchical scaling ladder. Scaling up or down is implemented by
changing model grain size, extent, or both across successive domains of scale.
The Office of Research and Development's National Center for Environmental Research
23
-------
2000 STAR Regional Scale Analysis and Assessment Progress Review Workshop
Index of Authors
Berk, R.A., 7
Burges, S., 15
Eshleman, K.N., 3
Groffman, P.M., 17
Law, B.E., 19
Likens, G.E., 8
Martin, T.E., 20
Pace, M.L., 9
Raney, J., 21
Wiens, J.A., 10
Wu, J., 23
The Office of Research and Development's National Center for Environmental Research 25
-------
U.S. Environmental Protection Agency
Region 5, Library (PL-12J)
77 West Jackson Boulevard, 12th FtaBT
Chicago. II 60604-3590
-------
g
9
f
------- |