UniMdl AMMphwIe R«i«aretJ and
•nvtaonmOTtt AMMMmmi Ubontoiy
Agmcy
R«s»arch and Devttopment November igoi
EPA 600/3-91/074
PROJE|CT REPORT
THE ROLE OF CLIMATE IN FOREST MONITORING AND
ASSESSMENT? A MEW ENGLAND EXAMPLE
-------
THE ROLE OF CLIMATE IN FOREST MONITORING AND
ASSESSMENT: A NEW ENGLAND EXAMPLE
Ellen J. Cooler, Sharon K. LeDuc, Lawrence Truppi1
Ecosystem Exposure Research Division
Atmospheric Research and Exposure Assessment Laboratory
Research Triangle Park, North Carolina
27711
Donald R. Block2
UNISYS Corporation
Research Triangle Park, NC
27709
November 1991
I/ On Assignment from the Air Resources Laboratory, National Oceanic and Atmospheric
Administration
21 Work completed while employed by ManTech Environmental Technology, Inc., Research
Triangle Park, NC
-------
DISCLAIMER
The information in this document has been funded in part by
the United States Environmental Protection Agency under contract
number 68-02-444. It has been subjected to Agency review and
approved for publication. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
-------
ABSTRACT
The development of climatological information products to
support ecological data collection and analysis is described. The
scope of research is narrowed to issues of direct interest to the
joint U.S. Environmental Protection Agency Environmental Monitoring
and Assessment Program and U.S. Department of Agriculture Forest
Service New England Forest Health Monitoring program.
Characteristics of climatological persistence and recurrence
that are especially critical to New England forest health and
productivity are identified. These include physical disturbance
events (tornadoes, high winds and wet snowfall), drought, growing
degree days and late spring freezes. Climatological data are
assembled and presentations developed based on the analysis issue
to be addressed: background (status and persistence); most recent
decade (short-term trends); and most recent sampling year (near-
term impacts). A Geographic Information System is used for
presentation, data management and analysis.
Major research findings focus on the application of climate
data and products to operational ecological monitoring and analysis
situations. Possible future activities are identified in the
areas of new climatologies, program design, database acquisition or
development and applied research. All these efforts would result
in significant contributions to the development of a more coherent
theory of natural disturbance and ecosystem response.
111
-------
TABLE OF CONTENTS
ABSTRACT ill
List of Figures vi
List of Tables ix
Acknowledgements x
I. Background 1
A. Project Overview 1
B. Conceptual Frameworks and Definitions for
Forest/Climate Research 2
1. Definitions 2
a. Weather and Climate 2
b. Scale 3
- mesometeorological events 3
- synoptic events 3
- landscapes 3
- patches 4
c. Disturbance 4
- spatial distribution 11
- disturbance frequency 11
- size of area 12
- magnitude 14
2. Conceptual Frameworks 14
a. Dynamic Stability 15
b. Recurrence and Persistence 17
c. Theories of Catastrophe and Chaos 18
d. Temporal and Spatial Hierarchies 19
C. Summary 20
IV
-------
II. Data and Methodology 21
A. Targeting the Study 21
B. Data Resources 24
1. Severe Weather 25
2. Tropical Cyclones 30
3. Drought 34
4. National Weather Service Cooperative
Network Data 36
a. Growing Degree Days 41
b. Last Spring Freeze 41
c. Last Spring Snowfall 43
C. Display Technology (CIS) 44
D. Summary 45
III. Example Disturbance Climatologies 46
A. Introduction 46
B. Baseline Conditions (1961-1990) 46
1. Severe Weather and Tropical Cyclones 47
2. Temperature and Precipitation 50
3. Drought 57
4 . Growing Degree Days 58
5. Last Spring Freeze 64
6. Last Spring Snowfall 64
C. Most Recent Decade (1981-1990) 70
1. Severe Weather and Tropical Cyclones 71
2. Combined Stress Analysis 72
3. Summary 78
D. Most Recent Year (1990) 82
1. October 1989 through September 1990 Weather 82
2 . Historical Comparison 83
3. Summary and Comparison to Monitored Data 87
IV- Report Summary and Future Work 90
V - References 93
Appendix A. FUTURE RESEARCH 100
Appendix B. CLIMATE DATA AND RESEARCH BIBLIOGRAPHY FOR
THE NEW ENGLAND STATES 103
-------
List of Figures
1. Developmental pathways of riparian forest patches
in response to frequent and infrequent disturbances
(after: Wissmar and Swanson, 1990) 13
2. Mechanical dynamic stability in terms of potential
energy (Godron and Forman, 1983) 15
3. (a) Disturbance regimes, (b) forest processes,
(c) environmental constraints and (d)vegetation
patterns, viewed in the context of space-time
domains (after: Gosselink, etal.,1990) 20
4. The New England Forest Health Monitoring Program
network of potential sampling sites 23
5. Digitized location of severe weather events, 1961-1990.... 26
6. Intersections of digitized severe weather events
with NEFHM Program sampling hexagons, 1961-1990 28
7. Location of tornado touchdown and path within a
forest sampling hexagon in southwestern New Hampshire,
1981-1990 29
8. North Atlantic cyclone tracks for 1938
(Neumann et al. , 1990) 32
9- The influence of weather related disturbance events
on a mixed coniferous and broadleaf deciduous forest
in New Hampshire (after Woodward, 1987) 33
10. Distance between "nearest" cooperative temperature
observation sites and potential NEFHM sampling hexagons... 38
11. Distance between "nearest" cooperative precipitation
observation sites and potential NEFHM sampling hexagons... 39
12. Relative intra-annual frequency of tornado and damaging
wind events in New England, 1961-1990 47
13. Distribution of tornado intensity for New England,
1961-1990 48
14. Mean annual New England precipitation, 1961-1990 51
15. Mean annual maximum New England temperature, 1961-1990.... 53
16. Mean annual minimum New England temperature, 1961-1990.... 54
vi
-------
34. Percentage of New England region impacted by
climate stress, 1961-1970 78
35. Distribution of hexagon and climate event
intersections, 1961-1970 79
36. Location of hexagons reporting four or more
intersections with climate disturbances, 1961-1970 80
37. 1990 annual maximum temperature departure from
30-year average conditions 84
38. 1990 annual minimum temperature departure from
30-year average conditions 85
39. 1990 annual precipitation departure from 30-year
average conditions 86
40. Potential forest sampling hexagons represented by
precipitation recorded at Squapan Dam, Maine 88
Vlll
-------
17. Area weighted time series of annual New England
precipitation, 1961-1990 55
18. Area weighted time series of annual minimum
New England temperatures, 1961-1990 55
19. Area weighted time series of annual maximum
New England temperatures, 1961-1990 56
20. Area weighted time series of annual New England
temperature range, 1961-1990 56
21. Area weighted time series of mean growing season
PDSI, 1961-1990 59
22. Monthly PDSI values for New England with a 1 in
100 year chance of occurrence, 1895-1990 60
23. Duration of New England drought episode (PDSI with
consecutive months of -2.0 or less) with a 1 in
100 event chance of occurrence, 1895-1990 61
24. Mean annual growing degree days (base = 10°C,
lower limit = 0°C) , 1961-1990 62
25. Area weighted time series of annual growing degree
days (base = 10°C, lower limit = 0°C) , 1961-1990 63
26. Mean date of last spring hard freeze (minimum
temperature of -2.2°C or less) , 1961-1990 65
27. Area weighted time series of last spring hard
freeze event dates (minimum temperature of
-2.2°C or less) , 1961-1990 66
28. Mean date of last spring snowfall, 1961-1990 67
29. Area weighted time series of last spring snowfall
dates, 1961-1990 69
30. Percent of New England region impacted by climate
stress, 1981-1990 72
31. Monthly PDSI values for New England with a 1 in
20 year chance of occurrence, 1895-1990 73
32. Distribution of hexagon and climate event
intersections, 1981-1990 75
33. Location of hexagons reporting four or more
intersections with climate disturbances, 1981-1990 77
VII
-------
List of Tables
1. Natural disturbances and associated ecosystems
present in the northeastern U.S. for which
studies are reviewed in Pickett and White (1985) 7
2. Some correlated disturbance, population and life
history characteristics in plant communities
(from Bazzaz, 1983) 8
3. Examples of climatic variables shown to affect
tree growth (Peer, 1990) 9
4. Proposed climatological variables, data and
ecological responses (Solomon, 1984) 10
5. USDA Forest Service New England Forest Health and
Monitoring Program objectives (USDA, n.d.) 24
6. Examples of climate disturbance (stress) products
for New England (see Chapter III) 45
7. Elements of a New England disturbance regime
climatology for which baseline examples have been
developed 46
8. Tropical cyclones impinging on the New England
states, 1961-1990 49
9. Elements of a New England disturbance regime
climatology for which decadal examples have been
developed 71
10. Elements of a 1990 New England disturbance regime
climatology 82
IX
-------
ACKNOWLEDGEMENTS
The authors wish to thank Keith Eggleston of the New England
Regional Climate Center for providing the 1990 New England
Climatological data, Richard Livingston and Preston Leftwich of the
National Weather Service for providing the digitized storm data
through 1989 and Vince Miller of the Weather Channel for providing
1990 preliminary storm data. Thanks also go to Karl Hermann for
additional GIS assistance and George Mapp for database programming.
-------
I. Background
I.A. Project Overview
The primary goal of this report is to define a working
hypothesis of climate/ecosystem interactions suitable for
operational environmental monitoring programs. Such an exercise is
necessary before climate research and analysis can become an
integral part of these programs. Although the research has
focussed on a specific cooperative EPA and USDA sampling project
initiated during 1990, insights are provided pertinent to the
general topic of climate/biosphere interactions.
The approach adopted is to establish a common ground of
concepts and. terminology to facilitate exchanges between the
climatological and ecological communities. This is accomplished
through the careful definition of key terms and a discussion of the
role of climate in current ecological theory (see section I.E.
below). The discussion provides an avenue for identifying major
environmental policy issues, establishing which of these are
significantly climate related, and initiating work on the
necessary data and analysis products.
The hypothesis that is developed is that the relationship of
climate (the synthesis of weather) to ecosystems depends on the
time and space scale of the biological system to be monitored and
the analyses to be performed. Climatological averages and ranges
reflect one set of physical bounds on the number of possible
ecosystem states represented within a landscape. Inter-annual
climate variability acts as ecosystem disturbances. Disturbance
phenomena impact ecosystem health and productivity as measured by
changes in growth, system energetics, plant populations and species
characteristics. Disturbance events impact the persistence and
recurrence of ecological states. Disturbance events vary with
geographic location and can be described by their frequency, size
and magnitude. Many EMAP and FHM indicators, particularly those
based on directly monitored observations, may not be correctly
interpreted and valid associations drawn without consideration of
mitigating or compounding climatological factors. A climatological
description of background persistence and disturbance regimes is
essential if associative (EMAP) or predictive modeling studies
(FHM) are to include natural as well as anthropogenic forces of
ecosystem change.
This hypothesis is illustrated with data drawn from a
particular region. The focus of this effort is the presentation of
relatively familiar climatological products and ideas in new
ecological contexts. Chapter II describes the cooperative
Environmental Monitoring and Assessment Program (EMAP) and New
England Forest Health Monitoring (NEFHM) programs, presents current
-------
data resources, future data resources, analysis techniques and
display technologies which may be employed. The summaries and
analysis products themselves are presented in Chapter III. A
summary and listing of future research and application
opportunities suggested by this reported are provided in Chapter IV
and Appendix A. Additional climatological references of potential
interest for work in the New England area are presented in Appendix
B.
I.E. Definitions and Conceptual Frameworks for Forest/Climate
Research
I.B.I. Definitions
Because the study of climate/ecosystem interactions is a
rapidly evolving field, terms must be carefully defined and
judiciously used. The definitions that follow have been selected
to establish a common language to facilitate communication between
ecological and climatological communities. As such, some may
appear to be intuitively obvious. Others represent the selection
of one particular use from several possibilities. All provide the
building blocks for research into (1) climate/ecosystem
interactions, (2) the role of climate and climate research in EMAP
and (3) the role of EMAP and ecology in climate change detection,
analysis and impact assessment.
I.B.I.a. Weather and Climate
Weather is the state of the atmosphere, mainly with respect to
its effects upon life and human activities (Huschke 1989) . As
distinguished from climate, weather consists of the short-term
(minutes to months) variations of the atmosphere.
Climate, on the other hand, is the synthesis of the weather.
It is the long-term manifestation of weather (Huschke, 1989) . More
rigorously, the climate of a specified area is represented by the
statistical collective of its weather conditions during a specified
interval of time (usually several decades).
These are important distinctions for climate/forest research.
Although the weather event of interest may take place over the span
of minutes (severe weather) or months (drought), forest ecosystems
may often respond on the order of seasons to decades. Thus, the
appropriate means of analysis is generally climate rather than
weather. But, as implied by the definition of climate, this does
not mean that we are limited to analyses of "average" conditions,
(although for very long-term large area studies this may provide
the most useful information). By definition, climate includes
-------
weather variables that describe the state of the atmosphere over
time. This is done through comparison to itself (normal, deviation
from normal, extremes, variability) or through associations of
weather variables (i.e., a synthesis of weather conditions).
Examples of the latter would include the Palmer Drought Severity
Index or Showalter Stability Index. In each case, a synthetic
variable or measure of atmospheric condition has been constructed
as a function of several coincident atmospheric conditions.
I.B.l.b. Scale
The resolution of ecological and environmental time and space
scales is a critical issue for climate impact assessment research.
Many discussions are available in the literature, but Rosswall et
al. (1988) contains a recent collection of essays that deal
directly with the resolution of scale to address integrated global
change issues.
Mesometeorological events occur between meteorological
stations, or at least well beyond the range of normal observation
from a single point (Huschke, 1989). The types of major weather
phenomena that are small enough to remain undetected within a
normal observation network include tornadoes, thunderstorms and
immature tropical cyclones. This scale also includes events
associated with local geographic influences such as isolated areas
of freezing temperatures or flash flooding. Fujita (1989) includes
events occurring in the range of 50 to 500 km, with duration of 30
min to 5 days and maximum winds of 25 to 60 m s"1.
Synoptic (cyclonic) scale events include the migratory high
and low pressure systems of the lower troposphere, with wave
lengths of 1000 to 2500 km. This scale would also include events
resulting from the presence of these systems such as drought,
freezing precipitation, freezing temperatures, general flooding and
hurricanes. Fujita (1989) includes events ranging from 200 to 2000
km, lasting 1 to 15 days with maximum windspeeds ranging from 10 to
90 m s"1.
Landscapes are heterogeneous land areas composed of a cluster
of interacting ecosystems that are repeated in similar form
throughout (Forman and Godron, 1986). Landscape ecology deals with
large, connected areas and emphasizes spatial patterns and spatial
interactions. Landscape ecology often focuses on studies of
spatial patterns subject to some combination of natural and
anthropogenic disturbances. This contrasts with the term
"ecosystem" which is used to emphasize more localized biotic and
abiotic functional interactions. There is considerable overlap in
the two terms (ecology and landscape ecology) since spatial pattern
and ecological process are inseparable. Both are influenced by the
scale of inquiry. One example of such a complex system is a river.
-------
A river is a complex system of many parts that function together as
an integrated whole. Headwaters, upland slopes, floodplains,
terraces, and river channels are all spatially and structurally
integrated and interrelated. River systems cannot be fully
understood by a study of these individual parts; their
interrelationship is a fundamental property of the system
(Gosselink, etal., 1990). Woodward (1987), IGBP (1990) and others
state that climate (the long-term synthesis of weather) impacts the
distribution, persistence and recurrence of vegetation on the
landscape scale. Given this definition, then, appropriate
climatological products for landscape analysis could include
regional summaries (maps) of climate means, extremes, ranges and
other relevant statistics.
Patches are plant and animal communities surrounded by an area
with a similar community structure or composition (Godron and
Forman, 1983). Patches may or may not have distinct boundaries.
Distinct patch types in natural vegetation are normally produced
through a variety of mechanisms. Spot disturbance patches result
from disturbance of small areas. Environmental resource patches
result from the heterogeneous spatial distribution of environmental
resources. Remnant patches result from disturbance surrounding
small areas. In natural vegetation influenced minimally by human
activities, patches are generally long, narrow, irregular in shape
and few in number. They are caused by a number of natural factors
including fire, insect outbreaks or hurricanes.
Concepts bearing on patch definition and patch dynamics are
summarized in a volume edited by Pickett and White (1985). The
range of essays presented describe dynamics that span a temporal
scale of 10 to 1000 years and spatial scales of 10"* to 106 m2. This
range reflects the variety of biological systems and the variety of
scales within each system on which disturbance effects usually
occur. Shugart et al. (1988) propose that mesoscale climate is
appropriate for gap or patch dynamics analysis and successional
models. The analysis of certain synoptic scale events should be of
value for these activities as well.
I.E.I.e. Disturbance
Disturbance, as defined by Forman and Godron (1987) , is an
event that causes a significant change from the normal pattern in
an ecological system such as an ecosystem or landscape. This is
nearly identical to the definition of ecosystem stress (Barrett,
1981). Climatologists would more than likely choose to call such
an event a perturbation rather than disturbance, but White and
Pickett (1985) contend that the use of perturbation in an
ecological setting is, most often, inappropriate. In general,
perturbation has been used by ecologists with a whole-system
orientation in the sense of any change in a defining characteristic
-------
of a system. The problem with applying this term to natural
systems is that it is difficult to separate perturbation from the
background variance in system parameters. White and Pickett (1985)
suggest limiting the use of perturbation to very narrowly defined
situations. They propose the following definition: A disturbance
is any relatively discrete event in time that disrupts ecosystem,
community or population structure and changes resources, substrate
availability or the physical environment. This definition includes
environmental fluctuations and destructive events, whether or not
these are perceived as "normal" for a particular system. This is
particularly important for relatively rare but large-scale
destructive events such as hurricanes and massive flooding that
might be omitted under the Forman and Godron definition. To
establish a common basis throughout this report the term
disturbance will be used as defined by White and Pickett. This
means that a disturbance can be a climatological perturbation, but
it is not a necessary condition for the event to be of interest.
The remainder of this discussion will focus on developing a more
precise definition of climatological disturbances and their
relationship to ecosystem health and productivity.
To restate an important point, a disturbance is any relatively
discrete event in time that disrupts ecosystem, community, or
population structures and changes resources, substrate availability
or the physical environment. The most obvious role that
disturbance plays in ecosystems is in the deflection of a community
from some otherwise predictable successional path (Pickett and
White, 1985; also see section I.B.2.C.). Models that describe the
successional implications of disturbance are discussed in Shugart
et al. (1988).
Gap models are a subset of a class of forest succession models
called individual-tree models because the models follow the growth
and fate of individual trees. Under optimal growth conditions, the
growth of a tree is assumed to occur at a rate that will produce an
individual of maximum recorded age and diameter. Modifications
reducing this optimal growth are imposed on each tree based on the
availability of light and, depending on the specific model, other
resources. Depending on the specific model, growth may be further
reduced as climate stochastically varies. Shugart et al. (1988)
limit their discussion to mesoscale meteorological phenomena
including tropical cyclones, mesoscale convective complexes, severe
thunderstorms and frozen precipitation. Such events may generate
dramatic changes in the successional framework, but other more
subtle factors such as drought or unfavorably cool or wet growing
season conditions that can impact general forest health and
productivity over an extended period of time are omitted. Michaels
and Hayden (1987) discuss climatological data sources and suggest
ways in which acute climate disturbances can be introduced into
such successional models. Chronic stress has been simulated using
the FORET (a tree growth and development model) by modifying the
growth curves of specific individuals or species by a fixed percent
-------
and following the impact on primary productivity and population
distribution (Auerback, 1981).
Table 1, adapted from White and Pickett, lists some natural
disturbances and the ecosystem in which literature has been
reviewed that fit this broader definition of disturbance. Only
weather related disturbances and terrestrial ecosystems represented
in the Northeastern U.S. have been included. Note that chronic as
well as acute climatological disturbances have been included.
Table 2, adapted from Bazzaz (1983) contains a summary of
correlated disturbance, population, and life history
characteristics in plant communities. Bazzaz (1983) describes in
some detail the role of disturbance in the determination of
population characteristics. Reiners (1983) presents a similar
discussion for ecosystem energetics. This includes measures of
ecosystem productivity such as above and below ground biomass as
well as other soil chemistry measurements. Similar studies for
aquatic as well as terrestrial ecosystems are summarized in Barrett
and Rosenberg (1981). Pickett and White (1985) summarize a number
of essays concerning natural disturbance and patch dynamics by
saying that disturbance has demonstrated effects on community
characteristics, including richness, dominance, and structure. The
functional attributes of ecosystems are also governed to some
extent by disturbance. Nutrient cycling and energetics respond to
disturbance as well as to biotic and abiotic opportunities and
limiting factors.
Table 3, adapted from Peer (1990), contains a summary of the
role of climate in recent forest growth models. Table 4, taken
from Solomon et al. (1984), contains relationships between specific
climatological variables (means, frequencies, extremes) and various
forest response indicators. This kind of information, in addition
to that presented in Table 2 and discussed in Reiners (1983) is
particularly critical if disturbance climatologies are to be
related to monitoring programs such as EMAP and FHM. Proper
interpretation of field samples lead to more accurate
identification of trends and potential sources of ecosystem stress
for policy analysis and for detection of local manifestations of
global environmental change (e.g., greenhouse warming). Although
model development is not a primary goal of this study, climate
analyses that directly support the integrated assessment of
ecosystem status and trends can lead to a more complete
understanding of forest/climate relationships and to the
development of more accurate models of natural and anthropogenic
influences on forest dynamics.
-------
Table l. Natural disturbances and associated ecosystems present in
the northeastern U.S. for which studies are reviewed in
Pickett and White (1985).
Disturbance
Fire
Hurricane
Other windstorms
Ice storm
Ice Push on shores
Freeze damage
Fluctuating water levels in
basins
Droughts
Alluvial processes
Coastal processes
Salinity changes
Insect outbreaks
Disease
Ecosystem of geographical area
Boreal forest
Temperate forest
Coastal plain
Terrestrial
Temperate forest
Temperate forest
Temperate and Boreal forests
Various
Various
Temperate forest
Various
Various
Various
Various
Deciduous forest
-------
Table 2. Some correlated disturbance, population, and life history
characteristics in plant communities (from Bazzaz, 1983).
A. Disturbance characteristics with relevance to plant response
Size
Intensity
Frequency, regularity and predictability of occurrence
Duration
Seasonal time of occurrence
Level of environmental heterogeneity within the disturbed
area
Nature of the biotic neighborhood
B. Population characteristics responsive to disturbance
Density and dispersion
Growth rate
Survivorship; age and size structure
Levels of gene flow in the population
Degree of relatedness among the members of the population
Organization of variation within the population
Strength of competitive interactions
Niche breadth and niche overlap
Strength of interactions with other trophic levels
Plant life history characteristics responsive to disturbance
Spatial and temporal dispersal ability
Seed germination
Seedling establishment and growth
Reproductive strategies
Breeding system
Fecundity
Reproductive allocation and packaging
8
-------
Table 3. Examples of climatic variables shown to affect tree growth (Peer, 1990)
CLIMATIC
VARIABLES
Fall precipitation =>1.3cm
Cold falls and winters
July precipitation
High June precipitation
Low fall rainfall
Intense summer precipitation
Cumulative air temp, degree days
Total precipitation
Monthly average temperature
Total rainfall
June temp, previous year
November temp, previous year
January temp, current year
July precip. previous year
March precip. current year
June precip. current year
July precip. current year
34 variables based on monthly
temperature and precipitation
SPECIES AND LOCATION
conifers (8 species)
Minnesota, Wisconsin, and
Michigan
hardwoods (14 species) same
locations as above
red maple (Acer rubrum) ,
northern red oak (Quercus rubra) ,
paper birch (Betula papyrifera) ,
all in northern Michigan
red spruce (Smoky Mountains)
hemlock (Pennsylvania)
European silver fir
(Bad Herenalb, FOR)
DEPENDENT
VARIABLE
diameter at
breast
height (dbh)
diameter
growth
tree ring
width
tree ring
width
tree ring
width
COMMENTS
Corr. coef. did not
exceed .42; Species varied
in climate sensitivity;
generally conifers more
sensitive than hardwoods
Proportion of variance
explained by regression
ranged from .44 to .65
Demonstrated that climate
sensitivity varied over
time
R-square=.64; hemlock is a
climatically sensitive
species
Method suitable for
dealing with both change
and sudden shocks
-------
Table 4. Proposed climatological variables, data and ecological reponses (Solomon, 1984).
Affedted organisms,
life-stage, or environmental
varieties
Primary
responses
Secondary
responses
Climate variables
Tenperature variables
Frequency, intensity of winter low
temperature
Frequency, intensity of spring frosts
Cumulative grown)ng season warmth
Frequency, intensity, length of heat
waves in midsummer
Frequency, intensity of fall frosts
Precipitation variables
Frequency, length of flooding
Frequency of spring soaking rains
Growing season precipitation as days
below wilting point of soil moisture
Frequency, intensity of growin season
droughts
Instense wind storms, ice storms
Saplings
Trees
Spring-terminating seeds
Trees
Saplings, trees
Soil biota
Seedlings
Fire frequency
Fall-germinaling seeds
Seedlings
Saplings, trees
Seedlings
Seedlings
Saplings, trees
Soil biota
Seedlings
Mature trees
Fire frequency
All trees, largest
trees, shallow-rooted
trees
X Death of saplings
% Loss of buds, branches
X Death, by species
% Defoliation
X Growth of tree, by species
X Nutrient release rates
X Death, by species
X Death, damage, reproduction by
life-stage, and species
X Death, by species
X Death
X Death, slowed growth, by species
X Death from soil bioat, by species
X Death if too little/too much
by species
X Growth of tree, by species
X Nutrient release rates
X Death by species
X Slowed, stopped growth,
by species
X Death, damage; reproduction by
life-stage and species
X Death, damage; enhanced
reproduction by species
Loss of reproduction, tree damage
X Growth of tree, by species
X Increased disease, mortality
X Growth of tree, by species
Loss and gain of pathogen
population
X Increased disease; death by species
X Increased disease, death by species
X Growth of trees, by species
X Increased disease, death by species;
Loss and gain of pathogen population
X Increased disease, mortality
10
-------
Wissmar and Swanson (1990) suggest that ecotones may be
particularly sensitive to disturbance regime. Ecotones are zones
of transition between adjacent ecological systems having
characteristics uniquely defined by space and time scales and by
the strength of interactions between systems (Holland, 1988). They
are considered to be sensitive to gradients of limiting factors and
landscape changes caused by physical and biological disturbances.
A basic scientific problem in the evaluation of landscapes and
their ecotones is the lack of testable models with short- and long-
term predictive capabilities (Wissmar and Swanson, 1990). The
difficulty concerns not knowing the extent to which characteristic
or repetitive changes in ecotones and ecosystems of a landscape are
caused by disturbances of low frequency and high magnitude
extremes. We understand that large disturbance events can dominate
the main trends of change. Yet we do not know which events will be
formative, or how to recognize temporal and spatial sequences of
events and the ability of ecotones and ecosystems to recover to
characteristic persistent states. Wissmar and Swanson (1990)
present information about representative disturbance recurrences
and recovery phases for disturbed hillsides and rivers, but such
landscape knowledge for both physical and biotic components of
ecotones and ecosystems is generally lacking.
White and Pickett (1985) summarize disturbance characteristics
which can be used as a guide to the development of disturbance
climatologies. Key descriptors include spatial distribution,
frequency, area and magnitude. These attributes are in fundamental
agreement with descriptions contained in Reiners (1983) and Bazzaz
(1983) .
Spatial distribution includes the relationship of the
disturbance to geographic, topographic, environmental and
ecological community gradients. Table 1 suggests some guidelines
for the geographic distribution of disturbance events. The list
should be considered only an initial starting point. Summaries of
other disturbance studies for specific geographic areas or
applications include Peer (1990) and Solomon et al. (1984).
Disturbance frequency is the mean number of events per time
period. Frequency is often used for probability of disturbance
when expressed as a decimal fraction of events per year.
Disturbance frequency is of particular interest to climatologists
as well as ecologists. Overpeck et al. (1990) report General
Circulation Model (GCM) results that suggest changes in disturbance
frequency resulting from greenhouse warming. The authors used a
mixed-species, mixed-age stochastic stand-simulation model to
estimate the impact on forest biomass of these changes. Given the
uncertainty in GCM scenario results, an alternative approach is to
explore ways in which changes in forest health and productivity can
suggest the existence of trends in disturbance event frequency.
Reiners (1983) suggests measures of above, as well as below, ground
11
-------
biomass as an indicator of changes in disturbance frequency. He
suggests that as disturbance frequency goes up, biomass and
associated detritus production will decrease. This is reflected in
the model generated results of Overpeck et al. (1990) . Bazzaz
(1983) states that frequency of disturbance may regulate age
structure of the population, life cycle and reproductive
strategies.
Reiners (1983) also discusses the "kind" or "uniqueness" of
the disturbance. Species composing an ecosystem must have, to
varying degrees, adaptations to normally occurring disturbances.
The more frequent the disturbance, the stronger the representation
of better adapted species and theoretically, the stronger the
selection pressure for adaptive traits. Thus the impact of a
unique disturbance is likely to be more profound than that of a
"normal" disturbance. Similar distinctions are emphasized by White
and Pickett (1985) in their review of the terms "disaster" and
"catastrophe." A "disaster" occurs so frequently that it is likely
to occur within the life cycles of successive generations, while a
"catastrophe" occurs rarely, so that it is unlikely to be
experienced as a repeated selective force. A disaster would be
likely to increase fitness through selection, while a catastrophe
would generally decrease fitness. Figure 1 illustrates differences
between developmental pathways of frequently and infrequently
disturbed riparian forest patches. Pathway 1 includes stand
initiation and exclusion stages during reactive and recovery phases
in ecotones following disturbances. Pathway 2 includes stand
exclusion, understory reinitiation, and old growth stages after the
recovery and during the persistence phases.
Size of area disturbed can be expressed as area per event,
area per time period, area per event per time period, or total area
per disturbance type(s) per time period. It is frequently given as
a percentage of total available area. Disturbance size ranges from
very small gaps created by the death of an individual in an
herbaceous community, to the breakage of a tree limb to the fall of
a single tree creating a canopy gap, to the fall of several trees,
to very large disturbances created by fires and windstorms. Bazzaz
(1983) states that disturbance size and heterogeneity interact in
determining size and identity of plant populations. Reiners (1983)
states that areal extent impacts system energetics through
environmental modification and re-colonization potential. A larger
disturbed area will, within limits, create more extreme
microclimates for regenerative processes. Depending on the form of
ecosystem disruption generated by the disturbance, a larger
catchment for runoff can result, leading to a higher potential for
erosion on upland sites and deposition in lowland sites. A larger
disturbed area, particularly as it takes a round and fully
contiguous shape, creates a longer dispersal distance for
recolonizers, including both the spores and seeds of plants and
vectors such as specific and critical pollinators.
12
-------
Figure 1. Developmental pathways of riparian forest patches in
response to frequent and infrequent disturbances. The
letter 't1 indicates different transition times for
riparian patches (successional stages)
(Wissmar and Swanson, 1990).
PATHWAY # 1
(Response to frequent
disturbances)
Vegetation remains In Initiation
and exclusion stages
PATHWAY # 2
(Responses to Infrequent
disturbances)
RECOVERY PHASE
Vegetation patches remain In
Initiation and exclusion stages
REACTION PHASE
\
Vegetation patches develop via
exelusion,understory reinitiation
and old growth stages A
PERSISTENCE PHASE
13
-------
Magnitude of disturbance can be characterized by intensity and
severity. Scaling intensity is relatively simple within a
disturbance type and for a particular ecosystem. The Palmer
Drought Severity Index for drought and the Fujita scale for
tornadoes are examples of scaled disturbance intensities. Reiners
(1983) points out, however, that at present we have no means of
making comparisons of intensity between disturbance types, or of
predicting responses of different ecosystems to the same intensity
of a particular disturbance. Intensity responses may be non-
linear, not monotonic and threshold behavior, particularly at the
ecosystem level, is likely. These thresholds may be critical
points at which recovery rates are significantly slowed, or more
importantly, at which recovery to the original state is impossible.
Defoliation either by pests or weather events is a case in point
(Stephens, 1981; Johnson, et al., 1988).
I.E.2. Conceptual Frameworks
The importance of climate to the health, productivity and even
the existence of particular ecosystems has long been recognized
(Woodward, 1987) . But the field of ecology and, along with it, the
study of climate and ecosystems, has evolved rapidly during the
last two decades. Prior to this, most ecological research focused
on the observational characterization of ecological patterns.
Research consisted primarily of qualitative descriptions with few
theories based on physical or mathematical principles. In recent
years, interdisciplinary concepts of ecosystem behavior and
dynamics have gained prominence. Quantitative models have been
developed for relationships that had previously been expressed only
in qualitative terms.
For the applied climatologist, an introduction to these
theoretical "frames of reference" serve two functions. First, it
emphasizes the use of theory and mathematical techniques that are
also found in predictive climate and meteorological research.
Markov chains and Chaos theory are two such examples. Second, it
illustrates the limited role of climate and weather in previous
ecological research. This is as much the responsibility of
climatologists as ecologists. Ecologists performing climatological
analyses introduce new perspectives concerning climate/biosphere
interactions, but to take advantage of these contributions, the
applied climatologist must gain an appreciation of dynamic bio-
centric as well as more familiar atmosphere or man-centered
systems.
From the ecological side, the selected frameworks will be
familiar, but the role of climate in these settings is highlighted.
In some cases, climate interactions have always been present
implicitly, but rarely expressed explicitly. For instance, under
the dynamic stability framework, the role of energy dynamics is
14
-------
discussed at length, but the influence of climate and climate
disturbances on the storage or release of ecosystem energy is less
apparent.
The discussion that follows highlights three conceptual
frameworks proposed in the recent ecological literature. These
approaches include dynamic stability, Markov chain analysis and
catastrophe theory. The selected frameworks are presented both for
their range of disciplinary appeal, but also for their similar
portrayal of climatological influences.
I.B.I.a. Dynamic Stability
The first approach appears frequently in the literature of the
late 1970's and early 1980's. It borrows prominently from physics,
specifically the field of mechanics. A representative example is
illustrated in Figure 2 adapted from Godron and Forman (1983).
Living organisms often start from highly stable states represented
as state B in Figure 2. Organisms build locally metastable
ecosystems represented by points C, D or E. At these points, a
system will return to equilibrium after small (but not large)
displacements. The greater the energy required to displace the
system from equilibrium, the more metastable that system is said to
be.
Figure 2. Mechanical dynamic stability in terms of potential
energy (Godron and Forman, 1983).
o
tr
UJ
z
UJ
o
Q.
Most
Metastable
Least
Metostable
'Most Stable
15
-------
Landscapes represent an interesting special case for this
framework. As defined earlier, a landscape is made up of many
smaller patches. Borman and Likens (1979) develop an example of
this framework applied to forest clear-cuts. The authors example
begins at a highly metastable point located to the left of point A
of Figure 2. This point represent a system with high biomass
(potential energy). Energy, in the form of some disturbance event,
clear-cutting in the Borman and Likens example although fire or
hurricane would work in a similar fashion, is input to the system
(point A) and the biomass is released until point B is reached.
Disturbance and increased decomposition promote an increase in
available water and nutrients. As time passes, living biomass
again accumulates and the system progresses through metastable
points C, D and E. Based on output of the forest growth model
JABOWA (Botkin, et al., 1972), Borman and Likens (1979) suggest
that dynamic equilibrium, or shifting mosaic steady state is
achieved when the total biomass across all patches begins to
oscillate about a mean. When a dynamic equilibrium is reached, the
proportions of landscape patches in states B, C, D or E remain more
or less constant with time, but the state of any individual plot
may change as it gains or looses biomass (points L, M, N or Z) .
For example, new patches in state B arise when gaps are created by
the fall of large trees. This results in a "blip" in the level of
biomass which is later compensated for by biomass gains (increased
metastability) in other patches.
Under a dynamic equilibrium framework, climate plays two
roles. First, climate means and long-term patterns of variability
impose upper limits to the accumulation of potential energy in the
ecosystem. Second, climate can act to move systems from metastable
states to higher or lower states. For example, metastability can
be increased by directly supplying or facilitating the storage of
additional energy (e.g. Photosynthetically Active Radiation (PAR),
moisture, conditions favorable for nutrient fixation).
Metastability could be decreased by bringing about the release of
stored energy through, the addition of mechanical energy (wind,
hurricane), heat energy (lightening and fire) or direct removal of
potential energy (e.g., reduced available moisture resulting from
drought), thus triggering decomposition and the loss of potential
energy from the system. The importance of this framework to
ecosystem status and health monitoring is that these climate
related energy exchanges take place continually and are part of
normal ecosystem behavior. Changes in these relationships or
patterns of climate related energy accumulation and release may
indicate changing forest status, health or define the initiation of
forest or climate trends. More detailed discussions of ecosystem
energetics including climatological aspects are provided in Reiners
(1983) and several papers in Zonneveld and Forman (1990).
16
-------
I.B.2.b. Recurrence and Persistence
This conceptual framework is closely related to principles of
Markov processes. The Markov chain is a special case in which a
finite (or countably infinite) number of outcomes or states are
possible. A first order Markov chain is one in which each state
depends only on the immediately preceding state.
Models using Markov techniques are useful as a first
approximation to changes from one ecological state to another when
only the probabilities of the transitions can be estimated
(Jeffers, 1988). Within major states, substates can also be
modelled as a Markov chain. Markov models (and related approaches)
have been used successfully as landscape modeling paradigms in both
applied and basic contexts. Markov models of forest succession are
mathematically and conceptually the most straight forward of the
succession models that are presently in use (Shugart et al. , 1988).
This approach is also found in more complex "gap" models. Several
examples of the use of Markov process models for successional
modeling are mentioned in Weinstein and Shugart (1983).
Early stand dynamics models held climate constant using only
average conditions. Climate impacted forest composition only
through growth response to temperature and competition for light
and moisture. Movement through the successional chain was driven
by forest-gap initiation and overturn through old-age mortality -
Later research recognized that changes in disturbance or
climatological regime short-circuits these slower processes and
modifies the rate of successional change. Overpeck et al. (1990)
tested the sensitivity of a widely used stochastic stand-simulation
model (FORENA, Shugart, 1984) to changes in disturbance frequency.
The practical impact of climate disturbance in these simulations
was to reset the Markov chain to earlier successional stages. More
frequent disturbance resulted in greater dominance by early
successional species.
As yet, Markov chains have not been employed to simulate
temporal patterns of large scale disturbance, but research that
explores the use of Markov chains to predict large-scale
atmospheric patterns is underway (S. LeDuc, personal
communication). The need to relate such large-scale circulations
to models of forest ecosystem dynamics is discussed in Michaels and
Hayden (1987).
Botkin (1980) uses the concept of Markov chains to address
some highly general ecological concepts. He suggests that ecosystem
behavior through time can usually be described by two
characteristics. First, particular ecosystem states come and go, or
recur. Second, the recurrence of ecosystem states depends on the
total amount of variation (the total number of possible states)
which, in turn, must be bounded. This suggests the concepts of
ecosystem persistence within bounds and the recurrence of specific
17
-------
ecosystem states.
Under this framework, long-term patterns of climatological
means and variability can influence the establishment and
persistence of dominant ecological landscapes, i.e., climate
features set the bounds for the state space (Woodward, 1987).
Conversely, climate is dynamic, with weather conditions that can
vary widely from one year to the next. Climate can, therefore,
affect the range of variation (the number of possible states), the
rate of recurrence of particular states and the average time that
the ecosystem is in any of its possible states.
I.B.2.C. Theories of Catastrophe and Chaos
Catastrophe theory has recently attracted considerable
attention in studies of non-linear mathematical dynamics and chaos
theory- Chaos theory has shown that regular equations can produce
irregular behavior. Minor variations in the system's initial state
cause behavior that is, in effect, unpredictable even though
overall system change is predictable. It is strictly determinate
in a mathematical sense. Catastrophe theory suggests that a
gradually changing system (with its associated characteristics)
converges on and crosses particular points (chaotic attractors).
Only a slight change in the immediate vicinity of a point will
divert the system in a quite different direction. A change in
system non-linearity acts in a fashion similar to changes in
initial system state. Not only might the final equilibrium
condition be altered, but the ability of the system to converge to
equilibrium may be eliminated (Gleick, 1987) . One example which
could lead to a system of equations that contain changing non-
linearities is plant growth and development under variable moisture
stress.
Forman and Godron (1986) propose that major alterations in
landscape development take place in this way. An abrupt change in
the distribution of climatological events is one factor that, alone
or in combination with other factors, can push an ecosystem beyond
its critical threshold point. For instance, tree species that
normally tolerate degraded air quality conditions may experience a
precipitous decline when frequency and intensity of severe winter
conditions change. The determination of the short or long term
nature of such interactions would influence the interpretation of
present status and monitored trends in forest health (Johnson et
al., 1988) .
18
-------
I.B.2.d Temporal and Spatial Hierarchies
To fully understand the role of climate (changing climate) in
an ecological context, its place in a bio-centric temporal and
spatial hierarchy must be considered. In the context of the
present study, compatibility of spatial scale is fairly
straightforward, but temporal scale must be approached more
cautiously. Although the disturbance event (e.g., windthrow) may
take place in the space of a few minutes, it is the response of the
biological system to this disturbance that is of interest. For
example, even though a tree is partially uprooted, it may take
several seasons to die. Changes in crown density and nutrients may
respond on the order of weeks to months to prolonged drought, but
normal growth and reproduction may not be observed for one or more
growing seasons even with a return to more moderate temperature and
moisture regimes. This represents a radical departure from more
traditional mono-crop agriculture applications common in the
climate literature that most often deal with single or short-term
events occurring within a particular growing season. With the
exception of extremely rapid-growing species such as loblolly pine,
forest/climate impacts span periods from one to ten years and
frequently include multiple, interacting disturbance events.
Figure 3 relates time and space distributions of disturbance
regimes, forest processes, environmental constraints and vegetation
patterns. In the case of disturbance regime, spatial scales fall
neatly into the mesoscale and synoptic scale definitions provided
in Huschke (1989). The temporal scale represents the return period
of the disturbance event. As an example, one interpretation of
Figure 3 is that a windthrow event with a return period of 1 in 100
to 1 in 1000 years and impacting an area on the order of 1000 m2 can
be expected to affect tree replacement processes within topographic
constraints and may manifest itself in changes in gap (canopy
disruption) and stand or serai stage vegetation pattern.
Climate change events, typified in Figure 3 by glacial cycles,
is defined as spanning more than 10000 years and impacting more and
10000 km2. From Figure 3, affected processes include species
extinction and impacts would be noted in vegetation patterns of
ecological provinces and biomes. If, as some studies suggest,
climate change takes place on smaller spatial scales (i.e., 105 m2)
and over a shorter time (i.e., 102 years), new impacts and
constraints can be quickly noted. For instance, under glacial
change, the importance of disturbance regime all but disappears.
Disturbance reemerges as an important element under a scenario of
rapid environmental change. Secondary succession and species
migration processes emerge as new impact areas. Soil properties
(pedogenesis) and fluxes of moisture and energy become limiting
constraints. Changes in stand and cover patterns may become
evident.
19
-------
Figure 3. (a) Disturbance regimes, (b) forest processes, (c)
environmental constraints and (d) vegetation patterns, viewed in
the context of space-time domains (Gosselink, et al., 1990).
a -
SPATIAL SCALE (lofl m'l
C J
UICRO-
CNVMONS7
b -
10 13
CO.,.1)
SCALE (lo« m1)
I.e. Summary - A Working Hypothesis of Climate/Ecosystem
Interactions
The relationship of climate (the synthesis of weather) to
ecosystems depends on the time and space scale of the biological
system to be monitored and the analyses to be performed.
Climatological averages and ranges reflect one set of physical
bounds on the number of possible ecosystem states represented
within a landscape. Inter-annual climate variability acts as
ecosystem disturbances. Disturbance phenomena impact ecosystem
health and productivity as measured by changes in growth, system
energetics, plant populations and species characteristics.
Disturbance events impact the persistence and recurrence of
ecological states. Disturbance events vary with geographic
location and can be described by their frequency, size and
magnitude. Many EMAP and FHM indicators, particularly those based
on directly monitored observations, may not be correctly
interpreted and valid associations drawn without consideration of
mitigating or compounding Climatological factors. A climatological
description of background persistence and disturbance regimes is
essential if associative (EMAP) or predictive (FHM) modeling
20
-------
studies are to include natural as well as anthropogenic forces of
ecosystem change.
The remainder of this report applies this hypothesis to the
New England region of the U.S. Data availability, analysis and
display options are discussed in section II. Climate analyses are
performed to address three levels of inquiry: large-scale
climatologies that establish system boundaries (e.g. means and
extremes); climatologies of shorter duration and smaller scale
disturbances that are characterized by frequency, size and
magnitude; and climate summaries that permit comparisons of recent
regional and local conditions to pre-determined baseline or
historical conditions.
II. Data and Methodology
II.A. Targeting the Study
The vast array of possible climate/ecosystem interactions that
could be examined is substantially narrowed by "targeting" the
analysis to a particular setting. In this case, the cooperative US
EPA/EMAP-Forests pilot and USDA Forest Service New England Forest
Health Monitoring (NEFHM) project has been selected.
EMAP is a long-range program to monitor status and trends in
the condition of the major ecological resources of the U.S. A
discussion of the sampling design and landscape characterization
under EMAP is presented in Norton and Slonecker (1990) . Six broad
ecological categories have been defined within EMAP, one of which
is Forests. The use of indicators to measure ecological status and
trends for forests as well as the other resource categories is
discussed in Hunsaker and Carpenter (1990). An indicator is
defined to be a characteristic of the environment that, when
measured, quantifies habitat characteristics, the magnitude of
stress, degree of exposure to the stressor, or degree of ecological
response to the exposure. Under this definition, for example,
drought is a climate stressor, but Palmer Drought Severity Index
quantifies the magnitude of stress and so is a stress indicator.
Climate as a whole could also be considered a landscape indicator.
A landscape indicator is defined as a characteristic of the
environment used to describe spatial distribution of physical,
biological and cultural features across a geographic area.
Although all examples of stress and stressors given in Hunsaker and
Carpenter (1990) are negative, a stress does not necessarily
correspond to a decline in forest status or health. The authors'
definition of stress indicator suggests only "changes in exposure
and habitat." This conincides with literature definitions of
ecological stress which closely resemble our earlier definition of
disturbance (Barrett, 1981). Throughout the remainder of this
21
-------
report, disturbance and stress will be used interchangeably with no
value judgment as to the detrimental or beneficial nature of the
resultant ecosystem response.
The U.S.D.A. Forest Service (USDA-FS) has designed a national
system to annually monitor the health of the Nation's forests as an
integral part of the USDA-FS Global Change Program proposed under
the Forest Ecosystems and Atmospheric Pollution Research Act of
1988. The program involves coordination with the USEPA's EMAP.
This new program, Forest Health Monitoring (FHM) is a multi-tiered,
long-term process to 1) detect unexpected deviation from
established baseline conditions or trends, 2) identify causes of
change, and 3) define basic relationships sufficiently to predict
consequences. Detection, evaluation and research monitoring levels
are planned. Implementation of the FHM program began during the
summer of 1990 in New England (NEFHM).
The initial year of the cooperative NEFHM/EMAP-forest project
consisted of 263 nearly equally spaced sampling areas, each being
equivalent to an EMAP sampling hexagon. EMAP plots are 40.6 km2
hexagons with a density of a one-sixteenth sample of total area.
These points represent potentially forested locations. Some
hexagons fell in urbanized areas so that the final sampling network
consisted of 206 forested hexagons distributed throughout
Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island and
Vermont (Figure 4). As time goes on, present forested sites may
become urbanized or previously inaccessible sites may become
available. Although 206 locations were sampled during 1990, all
263 sites remain available for future sampling seasons. Specific
NEFHM objectives are provided in Table 5. Lists of the variables
selected for observation are provided in USDA Forest Service (no
date). The initial sampling project was completed during the summer
months of 1990. A summary of this collection season is provided in
Brooks et al. (a) (in press).
22
-------
Pigur
e 4- The
Program
network
O 0 O O O
O O O O O (
O O O O O O
O O O O O O <
0 P O • 0 O
O O O 0 0 O t
oooooooco
000000300
oooooooooo
> o o o o o o o o o
O O O O 0 O
23
-------
Table 5. USDA Forest Service New England Forest Health and
Monitoring Program objectives (USDA, no date).
To characterize the following forest conditions:
a. growth rates
b. tree vigor
c. soil/site
d. stand composition
e. landscape
To characterize the following potential forest stressors:
a. insect and disease pests
b. climate (long-term)/weather (short-term)
c. atmospheric pollution/deposition
d. other (direct anthropogenic activities, e.g
harvesting)
To quantify changes in forest conditions, and
To correlate changes in forest condition with potential
forest stresses.
II.B. Data Resources
Section I has suggested general frameworks for describing
climate features relevant to the analysis of forest landscapes and
patch dynamics. The importance of climate related disturbance
regime to these frameworks is clear. For example, disturbance
related mortality impacts tree population structure, community
structure, the release of light, nutrient and moisture resources,
resources stored by decomposers and the creation of new resources
such as complex organic compounds and decomposer habitat (Franklin
et al., 1987). Climate and forest scientists agree that changes in
disturbance regime can result in major ecosystem impacts and could
be an early indicator of greenhouse warming (Overpeck et al., 1990;
Michaels and Hayden, 1987; Graham et al., 1990). To be able to
more fully understand forest/disturbance interactions and to detect
changes in these relationships, data must be assembled and baseline
climatologies constructed.
Data sets bearing on several major categories of ecosystem
disturbance have been assembled. Methodologies are explored to
illustrate the relevance of specific climate features to forest
status and health assessment. The pertinent climatological data
sets are outlined below. Examples of their use in selected climate
analysis products will be presented in section III.
24
-------
II.B.I. Severe Weather (High Winds and Tornadoes)
Windthrow, the toppling or uprooting of trees by wind, can be
a frequent disturbance phenomenon in northern and New England
forests. Catastrophic windthrow in presettlement Wisconsin is
discussed in Canham and Loucks (1984). They conclude that the
majority of windthrow damage conforms to downburst phenomena such
as those described in Fujita (1978). Bormann and Likens (1979)
take a larger spatial view and conclude that the importance of
such events varies with geographic location, but windthrow is a
frequent disturbance throughout the Northern Hardwood Forest area
of North America. Ferris (1980) reports that windstorms are not
uncommon in Maine and mature spruce-fir stands are particularly
susceptible to windthrow because they have shallow root systems.
Henry and Swan (1974) summarize an analysis of the affects of
windthrow on an experimental forest in southwestern New Hampshire.
Tornado and damaging wind data bases were obtained from the
National Weather Service's (NWS) National Severe Storms Forecast
Center (NSSFC) in Kansas City, Missouri. They contain digitized
records from 1950 (for tornadoes) or 1955 (for wind) through 1989
and have been updated for the New England states through September
of 1990 using preliminary weather service reports. Similar records
have been acquired from Environment Canada, Atmospheric Environment
Service to supplement data-sparse portions of northern New England.
Michaels and Hayden (1987) suggest that frequency, strength
and area affected are important characteristics of severe storms
for the estimation of atmospheric effects on ecosystems. These
correspond with disturbance frequency, magnitude and size
characteristics recommended by Pickett and White (1985). A number
of descriptors are available from the NSSFC database; only a few
are selected for analysis. In the case of tornadoes these include
location of touchdown, direction, length and width of path and
strength as described by the Fujita scale given below (Fujita,
1981).
Funita Tornado Classification categories
F-scale maximum estimated wind velocity
(m/s)
F-0 18 - 32
F-l 33 - 50
F-2 51 - 70
F-3 71 - 92
F-4 93 - 116
F-5 117 - 142
Although these may be critical characteristics, time-series
taken from the best digitized weather records available must be
used with care. Reporting and verification criteria vary throughout
the record. The tendency of reported events to converge on
population centers is also easily noted in a 30-year summary map of
severe weather events (Figure 5).
25
-------
Figure 5. Digitized location of severe weather events, 1961-1990
A WIND EVENT
o ter t
knots)
(jjrep ter than
TORNADO EVENT
Spot location
(Tornodo trockj
ore indicated
•i th a line)
26
-------
Because of these limitations, time series of tornado
characteristics are not presented in any of the climate summaries.
Particularly when detection of trends and changes in disturbance
regimes are priorities, observational biases in this historical
database are too great. Analyses of other characteristics such as
tornado strength (magnitude) and area affected (size) may also
suffer from similar biases, but, if it is assumed that the storms
randomly hit populated areas, a pooled sample of these population-
centered storms should still provide reasonably reliable general
information concerning the characteristics of tornadic events in
New England. Aggregate summaries of monthly distribution(intra-
annual frequency), F-scale (intensity) and extent of affected
surface area (size) are presented in the climate summaries (section
III) .
Reporting errors aside, other problems arise when attempting
to relate point observations of storm damage to an operational
monitoring program such as EMAP or FHM. Figure 5 displays the
location of reported severe weather events. Figure 6, generated
with a GIS, represents the intersection (co-location of latitude
and longitude coordinates) of New England tornado events of the
most recent decade and the NEFHM sampling hexagons. First,
population bias is again evident. Second, the impression from the
sampling network alone is that tornadic events are far less
frequent than the weather database suggests. That is, the
likelihood of noting the effects of these events at a forest
sampling point is quite small. This latter point is even more
evident in Figure 7. Figure 7 is a representation of a specific
hexagon in southwestern New Hampshire in which tornado events were
reported. The dashed lines surrounding the tornado locations
represent imprecision in the coordinate locations of the digitized
weather record. The large open circle represents the forest
sampling location within the hexagon.
The implications of these comparisons are twofold. First,
tornadoes and other severe storm events may best be summarized in
an aggregate, landscape, manner. Knowledge of the individual
events will be essential to future forest model development and
interpretation of site specific health indicators (e.g., mortality
and crown condition) , but will likely not be useful for the
detection of overall resource condition and trends. Second,
remotely sensed data from Doppler radar and space based earth
observing systems should play a vital role in establishing accurate
regional summaries. For example, by the turn of the century it is
anticipated that instruments on the Earth Observing System will
provide global coverage; surface winds are expected to be measured
at 100 km resolution with errors less than 5 ms"1 (Baker, 1991) .
Although this still represents a fairly high degree of error as
compared to in situ instrumentation, it will fill large geographic
"gaps" in our present databases.
27
-------
Figure 6. Intersections of digitized severe weather events with
NEFHM Program sampling hexagons, 1961-1990.
00*0
o o o o o o
o o o o o
o o o o o o
o o • o o
o o o o o o o
OOOOOOOO
oooooooo
oooo»ooooo
• o o * o / •
oo »ooo/o oo
o o o o o
o o o •
• o • • > o o o
o o o /* o o
o o I o o o • *
k O • •
« • k O • IO
">T - __^/-«
-7 ^ V*
28
-------
Figure 7. Location of tornado touchdown and path within a forest
sampling hexagon in southwestern New Hampshire, 1981-
1990.
hexagon area = 40 km
hexagon diameter = 8 km
• wind event
— tornado track
o FHU samp It site
Direction and d i stance
to nearest coop site
2
29
-------
Wind reports contained in the NSSFC database are events that
meet certain criteria. There must be verified peak gusts in excess
of 26 ms"1 (50 knots) for inclusion in the database. On occasion,
no windspeed measurements are available and gusts are estimated
from damage surveys. In these cases it can be determined that
windspeeds in excess of 26 ms"1 occurred, but the exact value of the
peak cannot be established.
Because of reporting uncertainties similar to those of the
tornado database, no time series summaries have been prepared. A
seasonal frequency distribution (intra-annual frequency) of events
is presented for wind. The data do not support a wind speed
magnitude analysis. Area impacted (disturbance size) information
are not available in digitized form.
Additional events of interest are reported in the serial
publication, Storm data (USDOC, 1991). The digitized severe
weather data that has been analyzed here is abstracted from these
publications. Storm Data may be of limited "real time" help to the
FHM program since its publication schedule runs six or more months
behind the event. However, it is useful for information about the
previous season. For instance, data are available in October 1990
for storms which occurred in February 1990.
II.B.2. Tropical Cyclones
The passage of tropical cyclones or hurricanes represents a
synoptic scale, ecosystem level disturbance. Severe storm
(mesoscale) events such as high winds, hail and tornadoes may
accompany or be imbedded in these systems, but overall, impacts are
more widespread. The terms tropical depression, tropical storm, or
hurricane are assigned depending on whether the sustained surface
winds near the center of the system are, respectively, < 17 ms'1, 17
ms'1 to 32 ms"1 inclusive, or > 32 ms"1. Tropical cyclones are not
archived (or named) unless they reach at least tropical storm
strength. The maximum wind speed often must be inferred from
indirect evidence, and a wind speed is subjectively assigned by the
analyst after considering all available information.
The passage through the New England area of disturbances
associated with or derived from North Atlantic tropical cyclones
is not uncommon. Most often, the original tropical system has been
highly modified by contact with continental air masses and is rated
as extratropical by the time New England is reached. When a
tropical cyclone becomes extratropical, the size of the circulation
usually expands, the speed of the maximum wind decreases, and the
distribution of the winds, rainfall, and temperatures around the
center of the cyclone become increasingly asymmetric. While these
characteristic features develop, some tropical features, such as a
30
-------
small area of strong, often hurricane-force winds near the center
and extremely heavy rainfall, may be retained for a considerable
time.
A 1938 New England cyclone is a good example of a storm which
was technically classified as extratropical, but which still
maintained hurricane-like characteristics (Neumann, et al., 1990).
Figure 8 shows the track of this system, labeled as storm number 4
for that year, as it passed through the region. Henry and Swan
(1974) summarize the ecological impacts of damage associated with
this storm while still rated at hurricane strength at a research
forest site in southwestern New Hampshire (Figure 9). Historical
reconstruction at this site indicates the forest canopy remained
essentially intact between 1665 and 1897. After this date, four
windstorms, culminating in the 1938 hurricane, completely destroyed
the canopy trees and set in motion the growth of a new forest. The
authors compare forest characteristics prior to the 1897 to 1938
disturbance period (pre-hurricane) to those of the period 1938-1967
(post-hurricane). They conclude that the post-hurricane forest
exhibits higher tree density, smaller stem diameters and changed
species composition. Although the pre-hurricane and post-hurricane
forests have beech (Fagus grandifolia), white birch (Betula
papyrifera Marsh.) and hemlock (Tsuga canadensis) in common, their
composition differs in that the current forest also contains red
maple (Acer rubrum L.), black birch (Betula lenta L.), sugar maple
(Acer saccharum Marsh.) and striped maple (Acer pensylvanicum L.).
White pine (Pinus strobus L.), present before the hurricane, is
absent in the post-hurricane forest.
Neumann et al.(1990) compile map displays (referred to as
"best tracks") of annual and multi-year tropical storm track
summaries in the North Atlantic. They represent best estimates of
the smoothed path of the cyclone eye as it moves across the earth's
surface. These tracks should be considered as the average path of
the larger scale storm circulation system and not necessarily the
precise location of the eye at any given time.
Since the introduction of continuous weather satellite
surveillance,, there is little chance that a tropical cyclone will
go undetected. There also is a high probability that the center of
the storm can be located within 46 km of its actual position and
the intensity determined to within 5.1 ms'1 of its actual intensity.
Since all of the storm tracks and intensity classifications for the
1964 through 1989 Atlantic hurricane seasons were prepared with the
benefit of satellite imagery (as well as aircraft reconnaissance
and other data), the track accuracy should be near optimum,
considering the scale of the maps and the scale of motion depicted.
31
-------
Figure 8. North Atlantic cyclone tracks for 1938 (Neumann et al., 1990).
U.S. DEPARTMENT OF COMMERCE, WEATHER BUREAU
NORTH ATLANTIC HURRICANE TRACKING CHART
-------
Figure 9. The influence of weather-related disturbance events on
a mixed coniferous and broadleaf deciduous forest in New
Hampshire (after Woodward, 1987).
x Main
•torey
t
tnereaaina
Dominance
1OO -]
80 -
60 -
40 •
2O
O -
Development of
main canopy
Pinua
Tsuga
Picea
Quercua
Acer
Pop,,In.
Pinua —
T»ug
Tauga
(Date) 16BO 17OO 174-O 178O 182O 1B6O
I
Fire
There is some suggestion that there has been an increase in
U.S. east coast storminess since the early 1940's and a shift in
dominant storm track to a position further off the coast (Hayden,
1981). Hayden concludes that these trends are part of a secular
variation of a longer time scale. Future trends in tropical and
extratropical storms and storm tracks under greenhouse warming is
a hotly debated topic. While some scientists argue for increases
in the destructive potential of hurricanes (Emanuel, 1987; Emanuel,
1988) and a higher frequency of hurricanes (AMS Council and UCAR
Board of Trustees, 1988) under greenhouse warming, others maintain
there will be no change in the frequency of occurrence of
Atlantic/Caribbean hurricanes and a significant decrease in the
intensities of such storms (Idso, et al, 1990).
For this report, North Atlantic cyclone events during the most
recent 30 years are enumerated. Magnitude is summarized by life-
stage at landfall and during its passage through the New England
region. Size is broadly summarized by listing the states through
which the storm path passed. Each storm is somewhat unique and
should be individually researched if site specific detail is
required.
33
-------
II.B.3. Drought
Next we move from disturbances resulting in direct windthrow
damage and mortality to a disturbance with more subtle implications
for forest ecosystem health. Drought impacts many aspects of
forest health. As a drought develops, leaf cells progressively
dehydrate and the reduced turgor of leaf cells inhibits cell
expansion (Kozlowski, 1985). At the same time the stomata close,
thereby reducing the rate of photosynthesis by impeding absorption
of C02 through stomatal pores. Changes in hormonal growth
regulators also occur. If drought is prolonged, the reduced
transport of carbohydrates and hormonal growth regulators from
leaves to roots will reduce root growth which will decrease
absorption of water and mineral nutrients. Production of fruits
and seeds can be arrested by drought at any stage of reproductive
growth, including flower bud initiation, opening of flowers,
pollination, fertilization, embryo growth, or fruit and seed
enlargement. Trees in chronic drought areas typically allocate an
abnormally large carbon component to their roots (Smith, 1990).
Field observations of prolonged droughts in the Allegheny Plateau
region of northwestern Pennsylvania result in heavy mortality of
shallow-rooted tree species such as hemlock (Tsuga canadensis) and
yellow birch (Betula alleghaniensis) (Runkle, 1985). Drought has
been implicated as an initiator or important contributor to many
forest declines (Smith, 1990).
A secondary impact of drought is pest infestation. For
instance, spruce budworm outbreaks occur in Maine in stands of
overmature balsam fir under warm, dry May and June conditions
(Ferris, 1980). Trees begin to die after 5 years of defoliation
and mortality is nearly complete after 8 years. Trees under
drought stress may also have lower resistance to pests such as bark
beetles (e.g., reduced resin exudation)(Graham, et al, 1990). In
pine-oak forests of the southeastern United States, drought
predisposes most of the resident pines to successful invasion by
the southern pine beetle (Michaels et al., 1986)
One measured forest response to drought is tree ring width.
A summary of current tree ring analysis techniques is provided in
Fritts and Swetnam (1989). The authors also provide examples of
dendrochronological applications to the assessment of spruce
budworm effects on forest growth, forest decline and climate
variability and change. Puckett (1981) reconstructs a drought
climatology using tree rings and Zahner and Myers (1989) use an
index of drought severity to help identify sources of stress as
reflected in tree ring histories.
One prominent indicator of meteorological drought is the
Palmer Drought Severity Index (PDSI) (Palmer, 1965). The PDSI
represents the deviation of meteorological conditions from
climatological "normals" for some specified area or location.
34
-------
Although the temporal characteristics of the algorithm makes it
poorly suited for estimating the impact of short-term drought on
biological systems (Sadowski, 1975; Cooter,1982; Alley, 1984), it
appears to be a reasonable measure of longer duration drought
stress for forest ecosystems (e.g. Puckett, 1981 and Zahner and
Myers, 1989) . The PDSI response time is of the same order as forest
growth and development, and its use to define large-scale
climatological patterns is well established (Karl, 1983; Eder et
al., 1987).
The PDSI uses mean monthly temperature and precipitation data
and a hydrologic accounting system based upon simple soil and
atmospheric demand models. Moisture is removed linearly from the
top layer of the two-layer soil model. Moisture removal from the
underlying layer is proportional to how close the profile is to
moisture capacity. Potential atmospheric moisture demand is a non-
linear transformation of mean monthly temperature and monthly
precipitation. Actual demand is a function of the potential and
the moisture available to meet the demand. Once atmospheric
demands are met, the remaining precipitation is assumed to refill
the soil profile linearly from the top layer, down. Any
precipitation in excess of that needed to refill the soil profile
and meet atmospheric demand (evapotranspiration) is treated as
runoff.
Palmer (1965) documents the original development of the PDSI.
Monthly PDSI for climate divisions across the U.S. from 1895
through 1990 have been digitized and are distributed through the
National Climatic Data Center (NCDC). Climate Divisions (CDs) are
groups of political (e.g. counties) or geographic (e.g. hydrologic
basins) entities considered to be climatologically homogeneous.
Documentation for these data is available in Karl et al. (1983).
In this study, drought frequency is represented in two ways.
First, time series of area-weighted PDSI values are computed.
Growing season (April-September) means are provided for a baseline
summary and individual months are provided for shorter term
summaries. Statistics of PDSI must be examined with care because
of the influence of the method of calculation. For instance, a
baseline summary of "mean PDSI" would be meaningless if the
baseline was equivalent to the normalizing period of the time
series. By definition, the mean PDSI for this period should be
zero. Likewise, the frequency distribution of PDSI values is
predefined by the characteristics of the normalizing period.
Drought disturbance magnitude has two components, intensity
and duration. Frequency, intensity and duration are summarized in
the baseline climate summary using a return period presentation.
The return period is the amount of time expected to pass before
these or more extreme conditions are experienced. Figure 3
suggests that a 100 year return period would be an appropriate
definition of drought disturbance. The entire computed PDSI record
35
-------
is used in the baseline summary. Frequency and intensity summaries
are based on mean growing season (April through September) PDSI
values. This eliminates higher frequency events that might not
appear in growth indicator measures such as changes in diameter at
breast height (dbh) and tree ring widths. Individual months are
examined to determine drought persistence. A PDSI value of -3.0 or
more negative indicates severe drought. Runs of consecutive months
with PDSI values of -3.0 or less were computed for the period of
record, and the number of consecutive drought months expected once
in 100 years of record was computed.
Size of drought disturbance is estimated in two ways. First,
the map presentations of intensity and duration provide some
indication of the area impacted by drought of similar character.
A second estimate is presented in a decadal summary. In this case,
the percent of the New England region impacted by drought of a
given magnitude is estimated for each year of the time series.
Although similar calculations could be performed for the entire
period of record and regional statistics computed, this kind of
pooled summary should be approached with caution. There are
several different characteristic drought episodes associated with
large-scale atmospheric circulation patterns, each having distinct
spatial and temporal characteristics. An average, aggregate area
mean could be misleading. A more detailed study including
stratification by circulation pattern may be more informative. For
those interested in trend detection, some scientists suggest that
greenhouse warming will first be detected by noting changes in such
large-scale circulation patterns (Michaels and Hayden, 1987) .
II.B.4. National Weather Service Cooperative Network Data
Much of the climatological data available for analysis are
from the National Weather Service's (NWS) cooperative network
stations "summary of the day." These daily arrays contain 24-hour
maximum temperature, minimum temperature, total precipitation,
snowfall and snow depth. At present, cooperative station density
is maintained nationally at a minimum of one station per 1600 km2.
Standard sample variables and methods of collection are detailed in
NWS (1989).
An initial array of potential weather sites were provided
based on a minimum 30-year period-of-record, current reporting
status and completeness of record. For a given station in this
network, several potential inhomogeneities are usually present
(Karl and Quayle, 1988). Changes in instrument location, the type
of instrument, the local environment, exposure and changes in
observational procedures average about five to ten potential
discontinuities per 100 years of record. Karl and Williams (1987)
provide examples of the magnitude and frequency of these
36
-------
inhomogeneities.
Next, the CIS was used to associate each forest sampling area
(hexagon) in Figure 4 with a weather observing site. Climatological
data for 1990 which were unavailable from the NCDC were obtained
from the Northeastern Regional Climate Center at Cornell
University.
Distributions of stations and station distances for NEFHM
hexagons and "nearest" temperature and precipitation data are
summarized in Figures 10 and 11. Overall, 43% of the sampling
sites have access to 30-year precipitation data within 15 km of the
site, 33% can utilize rainfall reports from between 15 and 30 km
away and 24% of sampling locations must use data more than 30 km
away. The maximum distance is 106 km. Most distances greater than
30 km are for sampling hexagons located in the population sparse
sections of northern Maine.
Twenty-nine percent of sampling hexagons are within 15 km of
a cooperative temperature site, while 28% of sampling sites have
access to data within 30 km and 33% of sampling sites must be
characterized by weather observations more than 30 km distant. The
maximum distance is 110 km. All distances greater than 42 km are
again for sampling hexagons located in northern Maine. Future
studies should examine the availability of Canadian data to fill
some of these gaps.
The significance of the spatial distribution of available
climate data for future forest and climate analyses should be
explicitly addressed. For instance, the use of "nearest" station
analysis may be adequate for regional summaries and status, but
more closely associated values, perhaps through statistical
interpolation, may be more appropriate for sampling point analysis
or for making associations between sampling points. The NWS and
the National Environmental Satellite and Data Information Service
(NESDIS) are in the process of considering various strategies for
modernizing the cooperative network. At this time, the NWS does
not anticipate any major changes in cooperative network density or
station locations arising from such efforts (Robert Leffler,
personal Communication).
Once the stations have been selected and assigned to sampling
sites, climatological variables of interest are designated. Basic
variables such as means and extremes of precipitation and
temperature have been identified as important to the NEFHM program
by Brooks et al. (b, in press) and Brooks et al. (c, in press) .
The relationships between these variables and forest response are
fairly well known and are presented in the climate summaries with
no additional explanation. Additional climate variables have been
selected for analysis based on the NEFHM project work plan, Solomon
et al. (1984), Kozlowski (1985) and Peer (1990). Cumulative
growing degree days, last spring freeze and warm spring snowfall
37
-------
Figure 10. Distance between "nearest" cooperative temperature
observation sites and potential NEFHM sampling
hexagons.
DISTANCE FROM FHM SAMPLE
TO NEAREST COOPERATIVE SITE
0 to 15 k i Iome t e r s
15 to 30 k i Iome t e r s
more than 30 kilometers
3P
-------
Figure 11. Distance between "nearest" cooperative precipitation
observation sites and potential NEFHM sampling
hexagons.
o • o o •
DISTANCE FROM FHM SAMPLE
TO NEAREST COOPERATIVE SITE
0 to 15 kilometers
15 to 30 kilometers
more than 30 kil ome t e r s
39
-------
dates were selected. Since the availability and use of these
secondary data may be less obvious, their derivation and
relationship to forest physiology and monitored response will be
discussed in some detail.
Disturbance regime characteristics are summarized in a similar
fashion for each of these primary and secondary variables. Mean
conditions are determined for the baseline summary to delineate
geographic and climatological boundaries for ecosystems. For
instance, high elevation locations can often be identified by
regions of cool seasonal temperatures. Rain shadow effects can be
detected as precipitation minima on the leeward side of these
cooler temperature areas. If we assume that at each location the
distribution of weather observations approximates the normal, the
mean also approximates the mode and so is a measure of greatest
frequency.
As in the case of drought, disturbance magnitude can be a
function of both intensity and duration. The PDSI relates present
conditions to historical record for a particular area. The index
at a point in time captures both intensity and duration aspects of
drought. For instance, an equally negative (severe) drought stress
could represent a short duration intense event as well as the
cumulative effects of long duration less intense events. This is
another reason that time series analysis of PDSI is a far more
useful tool than single value presentations. No such integrated
measures apropos to forest ecosystems exist for the climate
variables included in this discussion.
An alternative approach is to use event likelihood as a
surrogate for magnitude of intensity. Under this system, it is
assumed that existing vegetation has adapted to prevailing (most
frequent) climatological means and variability. Excursions from
these conditions represent situations that could result in
ecosystem change. The more unusual the event, the more likely it
is that some change will result. In this sense, frequency defines
the magnitude of the disturbance (stress). Following this system
we will arbitrarily declare an event with a local return period of
1 in 20 years to be mildly stressful (i.e., there is some chance
that an ecosystem change will result). An event with a return
period of 1 in 50 years is moderately stressful (i.e., there is a
moderate chance that ecosystem change will result). Using Figure
3 as an endpoint, an event with a 1 in 100 year return period is
considered to be extremely stressful( i.e., it is likely that some
ecosystem change will result from this event).
For this report, these probabilities have been estimated from
the instrumental record. Relatively few stations in the reporting
network have 100 or more years of recorded observations. Values
for returns greater than 50 years represent only rough estimates
based on the most recent history. Statistical techniques exist to
estimate more accurate values of rare events and should be explored
40
-------
in future studies.
Disturbance size is characterized by the percent of the New
England region reporting a specified level of disturbance. An
integrated size, frequency and intensity analysis is presented and
discussed in the decadal climate analysis (section III).
II.B.4.a. Growing Degree Days
Solomon et al. (1984) and Peer (1990) provide examples of
research associating cumulative growing degree days (GDD) directly
to forest growth. GDDs are a specific example of the heat unit
concept. This represents a measure of the energy available for
biologic activity or growth. The heat unit has proven to be useful
in a variety of biological settings. For instance, cumulative GDD
totals are associated with winter wheat phenology in Nuttonson
(1955). They are used to limit the number of daily timesteps
permitted for carbon transfer in corn plants in Jones and Kiniry
(1986). The use of degree day totals to predict Southern Pine
Beetle larvel survival and generation survival are discussed in
Gagne et al. (1980) . Species specific GDD ranges as used in a
forest stand simulation model are summarized in Solomon et al.
(1984).
GDDs are usually computed as a temperature deviation from some
physiologically significant base temperature. Upper and lower
temperature thresholds representing limiting conditions may also be
defined. In these cases, GDDs are not computed (i.e., GDD=0).
Solomon et al.(1984) states that forest processes begin between 0
and 10°C. According to Kozlowski (1985), growth of most trees is
optimal between 20 and 35°C, depending on the species. A GDD base
of 10°C is used for this illustration. A lower threshold is
defined as a 24-hour minimum temperature of 0°C. No explicit upper
"capping" temperature is used. Small annual GDD sums are assumed
to represent unfavorable growth conditions and are highlighted in
the climate stress analysis. Physical interpretation of GDDs must
be done cautiously- For instance, similar GDD totals could result
from either long periods of mild temperature conditions or short
periods of extremely warm temperatures. Growth response could vary
markedly given these two sets of conditions.
II.B.4.b. Last Spring Freeze
Primary forest response to late spring hard freezes are
defoliation and, in extreme cases, death. Gerardi and Grimm (1979)
summarize physiological stress to trees as the result of
defoliation. Stress on individual trees depends upon the species,
duration and degree of defoliation, stand site, location of
defoliation in the tree crown and the age and health of the tree at
the time defoliation occurs. Defoliation also affects seed and
f.l
-------
flower production, wood quality, foliation timing, respiration and
efficiency of photosynthesis, size, abundance and food storage
capacity, metabolic food conversion pathways and hormone production
in much the same way as drought. Removal of new foliage, such as
that at first flush in the spring, decreases growth of the upper
stem more than in the lower stem. Late season defoliation (from
early freezes or warm snow events) reduces twigs of woody tissue
and subjects them to winter damage. Specific growth losses from
defoliation include reduced height growth, radial growth losses and
reduction of shoot elongation. Stomatal resistance to transpiration
is also affected by defoliation. This resistance is less in fully
emerged, secondary foliage produced after defoliation than in
primary foliage that has escaped defoliation in oak and aspen;
however, as leaf surface area declines, the demand for moisture
decreases. The decreased demand for moisture improves the internal
water balance of defoliated trees if ample soil water is available;
but defoliation also may increase tree hydration. Drought
compounds the effects on the physiological make-up of trees during
and after defoliation.
A knowledge of interactive effects of natural and
anthropogenic disturbance factors is also critical if climate
change trends are to be detected. Forest response to winter or
freezing conditions is a good example of interacting stressors.
For instance, an exhaustive treatment of the response of red spruce
to winter conditions alone and in combination with anthropogenic
disturbances (ozone, SO2, NO2 and acid deposition) is contained in
Adams and Eagar (1989). A second summary that focuses exclusively
on the ecosystem-wide impacts of nitrogen deposition is presented
in Aber et al. (1989). Both reports discuss the effect of nitrogen
deposition on the development of frost-hardiness in spruce. These
research results could mean that future changes in freeze related
foliar damage to New England forests could be in response to
changing anthropogenic stressors rather than an indication of a
changing freeze disturbance regime.
A freeze analysis can be used to confirm or supplement
observations of adverse weather conditions that might cause growth
decline and tree mortality. Late spring freezes do not leave
sufficient sign for later identification even though, in
combination with other stresses, they have been implicated in most
forest declines (USDA Forest Service, no date). Knowledge of
possible climate related defoliation is also important for pest and
disease studies. Defoliation resulting from freezing temperatures
may combine with later pest related defoliation to contribute to
tree mortality. Spring freeze defoliation can remove a source of
food for pests and actually act to curb epidemics. Spring
defoliation can also weaken trees and predispose them to later pest
infestation. For example, Gerardi and Grimm (1979) report that the
action of shoestring root rot fungus, Arnillaria mellia, and the
two-lined chestnut borer, Agrilus bilineatus in the outer wood may
42
-------
be controlled by the chemical changes occurring there as a result
of defoliation.
Other factors related to frost/freeze damage are the duration
of stressful temperatures and the degree of cold hardiness in
affected plant tissues at the time of freeze (Kozlowski, 1985). The
NWS Cooperative Observer data set provides only 24-hour extremes
with no indication of persistence within the day- Supplemental
hourly data are available from a limited number of National Weather
Service First and Second Order observing stations. These data could
be used to perform more detailed studies if freeze duration within
a 24-hour period is determined to be a critical factor. A second
analysis could focus on minimum temperatures preceding the freeze
event (cold hardiness) and the duration across 24-hour periods of
the freezing event. In the present analysis, a hard freeze is
recorded if the 24-hr minimum temperature is -2.2°C or colder.
II.B.4.C Last Spring Snowfall
Snow has a variety of implications for forest ecosystem health
and productivity. An insulating blanket of snow protects tender
understory species from winter desiccation or damage from periodic
soil heaving. Snow also provides an important source of surface
water storage. Wintertime evapotranspiration is low, and
substantial supplies of moisture can be held until spring. A
slowly melting layer of snow in the spring releases moisture and
nutrients in a fashion that permits maximum vertical infiltration
and minimum surface erosion.
When ice crystals fall through a layer of relatively warm air,
the crystal surface melts. These "wet" flakes then stick to any
surface they come in contact with. The accumulated weight of this
trapped liquid water on leaves and branches causes loss of limbs
and uprooting of entire trees. One particular case, the "warm
snowstorm of May 9, 1977" is discussed in Gedzelman and Lewis
(1990) . This storm was one of the latest occurring major
snowstorms in the lower Hudson River Valley on record. The
accumulated weight of the wet snow on newly sprouted leaves caused
extensive destruction to trees (see impacts of defoliation in
section II.B.4.b.).
A traditional source of snowfall data is the NWS Cooperative
Observer Network. Unfortunately, these data are particularly poor
in terms of quality control and spatial coverage. But when used
with caution, they can provide valuable information for forest
scientists. For example, it is difficult to determine from the
digitized summary-of-day reports if a snowfall event was warm or
cold. As an alternative to extensive archival research, the
present study assumes that if, at a particular location, a spring
snowfall event is expected on a later spring date fewer than one
43
-------
year in 20, it is likely to be a warm snow event.
A discussion of snow data error and bias is presented in
Robinson (1988) . Although unavailable for the present study, a new
quality controlled set of snowfall data have recently been
developed based on the work of Robinson and can be ordered from the
National Climatic Data Center. Future work should consider warm
fall snow events, use of the Robinson data base and more accurate
determination of event occurrence from the digitized record.
II.C Display Technology (CIS)
Geographic Information System (GIS) software was used to study
these spatial relationships between the climatology of the New
England area and the sampling locations in the Forest Health
Monitoring Project. The GIS system used for the display and
analysis of these relationships was Environmental Systems Research
Institute's ARC/INFO. The types of spatial analysis performed were
proximity analysis, surface modeling and overlay analysis. The
procedures used for each of these three analyses are discussed
below.
For the proximity analysis, the GIS was used to associate
conditions measured at nearby National Weather Service Cooperative
Observer locations with the 263 potential NEFHM sampling locations.
The ARC/INFO "NEAR" command was used to identify the closest
cooperative observer station for each of the 263 NEFHM sampling
locations. Once identified, the information from the observer
station is linked to the associated NEFHM sampling location and
used to characterize the likelihood of forest response to
climatological disturbances.
Surface models were created from long term cooperative
observer station data for annual growing degree days, date of last
spring freeze and date of last spring snowfall using Dynamic
Graphics' Interactive Surface Modeling (ISM) software. The contour
lines generated from ISM during the modeling process were then
moved to ARC/INFO for overlay and display purposes.
The GIS was then used in the final overlay analysis which
characterizes individual sampling hexagons from a weather event
perspective. This process involves displaying and analyzing events
that occur in the same spatial domain. The events analyzed
included wind, hail and tornado events. The analysis produced maps
of individual hexagon sampling sites displaying with the associated
weather events that have occurred at that location. This analysis
also produced descriptive statistics on the number of weather
events (intersections) occurring within the sampling frame of 263
NEFHM locations.
44
-------
These analytical techniques were also used in combination to
further analyze relationships. For example, to estimate the
growing degree days at a particular sampling location the surface
model for the growing degree days was overlaid on the sample
locations. Similar analyses were performed for the spring snow and
freeze data.
II.D. Summary
Disturbance regime climatology examples have been prepared
targeted to the joint, 1990 EPA EMAP-forest and NEFHM project
conducted in six states in the northeastern United States. Weather
and climate databases have been assembled from a variety of
sources. Secondary or derived variables of potential interest to
these programs have been identified. The physiological impacts of
these variables on forest ecosystems have been summarized and
expected forest responses were noted. Table 6 summarizes all
climate disturbance products developed using the techniques
discussed in Chapter II that are applied to the New England region
in Chapter III. A blank entry in Table 6 indicates disturbance
attributes that have not been quantified for this report.
Table 6. Examples
England
of climate disturbance (stress)
(see Chapter III).
products for New
CLIMATE VARIABLE
Tornadoes
Wind
Tropical Cyclone
Drought
Temperature
Precipitation
Growing Degree
Days
Late Spring
Freeze
Late Spring
Snowfall
FREQUENCY
intra-annual
intra-annual
X
X
X
X
X
X
X
MAGNITUDE
intensity
only
intensity
only
intensity
duration
intensity
only
intensity
only
intensity
only
intensity
only
intensity
only
SIZE
X
X
X
X
X
X
X
X
45
-------
III. Example Disturbance Climatologies
III.A. Introduction
The results presented below represent examples of disturbance
(stress) climatologies. The use of, and caveats associated with
these products are discussed. A portion of this material already
appears in Brooks et al. (b, in press) and Brooks et al. (c, in
press) and is being used by the EMAP Forest and Assessment teams
(Kucera and Martin, 1991). Further operational and research
applications are anticipated in the future.
III.B. Baseline Conditions (1961-1990)
The first analysis characterizes aspects of the long-term, or
background climate of the New England region. Using the earlier
analogy of ecological persistence and recurrence, the background
climate defines, in part, the "bounds" for the recurring ecosystem
states. Table 7 summarizes elements of a disturbance regime
climatology for which baseline examples have been developed. With
the exception of event duration, frequency and magnitude of
disturbance (stress) are well represented at this level of
analysis. In most cases, event size is not characterized until the
decadal analysis.
Table 7. Elements of a New England disturbance regime climatology
for which baseline examples have been developed.
CLIMATE VARIABLE
Tornadoes
Wind
Tropical Cyclone
Drought
Temperature
Precipitation
Growing Degree
Days
Late Spring
Freeze
Late spring
Snowfall
FREQUENCY
intra-annual
intra-annual
X
X
X
X
X
X
X
MAGNITUDE
X
X
X
X
X
X
X
X
SIZE
X
X
46
-------
III. B.I. Severe Weather and Tropical Cyclones
These phenomena are presented together because their primary
impact on forest ecosystems is through windthrow. The primary
difference between these events is their frequency and size.
Disturbance frequency is presented in terms of intra-annual
variability for wind and tornado events and by enumeration for
tropical cyclones. Time series are not appropriate for severe
weather events because of observational biases discussed previously
and because of the infrequency of cyclone events.
Figure 5 illustrated that there have been numerous tornado and
damaging wind event reports throughout New England during the last
30 years. There has been confirmation of 278 tornadoes and 1351
damaging wind reports in the region from 1961 through September,
1990. Figure 12 summarizes the intra-annual frequency distribution
of these events. Figure 12 shows that the within-year distribution
of wind and tornado events are similar, with a wintertime minimum
and July maximum. This is to be expected since both phenomena are
generated by similar kinds of unstable atmospheric conditions.
Figure 12.
Relative intra-annual frequency of tornado and damaging
wind events in New England, 1961-1990.
o
CD
Z3
O~
CD
1OO
80
6O
4-0
20
Total - 278 ^Zl tornado
Total - 1351 E23 wind
r§
2 Z
I7L_ r-T_
1 2 3 A 5 6 7 8 9 1O 11 12
Month
Figure 13 contains a summary of the distribution of New
England tornado magnitude (F-scale rating). No intensity analysis
is performed for damaging winds because of poor records of peak
gusts. Windspeeds associated with F-0 events break branches off
trees and push over shallow-rooted trees. Windspeed associated
with any event rated F-2 or higher can result in large trees being
snapped or uprooted. Figure 14 indicates that these events account
for nearly 30% (about 83) of all tornadoes reported since 1961.
47
-------
There have been approximately three tornado events containing
windspeeds in excess of 89 ms'1.
Figure 13. Distribution of tornado intensity for New England,
1961-1990.
o
CD
CT
CL5
1OO
BO
eo
4O
20
Total « 278
1 2 3
F-Scale
The majority of reported tornadoes either touch-down only
briefly (point events) or have such short paths that they are not
recorded. Based on the available record, however, the average of
2.59 km2 is a conservative estimate of the area impacted by tornadic
events each year. The largest documented area impacted in a single
year between 1961 and 1990 was 24 km2. This damage was generated
by three storms reported in 1979.
Although not as common as tornado or wind events, tropical
cyclones pass through the New England area with some regularity.
Table 8 enumerates these events and estimates storm strength and
the states impacted. There have been nine storms in the last 30
years capable of damaging the forest canopy. The last confirmed
storm (through September, 1990) capable of producing significant
windthrow in the New England states occurred in 1976. The most
frequently impacted states are Connecticut, Massachusetts, Maine
and New Hampshire. For more detailed information concerning the
tracks and life stages of these events, see Neumann et al (1990).
48
-------
Table 8. Tropical cyclones impinging on the New England states,
1961-1990.
Year
1961
1961
1966
1971
1971
1972
1976
1979
1979
1985
1988
Name
-
Esther
Alma
Doria
Heidi
Carrie
Belle
Frederic
David
Gloria
Chris
Life
Stage*
T
D
EX
T EX
T EX
EX
H T
T EX
T EX
EX
EX
Dates
Sept 12-15
Sept 11-26
June 4-14
Aug 20-29
Sept 10-14
Aug 2 9 -Sept 5
Aug 6-10
Aug 29-Sept 14
Aug 25-Sept 7
Sept 16-Oct 1
Aug 21-29
States
Impacted
CT,MA,ME,NH
ME
MA
CT,MA,ME,NH
ME
ME
CT , MA , ME , NH
VT
ME,NH,VT
CT,MA,ME,NH
CT,MA
* H = Hurricane (windspeeds equivalent to F-l or greater)
T = Tropical Storm (windspeeds equivalent to F-0 to F-l)
EX = Extratropical Storm (windspeeds may vary from F-0 to F-l)
D = Tropical Depression (windspeeds of F-0 or less)
49
-------
III.B.2. Temperature and Precipitation
There are many possible averaging periods and derived variable
options to represent temperature and precipitation conditions.
Temperature extremes (annual maximum and minimum) and annual
precipitation (October through September) were identified as
particularly useful in monitoring forest health in the Northeastern
U.S. (Brooks, et al., b, in press). Persistence characteristics of
these variables are summarized in contour maps of mean conditions.
Frequency is illustrated by weighted time series graphs. Relative
magnitude of disturbance can be visually estimated from the time
series presentation. No size (area) summary is presented at this
time.
Contour map summaries of the most recent 3O years for mean New
England precipitation and temperature extremes are presented in
Figures 14, 15 and 16. Two features make these products unique.
First, atlas summaries have often focussed on the presentation of
all available data, rather than data selected for specific
applications. In this study, only data from stations that have
been associated with NEFHM program sampling locations are analyzed.
The selected station data are then spatially smoothed using a bi-
harmonic cubic spline to generate spatial contours. The degree to
which these data, some of which are from instruments located more
than 40 km away from the FHM sampling location, represent
conditions at a monitoring site has yet to be explicitly addressed.
Statistical estimation techniques, such as kriging or cokriging
could be applied to climate means (e.g., Dingman et al., 1988).
Interpolation of extreme values such as annual maximum and minimum
temperature is significantly more complex. The selection of
appropriate interpolation techniques should be dictated by the
needs of the monitoring program and will be addressed in future
research.
A second feature of the contour presentation is that annual
values are based on the months October through September rather
than a calendar year. This period was chosen so that a 1990 value
represents events leading up to and including the forest sampling
period and does not overlap previous or following growing seasons.
All values on contour or time series graphs are labeled with the
calendar year in which the averaging period ends. This label also
corresponds to the year containing months of significant tree
growth.
The most obvious ecological boundary effects illustrated by
these figures are a function of latitude and elevation. Latitude
and elevation play an important part in patterns of annual
precipitation (Figure 14) . Moist conditions prevail in the
southern portion of New England and on the windward side of
mountain peaks and extensive mountain ranges. Slightly drier
conditions are observed in the colder more northern latitudes and
50
-------
Figure 14. Mean annual New England precipitation, 1961-1990
PRECIPITATION
(in centime ten)
less than 45.0
ITTT1 45.0 to 60.0
60.0 to 75.0
75.0 to 90.0
90.0 to 105.0
105.0 to 120.0
120.0 to 135.0
135.0 to 150.0
150.0 to 165.0
165.0 to 180.0
180.0 and over
51
-------
leeward of significant mountain ranges. Latitudinal affects are
also reflected by zonal bands of northward-decreasing temperatures
in homogeneous geographic areas and are most notable in minimum
temperature patterns (Figures 15 and 16) . Distortions of these
broad patterns occur in mountainous areas. Elevation does not play
a dominant role in annual minimum temperature patterns across New
England, but is more apparent in patterns of annual maximum
temperature.
Temporal variability of regional temperature and precipitation
variables can be visually summarized by weighted time series
(Figures 17, 18, 19 and 20). Since the NEFHM sampling design
generates a nearly uniform grid of points, weighted area means are
constructed by associating climate values with their appropriate
sampling hexagons, summing and dividing the results by the total
number of hexagons in the region.
Area-weighted annual precipitation in New England ranges from
a low of 78.7 cm in 1965 to a high of 140 cm in 1983 (Figure 17).
The time series appears equally variable throughout the 30-years of
data and no statistically significant linear trend has been
detected. The largest year-to-year change, 45 cm, occurs between
1984 and 1985.
Inter-annual variability is a notable feature in the time
series of area-weighted minimum temperatures (Figure 18). Values
range from -32°C during 1971 to -23.5°C during 1980. Particularly
large year-to-year variability is noted from 1979 through 1985. A
small but statistically significant linear trend of .18 °C/year has
been estimated for the 30-year time series.
Periods of smaller year-to-year change are seen in the time
series of area-weighted annual maximum temperatures (Figure 19).
The greatest year-to-year change occurred between 1974 and 1975.
1974 was such an extreme year that it dominates our perception of
the slope. The combined effect of area-weighted maximum and
minimum temperature trends could narrow annual temperature ranges.
This, in fact is the case. A significant linear trend with
negative slope equal to that determined for the annual minimum
temperature has been estimated. Figure 20 indicates increased
annual range variability beginning in 1979. Such possible trends
in regional climatological averages and variability must be
interpreted carefully. For instance, for global or hemispheric
averages, it is likely that random inhomogeneities in the station
data will tend to cancel each other out (Karl and Quayle, 1988).
This is not as likely in the calculation of regional averages where
far fewer stations are used. Furthermore, station inhomogeneities
add artificial variance to the time series.
52
-------
Figure 15. Mean annual maximum New England temperature, 1961-1990
DEGREES CENTIGRADE
15.0 to is.o
18.0 to 21 .0
21.0 to 24.0
7777K 24.0 to 27.0
27.0 to 30.0
30.0 to 33.0
33.0 to 36.0
36.0 to 39.0
53
-------
Figure 16. Mean annual minimum New England temperature, 1961-1990,
DEGREES CENTIGRADE
••I -41.0 to -38.0
-38.0 to -35.0
-35.0 to -32.0
-32.0 to -29.0
-29.0 to -26.0
-26.0 to -23.0
-23.0 to -20.0
-20.0 to -17.0
-17.0 to -14.0
54
-------
Figure 17,
Area weighted time series of annual New England
precipitation (cm), 1961-1990.
p
R
E
c
i
p
i
T
A
T
I
O
N
CM
140
130 '
120 '
110
100
90
80
70 i,—
1960
1970
1980
1990
YEAR
Figure 18. Area weighted time series of annual minimum New England
temperatures (°C), 1961-1990.
T
E
M
P
E
R
A
T
U
R
E
C
-23 '
-24
-25 '
-26
-27 1
-28
-29
-30
-31
-32
-33
I960
1970
1980
1990
TEAR
55
-------
Figure 19. Area weighted time series of annual maximum New England
temperatures (°C), 1961-1990.
T
E
M
P
E
R
A
T
U
R
E
C
37
36
35
34
33
32
31 V-
1960
1970
1980
1990
TEAR
Figure 20. Area weighted time series of annual New England
temperature range (°C), 1961-1990.
R
A
N
G
E
C
66
65
64
63
62
61
60
59
58
57
56
1960
1970
I960
1990
YEAR
56
-------
In the New England forest case, such inhomogeneities may not
represent a significant problem in well instrumented areas where
each station may represent from 1 to 3 sampling locations.
However, significant bias may be introduced into a time series when
a larger number of forest sites must be represented by a single
weather observation. In New England, all such cases occur in Maine
- the state which also dominates the region in area. Thus, a
situation exists in which the greatest influence on the regional
estimate is derived from the region with the least data. Errors in
sparsely instrumented areas of Maine could easily be responsible
for visually detected trends in the regional time series (e.g.,
Figure 17). If studies of ecosystem behavior at these latitudes
are considered to be critical for climate trend detection and
impact analyses, additional long-term climate observation sites are
needed.
The accuracy of climatological data and summaries aside, a
second fundamental question concerns the importance of such changes
for forest ecosystems. For instance, if we assume that existing
vegetation is ideally adapted to prevailing climatological means
and extremes, how great a change is necessary before detectable
ecosystem changes occur? Although the majority of plant growth and
development can be generalized as responding to modal (most
frequent or dominant) climate conditions, it is extreme conditions
or events that most severely constrain species range and potential
productivity. Responses to trends such as that noted in annual
temperature range would most likely be observed at the margins of
species range (ecotones). Warmer minimum temperatures (Figure 18)
could create conditions in these areas favorable for expansion of
marginal southern species into existing ecotones, or facilitate
expansion of the ecotone itself. Although answers to some "how
important is this?" questions are already available in the
literature, other such questions require more detailed modeling
studies. The goal of this report is to develop ways of describing
climatological conditions so that associations between climate
conditions (e.g. means, trends and variability) and indicators of
forest status and health can be postulated and tested. This is a
fundamental EMAP activity. If such associations can be confirmed,
then research aimed towards the development of predictive response
models can be pursued. This could occur either within the FHM
program of entirely outside the EMAP/FHM initiative.
III.B.3 Drought
Drought frequency, intensity and duration products have been
prepared as a baseline analysis. The spatially aggregate nature of
the drought database makes contour presentations inappropriate. A
time series of regional mean growing season PDSI can be used to
estimate drought frequency (Figure 21) . In this case, Climate
Division (CD) PDSI values were assigned to each NEFHM sampling
hexagon within the division. The 1960s are dominated by drought
57
-------
Figure 21.
Area weighted timeseries of mean growing season PDSI,
1961-1990.
PDSI
3
2
1
0
-1
-2
-3 1
-4
I960
1970
1980
1990
YEAR
GE 4.00 EXTREMELY WET
3.99 - 3.00 VERY WET
2.99 - 2.00 MODERATELY WET
I.99 - I.00 SLIGHTLY WET
0.99 - 0.50 INCIPIENT WET
0.49 TO -0.49 NORMAL
-0.50 TO -0.99 INCIPIENT DROUGHT
-1.00 TO -1.99 MILD DROUGHT
-2.00 TO -2.99 MODERATE DROUGHT
-3.00 TO -3.99 SEVERE DROUGHT
LE -4.00 EXTREME DROUGHT
58
-------
conditions that peak at a region-wide PDSI value of nearly -4.0 in
1965. This was followed in the early to mid 1970s by wet
conditions. Regional growing season PDSI values have ranged from
"slightly dry" (-1.50) to "slightly wet" (+1.50) from 1975 through
1990.
Drought magnitude is described by intensity and duration.
Based on relative frequency, Figure 22 contains monthly CD PDSI
values expected to occur once in 100 calendar years of record.
Figure 3a indicates that disturbances of this order are
hierarchically compatible with a number of ecological process and
responses. Since nearly 100 years of computed PDSI values are
available for analysis (1895-1990), no statistical modeling is
required and these estimates may be considered to be relatively
accurate representation of drought disturbance events. Most
intense drought conditions have occurred in coastal Maine, western
and southern Vermont, western and coastal Massachusetts,
northwestern and coastal Connecticut.
New England drought magnitude expressed as duration is
summarized in Figure 23. Over the past 96 years, each CD has
averaged 28 multiple-month droughts of PDSI equal to or less than
-2.00. The values shown on Figure 23 were computed from a
frequency analysis of these data, and represent drought duration
expected once in 100 such multiple-month events. Length of the 1
in 100 chance drought varies from 23 months in Eastern
Massachusetts to 42 months in Western Vermont. There is
substantially more geographic variability in New England drought
persistence than in drought intensity.
III.B.4. Growing Degree Days
A contour map of mean growing degree days and a weighted time
series are presented as baseline products (Figures 24 and 25) .
Return periods have also been calculated and will be used in the
decadal analysis.
Figure 24 indicates the fewest number of degree days in
northern Maine and the mountains of New Hampshire. A tongue of GDD
maximums follows the Connecticut River valley northward into
Massachusetts. Other areas of warmth center on major metropolitan
areas of Massachusetts and Rhode Island.
An area weighted time series of annual GDD is presented in
Figure 25. No trend is evident in the time series. A maximum of
1030 GDD is estimated in 1988. A minimum value of 810 GDD is
estimated for 1986. Maximum year-to-year change appears to have
occurred between 1986 and 1987.
59
-------
Figure 22. Monthly PDSI values for New England with a l in 100
year chance of occurrence, 1895-1990.
-------
Figure 23.
Duration of New England drought episode (PDSI with
consecutive months of -2.0 or less) with a 1 in 100
event chance of occurrence, 1895-1990.
NUMBER OF CONSECUTIVE
• GE 50 MONTHS H
• 49 TO 40 MONTHS
• 39 TO 30 MONTHS
OH 29 TO 20 MONTHS
19 TO 10 MONTHS
MONTHS POSI LE -2
10 TO 0 MONTHS
00
61
-------
Figure 24.
Mean annual New England growing degree days
(base = 10°C, lower limit = 0°C) , 1961-1990.
62
-------
Figure 25. Area weighted timeseries of annual New England growing
degree days (base = 10°C, lower limit = 0°C) , 1961-1990.
1100
1000
D 900
D
800
700 <
I960
1970 1980
YEAR
1990
63
-------
III.B.5. Last Spring Freeze
A contour map of the mean last date on which a minimum daily
temperature of -2.2°C or less is reported in the spring and a
weighted time series of these dates are presented as baseline
products (Figures 26 and 27) . Return periods have also been
calculated and will be used in the decadal analysis.
Figure 26 shows essentially the same latitudinal and elevation
patterns as GDD. On average, the latest spring freeze dates occur
at northern latitudes and at higher elevations. Freezing
temperatures disappear earliest in the spring along the eastern
seacoast and large metropolitan areas.
Figure 27 contains an area-weighted time series of last spring
freeze dates. Between 1960 and 1980, the average date of the last
spring hard freeze event occurs progressively earlier in the
season. Although Figure 27 could be interpreted to reflect a
changing climate in New England, other potential sources of
observational or recording error should be investigated in future
studies. One potential source of error reported in Karl and Quayle
(1988) involves shifts in time of observation from evening
to a morning observation. This can create cooling bias in the data
record.
III.B.6. Last Spring Snowfall
A contour map of the 30-year mean date of the last spring
snowfall and a weighted time series of these dates are presented as
baseline products (Figures 28 and 29) . No accumulation (snow-on-
the-ground) is considered because of the poor quality of these
data. This variable, then, represents a risk of snow damage, but
not that damage has actually occurred. The later in the spring the
snow event occurs, the more severe the forest stress is assumed to
be and the greater the likelihood of defoliation. No underlying
physical assumptions concerning atmospheric conditions conducive to
warm snow events are made, and this presentation is meant only to
be illustrative.
Figure 28 suggests a geographic pattern of last season
snowfall similar to that of GDDs and late spring hard freeze dates
with at least one interesting exception. An unusual area of warm
and dry conditions (last snowfall prior to March 1) appears in
central Maine. Although this site is on the drier, leeward side of
the Appalachian Mountains, the record should be investigated to
further identify the source of this unusual signal.
64
-------
Figure 26.
Mean date of last spring hard freeze (minimum
temperature of -2.2°C or less), 1961-1990.
Dote of lost hard
-------
Figure 27
Area weighted timeseries of last spring hard freeze
event dates (minimum temperature of -2.2°C or less) in
New England, 1961-1990.
MO
c
A
L
E
N
D
A
R
D
A
Y
130
120
110
I960
1970
1980
1990
YEAR
66
-------
Figure 28. Mean date of last spring snowfall, 1961-1990.
Dote of last snowfall
before March 1
TTI March 1 to Kerch 20
^ Morch 21 to April 9
^H Apri I 10 to Apri I 29
after April 29
67
-------
Figure 29 contains an area-weighted time series of last spring
snowfall dates. Although the earlier record (1961-1980) is
difficult to interpret, there is a marked date increase (later in
spring) in the decade of the 1980s. This, in combination with
regional freeze data (Figure 27) suggests that during the 1980s the
risk of damage associated with late spring snowfall events may have
increased. Late spring hard freeze events that might delay bud
initiation occur earlier in the spring (Figure 27), while snowfall
events occur later (Figure 29). Note that the term "risk" is used
here. Given the present dataset, it cannot be stated with
certainty that the number of damaging snow events has increased. It
does not contain the information needed to verify the moisture
content of the snow. In the future, newly developed snowfall data
sets, used in combination with synoptic meteorological map analysis
could be used to more carefully define such events.
68
-------
Figure 29. Area weighted timeseries of last spring snowfall dates
in New England, 1961-1990.
c
A
L
E
N
D
A
R
D
A
Y
130
120
1 10
100
90
80
I960
1970
1980
1990
YEAR
69
-------
III.C. Most Recent Decade (1981-1990)
In contrast to the 30-year climatology, a 10-year study is
used to emphasize short-term variability of climate stress. This
also represents more closely the regional response time of forest
ecosystems to chronic climatological stress. Since detectable
ecosystem health and productivity response times are relatively
slow, substantial data sets will be required before true ecological
trend analyses over relatively small areas can be performed. Under
the present EMAP Forests four year sampling design, it is estimated
that detection of regional and national trends in response
indicators on the order of 1% per year is expected within 10-15
years (Hunsaker and Carpenter, 1990) .
The period 1981 to 1990 has been chosen to illustrate regional
climate fluctuations within a decade which approximates the time
interval since the last USDA Forest Inventory Analysis (FIA) Survey
(see table below).
DATES OF LAST FIA SURVEY
Connecticut 1984
Maine 1982
Massachusetts 1984
New Hampshire 1983
Rhode Island 1984
Vermont 1983
Table 9 summarizes the characteristics of the regional stress
regime for which sample climatologies covering the period 1981-1990
have been developed. A comparison of Tables 8 and 9 shows a
greater emphasis on spatial characterization in the decadal
analysis (Table 9) than in the baseline study- Since temperature
and precipitation have been addressed extensively in the baseline
discussion, they are omitted from the decade analysis. As is the
case throughout all these example climatologies, no effective means
of wind event magnitude and areal extent characterization has been
identified.
70
-------
Table 9. Elements of a New England disturbance regime climatology
for which decadal examples have been developed.
CLIMATE VARIABLE
Tornadoes
Wind
Tropical Cyclone
Drought
Temperature
Precipitation
Growing Degree
Days
Late Spring
Freeze
Late spring
Snowfall
FREQUENCY
intra-annual
intra-annual
X
X
X
X
X
MAGNITUDE
X
X
X
X
X
X
SIZE
X
X
X
X
X
X
III.C.I. Severe Weather and Tropical Cyclones
The most recent decade has been relatively quiet in terms of
large scale windthrow events. There has been a total of 58 tornado
events reported. The annual mean number of reported events dropped
from a 30-year value of 9 to only 6 per year. Seventy-five percent
of these storms were of strength F-2 or less. Only two tornadoes
of strength F-3 or greater were reported. Annually, the average
area impacted by tornadic events was only 1.6 km2.
The number of damaging wind events has shown a marked increase
in the last decade. It is not clear if this represents a physical
trend or if it reflects changes in reporting techniques. There
have been 832 damaging wind reports in the last decade. The decade
average of 83 events each year is nearly double that of the last 30
years. Incompleteness of data records prevents any generalization
of peak gusts or area impacted by these storms.
Only two tropical cyclone-spawned storm systems passed through
New England in the last decade (Table 8). Both were classified as
extra-tropical storms. The larger of the two, Gloria, passed
through Connecticut, Massachusetts, Maine and New Hampshire in
1985.
71
-------
In summary, the most recent decade has generally lacked large-
scale windthrow events such as tornadoes and well-developed
tropical cyclones. A significant increase in the number of
damaging wind events (most often described as downburst type events
associated with dissipating severe weather) is suggested by the
data, but verification of this increase has not been performed.
Research by Canham and Loucks (1984) for downburst events in
Wisconsin suggests that forest damage from individual events should
be on the order of 1 km2 or less.
III.C.2 Combined Stress (Disturbance) Analysis
Figure 30 contains a decadal summary of drought, GDD sums,
spring freezes and warm spring snowfall events in terms of the
percent of total area impacted. Small GDD totals, spring freeze
and snowfall dates are included if, based on the entire climate
record for a station, the current value is expected to occur fewer
than once in 20 years. Although more extreme conditions such as a
100 year freeze are suggested as appropriate by Figure 3a, the
relatively short period of record available does not support the
direct calculation of these values. Such rare events might be
statistically estimated through the use of extreme value theory in
future studies (Mearns, et al., 1984; Katz and Brown, 1991).
Figure 30.
Percent of New England region impacted by climate
stress, 1981-1990.
0
C£
LU
Q_
100
80
60
40
20
II freeze
gdd
snow
drought
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990
YEAR
72
-------
In order to determine a comparable drought return period, a
frequency analysis was performed using the minimum PDSI value
reported in each CD during each October through September calendar
year. Figure 31 presents the PDSI values expected in each CD once
in every 20 calendar years. These values range from -3.62 to
-4.10. To simplify calculations, a value of -4.0 was used to
declare a forest growth year to have suffered under drought stress.
Figure 31.
Monthly PDSI values for New England with a 1 in 20 year
chance of occurrence, 1895-1990.
73
-------
The purposes of Figure 30 are 1) to facilitate visual
identification of short-term trends in forest related climate
variables; 2) to identify the frequency and size of regional-scale
climate events; and 3) to identify associations of stressful
climate conditions. This figure does not quantify the amount of
forest damage or forest response to these stresses. The impact of
environmental stress will vary with ecological system (although all
of these measures reflect local conditions) and combinations of
stresses may result in threshold behavior (discontinuities) or non-
linear increases in damage. This figure represents a consistent
intensity of stress across stressors and permits, for this decade,
some generalizations of size/frequency relationships.
Figure 30 indicates that one drought event occurred (PDSI
during at least one month equal to or less than -4.00) in the New
England region during the last decade. Late spring freezes were
associated with at least a few sampling locations during nine of
the ten years. The freeze event of largest spatial extent occurred
in 1988 and impacted 8% of the region. Late spring freezes and
unfavorably cool growing conditions throughout the year (small GDD
totals) were not clearly associated. The most widespread GDD event
was reported in 1982 and affected 16% of the New England region.
The most extensive late spring snowfall event of the last decade
occurred in 1990 and impacted 20% of New England. Thus, for the
decade of the 1980s, at the level of stress represented by a 20
year return period, the largest scale events (spatially) were late
spring snowfall and low GDD totals.
Next, this stress information is related to the NEFHM program
sampling network. Figure 32 shows the distribution of the number
of event "intersections" with NEFHM sampling hexagons over the
decade. A hexagon intersection is said to occur when a disturbance
event is identified in the climatological record associated with a
particular NEFHM sampling hexagon. The maximum possible number of
intersections for each hexagon is 40 (drought, freeze, snow and GDD
for each of ten years). The areas of interest are the "tails" of
the distribution. Hexagons counts located at the lower tail (zero
intersections) represent areas experiencing few climate
disturbances (as defined by our four events and 20-year return
period). Using the earlier conceptual description of climate-
ecosystem interactions, these are areas that might be expected to
persist or show slow or little change. Dramatic changes in forest
status or health in these areas are likely associated with
something other than the climatological stresses examined here.
74
-------
Figure 32. Distribution of hexagon and climate event
intersections, 1981-1990.
>-
O
ID
O
LU
cr.
100
80 -
60 -
40 -
20 -
00 123456789
INTERSECTIONS
Hexagons in the upper tail represent areas subject to a
relatively larger number of stresses. Evidence of ecosystem
response to these factors could take the form of changes in
ecosystem structure or characteristics or changes in overall system
health (resistance to pest and disease injury). The present stress
analysis, however, is not complete. Disturbance persistence should
be included as well. Under the present system, a region that
experiences stresses that are few in number but persistent would
not be highlighted. The estimation of drought persistence is
straightforward and has been demonstrated in the background study.
Late spring freeze persistence could also be easily defined. The
development of a definition of persistence for GDD and late spring
snowfall is more difficult.
Figure 33 contains the location of hexagons reporting 4 or
more event intersections. This is equivalent to the upper 3% of
the frequency distribution shown in Figure 32. Five cooperative
75
-------
weather network stations define the three intersection areas of
Figure 33. Three of these five stations define the Massachusetts
area, indicating this is likely to be a climatologically active
region. Further investigation shows the Vermont area becomes more
firmly defined when the next lower intersection frequency category
is added to the map. Alternatively, the third area which is
located in New Hampshire remains poorly defined, even when hexagons
with lower intersection frequencies are added. This suggests
"noisy" weather data or local-scale phenomena rather than a pattern
of regionally important climatological disturbances.
76
-------
Figure 33 T
JJ- Locati
O O 0 O
O 0 O O O
O O O O O <>
O O O O O O
O O O O O O I
O O O O O O O
O O O O O O
ooooooooo,
OOOOOOO.OO
oooooooooo
0000000000^
OOOOOO/OOO O^J^/
77
-------
III.C.3. Summary
The results of the 1981-1990 analysis suggest that this was a
relatively quiet period with no significant mesoscale disturbances
(tropical cyclones), few tornadoes and only small scale (area)
disturbances throughout the region. By way of comparison, a
similar regional hexagon analysis was performed for the decade of
the 1960's. This is a period of particular interest because high
elevation red spruce decline is suggested to have been triggered by
a combination of climate and pollution stresses during this time
(Johnson et al., 1988). Figure 34 contains the same data
presentation as Figure 30. It is immediately apparent that this is
a more active climatological period. Three consecutive large scale
drought years dominate the figure from 1964 into 1966. Late
snowfall events also are more frequent and impact a larger area
than during the 1980's. Nearly one-third of New England was
impacted by a late spring freeze in 1961 and 20% was impacted in
1969. GDD totals do not appear to be of major importance in the
1960's. Figure 35 contains the distribution of hexagon/climate
stress intersections. Hexagons with four or more intersections
represent 31% of the 1961-1970 sample as compared to 3% of 1981-
1990 hexagons. The bulk of these sites is concentrated in the
Northern three New England states (Figure 36).
Figure 34. Percentage of New England region impacted by climate
stress, 1961-1970.
100
80
60
CJ>
C£
LU
Q_ 40
20
I . I freeze
[//A gdd
Vr^4 snow
^1 drought
Jk
n
n
1961 1962 1963 1964 1965 1966 1967 1968 1969 1970
YEAR
78
-------
Figure 35. Distribution of hexagon and climate event
intersections, 1961-1970.
CJ
z
LJ
LJ
LT
100
0 12345678
INTERSECTIONS
79
-------
Figure 36. Location of hexagons reporting four or more
intersections with climate disturbances, 1961-1970.
80
-------
Johnson et al. (1988) attribute red spruce declines in the
White, Green and Adirondack Mountains to a series of cold winters
in 1961, 1962 and 1963. Although there is agreement that this is
a likely source of decline at higher elevations, there is some
disagreement concerning the interpretation of tree-ring patterns
and mortality elsewhere (Van Deusen et al., 1991) . The time series
in Figure 34 does suggest unusual winter conditions in 1961 and
1963, but also indicates that these conditions were followed by
severe drought and continued unfavorable winter conditions in 1966,
1967 and 1969. Johnson et al. (1988) found conflicting evidence of
the association of PDSI with ring width, but drought as an
additional source of stress cannot be entirely discounted. Zahner
et al. (1989) found PDSI very useful in evaluating tree-ring
chronologies in conifers of the Southeast.
Johnson et al. (1988) found close relationships between cold
early winters prior to the growing season and warm late summers of
the previous years to red spruce ring chronologies. Such unusually
late warm conditions could easily be associated with the multiple
year drought of 1964-1966. As yet there is no physiological
explanation for these relationships other than perhaps the new
shoot material doesn't harden-off properly when the late summer is
too warm. In the future, if tree-ring chronologies are to be used
to track changes in disturbance regime by EMAP and the NEFHM
program; then an index of -late summer temperatures and the
distribution of wintertime temperature minima should be added to
the stress climatology-
Associative studies such as these and modeling activities
linking climate and ecosystem behavior in the northeastern United
States are anticipated in the future. Recent climate/forest
modeling activities are discussed in a special issue of Agriculture
and Forest Meteorology. The April, 1990 issue is devoted to the
Second Uppsala workshop on Modeling Forest growth Processes. A
meeting summary and review is presented by Perttu (1990).
Examples of forest modeling and associative studies for
climate change applications are discussed in Solomon et al.(1984),
Pastor and Post (1988), Joyce et al. (1990), Overpeck et al.
(1990), Fritts and Swetnam (1989) and IPCC (1990). Although a
relatively large body of research literature exists, the need for
more basic and applications research is identified by both the IGBP
and the U.S. Department of Agriculture Forest Service (IGBP, 1990;
USDA Forest Service, 1990).
81
-------
III.D. Most Recent Year (1990)
Summaries of climatological conditions in the most recent past
are primarily used to aid in the interpretation of the current
year's sampling program. An example is found in Brooks et al. (c,
in press). These products and analyses, in combination with
supplemental field observations, can be useful for the
identification of areas in need of special studies to identify
unknown sources of stress or mortality. Table 10 indicates those
disturbance regime characteristics summarized for the roost recent
sample collection period (1990). With the exception of wind and
tropical cyclone events, all identified disturbances are
characterized for 1990 in terms of their frequency, magnitude and
size.
Table 10. Elements of a 1990 New England disturbance regime
climatology.
CLIMATE VARIABLE
Tornadoes
Wind
Tropical Cyclone
Drought
Temperature
Precipitation
Growing Degree
Days
Late Spring
Freeze
Late spring
Snowfall
FREQUENCY
X
X
none reported
X
X
X
X
X
X
MAGNITUDE
X
none reported
X
X
X
X
X
X
SIZE
X
none reported
X
X
X
X
X
X
III.D.I. October 1989 through September 1990 Weather
Record warmth was reported during November 1990. This was
followed by a blizzard with sustained hurricane force winds and
gusts of 45 ms"1 in Maine during November. Winter weather impacted
Maine again during December, when the worst blizzard since 1987
struck the Presque Isle area. Many Vermont stations reported "the
coldest December on record." Ice accumulations of from 1.27 cm to
1.9 cm on trees and powerlines in Connecticut and Massachusetts
made a February ice storm the most serious since 1984.
82
-------
Severe weather was reported in May and June. Two weak
tornadoes were reported in Massachusetts during May- A small
tornado, three funnel clouds and 15 damaging wind events were
reported during June. Flooding was reported in New Hampshire and
Vermont throughout July and August. The heaviest 24-hour rainfall
in 50 years was reported in Massachusetts during August.
III.D.2. Historical Comparison
There were fewer than average tornadoes during 1990. Those
that were reported occurred earlier in the summer than usual. The
spring and summer wind events were in keeping with the past
climatology of the area, but the large number of reports associated
with a September storm outbreak was unusual. Peak wind gusts of 45
ms'1 in November, over 34 ms'1 in June and over 31 ms~' in August were
unusually strong. Figures 37-39 illustrate 1990 departures from
average temperature and precipitation conditions. Figure 30 tells
us that GDD, last spring frosts date and PDSI conditions were not
unusual over a significant portion of New England during 1990.
Mean annual 1990 New England temperatures follow the typical
south to north gradient evident in the long-term area record
(Brooks et al, c, in press). Slightly cooler than normal
conditions prevailed in southern Vermont and southeastern Maine,
with slightly warmer than average conditions elsewhere. Annual
maximum temperatures were near, or slightly cooler than, the 30-
year average (Figure 37). Annual minimum temperatures were below
the 30-year average for much of Maine and near to slightly above
average for the remainder of New England (Figure 38). This is in
keeping with the unusually cold winter conditions noted in the
weather summary.
An irregular pattern of precipitation was reported across New
England during 1990. With the exception of north-central and
southern coastal Maine, the region reported near-normal or wet
conditions. Some areas reported 25.4 to 38.1 cm of rain in excess
of their average annual totals (Figure 39).
Figure 30 indicates a fairly large portion of New England
experienced a late season (spring) snowfall event in 1990. This
event occurred on May 22 and 23 and was centered in northern Maine.
A few stations in Vermont and New Hampshire were impacted as well.
Frozen precipitation associated with this storm tended to be light,
with the exception of Squapan Dam, Maine, which reported
accumulations of 5 cm.
83
-------
Figure 37. 1990 annual maximum temperature departure from 30-year
average conditions.
DEGREES CENTIGRADE
^m -6.0 to -3.0
-3.0 to 0.0
0.0 to 3.0
3.0 to 6.0
84
-------
Figure 38
1990 annual ,ini». temperature departure fro* 30-year
average conditions.
DEGREES CENTIGRADE
j^ -g.O to -6.0
-6.0 to -3.0
-3.0 to 0.0
0.0 to 3.0
J.O to 6.0
85
-------
Figure 39. 1990 annual precipitation departure from 30-year
average conditions.
PRECIPITATION
(in centimeters)
-30.0 to -15.0
-15.0 to 0.0
0.0 to 15.0
15.0 to 30.0
30.0 to 45.0
86
-------
III.D. Summary and Comparison with Monitored Data
With only a few local exceptions, 1990 brought to a quiet
close a notably uneventful disturbance decade. The negative trend
in severe weather and tropical storm activity discussed in the
decade summary continued through 1990. No significant drought
stress was noted. Conditions generally favorable to forest growth
prevailed as is borne out by Brooks et al. (c, in press) . Isolated
cases of winter freeze damage were reported, but no locations were
given. This could coincide with extremely cold winter conditions
reported in Maine and Vermont.
Brooks et al. (c, in press) note an area of moderate to severe
crown dieback in Maine cedars, particularly in Aroostook County. No
explanation was immediately available. Although many other factors
could be involved, a cursory check of weather records for that area
suggests three possibilities. First, the Presgue Isle area of
Aroostook County reported record 24-hour snowfall in December of
1989 (76 cm) accompanied by sustained winds of 13 ms"1 and peak
gusts of 21 ms"1. Although evergreens in this region are well
adapted to such events, the early December blizzard was extreme by
any standard.
Second, Figure 30 indicates an unusual late spring snowfall
event during 1990. Further investigation shows that the event
centered on Aroostook County. The county is represented by three
precipitation stations, Caribou, Fort Kent and Sguapan Dam.
Temperature conditions were warm enough to suggest this was a warm
snow event. Records for Caribou and Fort Kent indicate this was
snow mixed with rain, but frozen precipitation accounted for only
a trace of that recorded and there was no snow accumulation.
Squapan Dam, on the other hand, reported 5 cm of accumulated frozen
precipitation from this storm which lasted from May 22 to 23 and
produced a total water equivalent (rain and melted snow or ice) of
3.1 cm. Figure 40 illustrates the locations of hexagons
represented by the Squapan Dam observation.
Finally, a severe storm passed through the area during August
of 1990 with winds in excess of 26 ms"1. This event would be a
factor only if sampling in this area was done late in the summer.
Since more detailed information concerning the location of crown
damage within the county is unavailable at this time, these three
sources of forest disturbance are only suggestions and cannot be
directly associated with the sample observations.
87
-------
Figure 40. Potential forest sampling hexagons represented by
precipitation recorded at Squapan Dam, Maine.
O O O O • •
o e> o o o o o
o o o o o o
ooooooo
ooooooo
oooooooooo
ooooo
ooooo/o
olo o o o o o
ooooo
000(0000
o o o o o o o
o o o o o o o
OOOOO
88
-------
IV. Report Summary and Future Work
Forest response research has established that climate plays an
important role in forest ecosystems, influencing both persistence
and recurrence characteristics of the landscape (Chapter I) .
Forests act as environmental integrators of both natural and
anthropogenic disturbances. New monitoring programs such as EPA
EMAP and the USDA FHM program hope to capture forest response to
changes in these stresses, but their observations cannot be
correctly interpreted without an ability to discern between natural
climate variability and signals of permanent climatological change.
Likewise, changes in ecosystem health and productivity can not be
accurately attributed to changes in societal behavior unless the
ecosystem response to natural disturbance can be better understood
and predictively modeled.
Pickett and White (1985) propose the foundation for a theory
of natural disturbance. One conclusion is that, "in order to
develop a theory of disturbance composed of unambiguous, testable
hypotheses and capable of making sound mechanistic predictions, the
relevant variables of disturbance must be established. These
include at least magnitude, frequency, size and dispersion."
Existing climatological products that are designed to appeal to a
general audience provide valuable background information. If,
however, we are to correctly interpret and model forest/climate
interactions as proposed by Pickett and White, we must be certain
to focus on the most pertinent aspects of the environment. This
includes making sure that researchers are provided with the climate
conditions being integrated by and inducing changes in the
communities under study (Chapter II).
This research addresses several critical climate impact
issues. First, while existing historical networks may be adequate
for landscape or regional analyses, local geographic influences
critical to the development of accurate predictive models (a stated
goal of the FHM) may be lost or misinterpreted by using
nonrepresentative climate observations. Ecosystem monitoring and
modeling programs that anticipate the availability of
climatological observations suitable for assessment and research
applications may be disappointed by reality. The availability and
characteristics of climatological data and information pertinent to
the research issue to be addressed should figure prominently in the
early stages of any ecological monitoring program. Too often it is
included as an afterthought. With many of today's critical issues
focusing on the presence, absence, change and detection of
climatological and anthropogenic influences on the biosphere, this
oversight could be very costly in terms of knowledge lost as well
as dollars expended.
Next, although there are usually far more climate data
available for analysis than for ecosystem status and health, there
89
-------
are serious quality control and representation problems. Some of
these issues are currently being addressed by the atmospheric
science community, but true, uniform quality control of
cooperatively observed weather conditions is a formidable task.
The ecological research community must be made as aware as possible
of particularly risky areas. Alternatives should be provided where
possible.
Finally, new data analysis technologies such as CIS offer
unique opportunities to integrate previously intractable multi-
media research issues. With increased flexibility comes new
challenges-to research problem definition. One issue illustrated
in this research is the problem of graphically representing an
inherently temporal process (climate\forest interactions and
dynamics) in a spatial setting (regional analysis). The discussion
of data biases and weighting scheme error illustrates that the best
solution to these problems is yet to be determined.
Remotely sensed and new ground-based monitoring programs
increase analysis opportunities. For example, Quattrochi and
Pelletier (1991) suggest that remote sensing technology can be used
to detect and record successional or cyclical ecosystemic changes
in response to disturbance, to identify the location and spatial
extent of stress response or disturbance events and to estimate the
magnitude of disturbance and level of stress response. To realize
this potential, there is first the need to establish the new data
in the assessment process. Even if remotely sensed data provides
improved characterization of spatial and temporal variability of
the climate, comparative relationships between historic data
sources and new sources must be established before they are
routinely incorporated into ecological assessment products.
Pickett and White conclude that the development of disturbance
theory requires an explicit statement of the parameters of systems
that can respond to disturbance. Predictions must be made in terms
of the variables of disturbance and the response parameters. The
research summarized here represents a first step in this process,
i.e., the presentation of the magnitude, frequency, size and
dispersion of known climate disturbance phenomena. The next step
is to explicitly link observations of forest status and behavior
with the unified climatological disturbance data base to identify
and more precisely define forest response parameters. This has
been done in a cursory fashion in Chapter III, but a long process
of research, integration and application remains before the role of
climate/forest interactions in the overall earth system can be
fully understood and predicted.
This point is illustrated by Appendix A, which contains an
outline of research opportunities suggested by this report. They
are not ranked but have been stratified into four categories.
These areas represent: (1) the development of new climate analyses
(climatologies), (2) fundamental program design issues, (3)
90
-------
acquisition of new data, and (4) applied climatological research.
This report illustrates the volume of research and applications
studies that can be applied to operational monitoring programs such
as EMAP and FHM. Appendix A suggests there is far more that can
and should be done to expand the use and value of climate data and
information to these programs.
91
-------
V. References
Aber, J.D., K.J. Nadelhoffer, P- Steudler and J.M. Melillo, 1989.
"Nitrogen Saturation in Northern Forest Ecosystems."
BioSciencer 39(6): 378-386.
Adams, M.B. and C. Eager (eds.), 1989. Air Pollution and Winter
Injury of Red Spruce, Proceedings of a workshop sponsored by
the Spruce-Fir Research Cooperative, Edinburgh, Scotland,
April 17-18, 1989, Northeastern Forest Experiment Station,
Amherst, MA.
Alley, W.M., 1984. "The Palmer Drought Severity Index:
Limitations and Assumptions." Journal of Climate and Applied
Meteorology, 23: 1100-1109.
AMS Council, UCAR Board of Trustees, 1988. "The changing
atmosphere - challenges and opportunities." Bulletin of the
American Meteorological Society. 69: 1434-1440.
Auerback, S.I., 1981. "Ecosystem Response to Stress: A Review of
Concepts and Approaches." Chapter 3 in Stress Effects on
Natural Ecosystems. G.W. Barrett and R. Rosenberg (eds.), John
Wiley & Sons, Chichester. pp 29-41.
Baker, W.E., 1991. "Wind Measurements Expected with the Laser
Atmospheric Wind Sounder." Preprints of the Seventh Symposium
on Meteorological Observations and Instrumentation. American
Meteorological Society, Boston, MA. pp 169-174.
Barrett, G.W., 1981. "Stress Ecology: An Integrative Approach."
Chapter 1 in Stress Effects on Natural Ecosystems. G.W.
Barrett and R. Rosenberg (eds.), John Wiley & Sons,
Chichester. pg 4.
Barrett, G.W. and R. Rosenberg, 1981. Stress Effects on Natural
Ecosystems. John Wiley & Sons, Chichester. 305pp.
Bazzaz, F.A., 1983. "Characteristics of Populations in Relation to
Disturbance in Natural and Man-Modified Ecosystems." Chapter
5.4 in Disturbance and Ecosystems; Components of Response.
H.A. Mooney and M. Godron (eds.), Springer-Verlag, Berlin.
pp259-275.
Bormann, F.H. and G.E. Likens, 1979. "Catastrophic Disturbance and
the Steady State in Northern Hardwood Forests." American
Scientist. 67: 660-669.
Botkin, D.B., 1980. "A Grandfather Clock down the Staircase:
Stability and Disturbance in Natural Ecosystems." In Forests.
Fresh Perspectives from Ecosystem Analysis. R.H. Waring (ed.),
Oregon State University Press, Corvallis, pp 1-10.
92
-------
Botkin, D.B., J.F. Janak and J.R. Wallis, 1972. "Some ecological
consequences of a computer model of forest growth." Journal
of Ecology. 60:948-972.
Brooks, R.T., M. Miller-Weeks and W. Burkman, in press a. New
England Forest Health Monitoring Program: 1990 Executive
Summary- USDA Forest Service, Northeastern Forest Experiment
Station, Radnor, PA.
Brooks, R.T., D. Dickson, W. Burkman, I. Millers, M. Miller-Weeks,
E. Cooter and L. Smith, in press c. New England Forest Health
Monitoring Program: 1990 Annual Report. USDA Forest Service
Northeastern Forest Experiment Station, Radnor, PA.
Brooks, R.T., T. Frieswyk, D. Griffith, E. Cooter and L. Smith, in
press b. New England Forests: A Baseline for the New England
Forest Health Monitoring Program. USDA Forest Service
Northeastern Forest Experiment Station, Radnor, PA.
Canham, C.D. and O.L. Loucks, 1984. "Catastrophic Windthrow in the
Presettlement Forests of Wisconsin." Ecology. 65(3):803-809.
Cooter, E.J., 1982. An Oklahoma Soil Moisture Climatology. Volume
6 of Operational Weather Modification, Oklahoma Climatological
Survey, Norman, Oklahoma, 184 pp.
Eder, B.K., J.M. Davis and J.F. Monahan, 1987. "Spatial and
Temporal Analysis of the Palmer Drought Severity Index over
the South-Eastern United States." Journal of Climatology,
7:31-56.
Emanuel, K.A., 1988. "The Maximum Intensity of Hurricanes."
Journal of the Atmospheric Sciences. 45(7):1143-1155.
Emanuel, K.A., 1987. "The dependence of hurricane intensity on
climate." Nature. 326:483-485.
Ferris, C., 1980. "The Forest System." Volume 2, Chapter 9 of An
Ecological Characterization of Coastal Maine (North and East
of Cape Elizabeth), S. I. Fefer and P.A. Schettig, eds.,
Department of the Interior, U.S. Fish and Wildlife Northeast
Region publication, FWS/OBS-80/29, Newton Corner, MA. 39 pp.
Forman, R.T.T., and M. Godron, 1986. Landscape Ecology. John
Wiley and Sons, New York. 619pp.
Franklin, J.F., H.H. Shugart and M.E. Harmon, 1987. "Tree Death as
an Ecological Process." BioScience. 37(8): 550-556.
93
-------
Fritts, H.C. and T.W. Swetnam, 1989. "Dendroecology: A Tool for
Evaluating Variations in Past and Present Forest
Environments." In Advances in Ecological Research, Volume 19,
M. Begon, A.H. Fitter, E.D. Ford and A. MacFadyen (eds.),
Academic Press, London, pp 111-188.
Fujita, T.T., 1978. Manual of downburst identification. Satellite
and Mesometeorology Research Project Research Paper No. 156,
The University of Chicago, Chicago, Illinois.
Fujita, T.T., 1989. "Mesoscale Classifications: Their History and
Their Application to Forecasting." Chapter 2 of Mesoscale
Meteorology and Forecasting. P.S. Ray (ed.), American
Meteorological Society, Boston. ppl8-35.
Fujita, T.T., 1981. "Tornadoes and Downbursts in the Context of
Generalized Planetary Scales." Journal of the Atmospheric
Sciences. 38(8):1511-1534.
Gagne, J.A., Coulson, R.N., Foltz, J.L., Wagner, T.L. and L.J.
Edson, L.J., 1980. "Attack and Survival of Dendroctonus
frontalis in relation to weather during three years in east
Texas." Environmental Entomology. 9:22-29.
Gedzelman, S.D. and E. Lewis, 1990. "Warm Snowstorms, a
forecaster's Dilemma". Weatherwise. 43(5): 265-270.
Gerardi, M.H. and J.K. Grimm, 1979. The History. Biology. Damage
and Control of the Gypsy Moth Porthetria Dispar (L.).
Fairleigh Dickinson University Press, Rutherford, pp 101-103.
Gleick, J., 1987. Chaos. Viking Press, NY. 352pp.
Godron, M. and R.T.T. Forman, 1983. "Landscape Modification and
Changing Ecological Characteristics." Chapter 2.1 of
Disturbance and Ecosystems: Components of Response. H.A.
Mooney and M. Godron (eds.), Springer-Verlag, Berlin, pp 12-
28.
Gosselink, J.G., L.C. Lee and T.A. Muir, 1990. "The Regulation and
Management of Bottomland Hardwood Forest Wetlands:
Implications of the EPA-Sponsored Workshops." Chapter 19 in
Ecological Processes and Cumulative Impacts: Illustrated by
Bottomland Hardwood Wetland Ecosystems. J.G. Gosselink, L.C.
Lee and T.A. Muir, eds., Lewis Publishers, Chelsea, Michigan.
p 646.
Graham, R.L., M.G. Turner and V.H. Dale, 1990. "How Increasing C02
and Climate Change Affect Forests." BioScience. 40(8): 575-
587.
94
-------
Hayden, B.P., 1981. "Secular Variation in Atlantic Coast
Extratropical Cyclones. " Monthly Weather Review. 109:159-167.
Henry, J.D. and Swan, J.M.A., 1974. "Reconstructing forest history
from live and dead plant material - an approach to the study
of forest succession in southwest New Hampshire." Ecology
55:772-783.
Holland, M.M., 1988. "SCOPE/MAB technical consultations on
landscape boundaries: report of a SCOPE/MAB workshop on
ecotones." Biology International. Special Issue, 17: 47-106.
Hunsaker, C.T. and D.E. Carpenter (eds.), 1990. Ecological
Indicators for the Environmental Monitoring and Assessment
Program. USEPA/AREAL, Research Triangle Park, North Carolina.
Huschke, R.E. (ed.), 1989. Glossary of Meteorology. American
Meteorological Society, Boston. 638pp.
Idso, S.B., R.C. Balling, Jr., and R.S. Cerveny, 1990. "Carbon
Dioxide and Hurricanes: Implications of Northern Hemispheric
Warming for Atlantic/Caribbean Storms." Meteorology and
Atmospheric Physics. 42: 259-263.
International Geosphere-Biosphere Programme, 1990. IGBP: A Study of
Global Change. Report no. 12, pp 6.1-1 to 6.1-30.
Jeffers, J.N., 1988. "Statistical and Mathematical Approaches to
Issues of Scales in Ecology-" Chapter 4 in Scales and Global
Change: Spatial and Temporal Variability in Biospheric and
Geospheric Processes. T. Rosswall, R.G. Woodmansee and P.G.
Risser (eds.), John Wiley & Sons, Chichester. pg 54.
Johnson, A.H., E.R. Cook, and T.G. Siccama, 1988. "Climate and red
spruce growth and decline in the northern Appalachians." In
Proceedings of the National Academy of Science USA. 85: 5369-
5373.
Jones, D.A. and Kiniry, J.R., 1986. CERES-Maize; A Simulation
Model of Maize Growth and Development. Texas A&M University
Press, College Station, TX. 194pp.
Karl, T.R. and R.G. Quayle, 1988. "Climate Change in Fact and in
Theory: Are we Collecting the Facts?" Climatic Change. 13:
5-17.
Karl, T.R. , L.K. Metcalf, M.L. Nicodemus and R.G. Quayle, 1983.
Statewide Average Climatic History. Historical Climatology
Series No. 6-1, NOAA/NESDIS/NCDC.
95
-------
Karl, T.R. and C.N. Williams Jr., 1987. "An Approach to Adjusting
Climatological Time Series for Discontinuous Inhomogeneities. "
Journal of Climate and Applied Meteorology. 1744-1763.
Karl, T.R., 1983. "Some Spatial Characteristics of Drought
Duration in the United States." Journal of Climate and
Applied Meteorology. 22: 1356-1366.
Kozlowski, T.T., 1985. "Tree Growth in Response to Environmental
Stresses." J. of Arboriculture. 11(4):97-lll.
Michaels, P.J. and B.P. Hayden, 1987. "Modeling the Climate
Dynamics of Tree Death." BioScience. 37(8):603-610.
Michaels, P.J., P.J. Stenger, and D.E. Sappington, 1986.
"Modelling changes in the epidemic range of Southern Pine
Beetle using temperature and objective moisture indices."
Theoretical and Applied Climatology. 37: 39-50.
National Weather Service, 1989. Cooperative Station Observations.
National Weather Service Observing Handbook No. 2, U.S. GPO,
Washington, D.C. 83 pp.
Neumann, C.J., B.R. Jarvinen, A.C. Pike, and J.D. Elms, 1990.
Tropical Cyclones of the North Atlantic Ocean. 1871-1986 (with
storm track maps updated through 1989), NOAA/NESDIS/NCDC,
Asheville, NC. 186 pp.
Norton, D.J. and Slonecker, E.T., 1990. The Ecological Geography
of EMAP- Geolnfo Systems. 1:32-45.
Nuttonson, M.Y., 1955. Wheat-Climate Relationships and the Use of
Phenology in Ascertaining the Thermal and Photo-Thermal
Requirements of Wheat. American Institute of Crop Ecology,
Washington, D.C.
Overpeck, J.T., D. Rind and R. Goldberg, 1990. "Climate Induced
Changes in Forest Disturbance and Vegetation." Nature.
343(4):51-53.
Palmer, W.C., 1965. Meteorological Drought. Weather Bureau
Research Paper No. 45. Washington, D.C. 58 pp.
Peer, R.L., 1990. "An Overview of Climate Information Needs for
Ecological Effects Models." Contract No. 68-02-4288,
USEPA/ORD/AREAL, Research Triangle Park, NC. 52 pp.
Pickett, S.T.A. and White, P.S. (eds.), 1985. The Ecology of
Natural Disturbance and Patch Dynamics. Academic Press, Inc.,
Orlando, 472 pp.
96
-------
Puckett, L.J., 1981. Dendroclimatic estimates of a drought index
for northern Virginia. U.S. Geological Survey Water-Supply
Paper 2080. 39 pp.
Quattrochi, D.A. and R.E. Pelletier, 1991. "Remote Sensing for
Analysis of Landscapes: An Introduction." Chapter 3 of
Quantitative Methods in Landscape Ecology, M.G. Turner and
R.H. Gardner (eds.), Springer-Verlag, New York. pp 51-76.
Reiners, W.A., 1983. "Disturbance and Basic Properties of
Ecosystem Energetics." Chapter 3.3 in Disturbance and
Ecosystems: Components of Response, H.A. Mooney and M.
Godron (eds.), Springer-Verlag, Berlin. pp 83-98.
Robinson, D.A., 1988. "Construction of a United States Historical
Snow Data Base." In Proceedings of the 1988 Eastern Snow
Conference. pp 50-59.
Rosswall, T., R.G. Woodmansee and P.G. Risser (eds.), 1988. Scales
and Global Change: Spatial and Temporal Variability in
Biospheric and Geospheric Processes. SCOPE (series) 35, John
Wiley & Sons, Chichester. 355 pp.
Runkle, J.R., 1985. Disturbance Regimes in Temperate Forests. In
The Ecology of Natural Disturbance and Patch Dynamics. S.T.A.
Pickett and P.S. White (eds.), Academic Press, Inc., Orlando,
pp 17-33.
Sadowski, A., 1975. Bulletin concerning Technical Procedures
Bulletin No. 130:CROP MOISTURE INDEX. NOAA/NWS bulletin to
recipients of technical procedures bulletin series
(mimeographed).
Shugart, H.H., 1984. A Theory of Forest Dynamics: The Ecological
Implications of Forest Succession Models. Springer-Verlag,
New York. 278pp.
Shugart, H.H., P.J. Michaels, T.M. Smith, D.A. Weinstein and E.B.
Rastetter, 1988. "Simulation Models of Forest Succession."
Chapter .8 in Scales and Global Change: Spatial and Temporal
Variability in Biospheric and Geospheric Processes.
R.Rosswall, R.G. Woodmansee and P.G. Risser, eds., John Wiley
& Sons, Chichester. pp!25-152.
Smith, W.H., 1990. "The Health of North American Forests: Stress
and Risk Assessment." Journal of Forestryf 88: 32-35.
97
-------
Solomon, A.M., M.L. Tharp, D.C. West, G.E. Taylor, J.W. Webb and
J.L. Trimble, 1984. "Response of Unmanaged Forests to C02 -
Induced Climate Change: Available Information, Initial Tests
and Data Requirements." Prepared for DOE under Contract, No.
W-7405-ENG-26, DOE/NBB-0053, Oak Ridge National Laboratory,
Oak Ridge TN. 93 pp.
Stephens, G.R., 1981. Defoliation and mortality in Connecticut
Forests. Bull. 796, Connecticut Agricultural Experiment
Station, New Haven. 13pp.
United States Department of Commerce, 1991. "Storm Data",
USDOC/NOAA/NCDC, National Climatic Data Center, Asheville,
North Carolina.
United States Department of Agriculture Forest Service, no date.
"New England Forest Health Monitoring Program: Project Work
Plan." Compiled by M. Miller-Weeks and D.I. Gagnon, FPM, USDA
Forest Service, Durham, NH.
Weinstein, D.A. and H.H. Shugart, 1983. "Ecological Modeling of
Landscape Dynamics." Chapter 2.2 in Disturbance and
Ecosystems: Components of Response. H.A. Mooney and M. Godron
(eds.), Springer-Verlag, Berlin. pp29-45.
White, P.S. and S.T.A. Pickett, 1985. "Natural Disturbance and
Patch Dynamics: An Introduction." Chapter 1 in The Ecology
of Natural Disturbance and Patch Dynamics. S.T.A. Pickett and
P.S. White (eds.), Academic Press, Inc, Orlando. pp3-9.
Wissmar, R.C. and F.J. Swanson, 1990. "Landscape Disturbances and
Lotic Ecotones." Chapter 5 in The Ecology and Management of
Aquatic-Terrestrial Ecotones. R.J. Naiman and H. Decamps
(eds.), Parthenon Publishing Group Inc., Park Ridge, New
Jersey. pp 65-89.
Woodward, F.I., 1987. Climate and Plant Distribution. Cambridge
University Press, Cambridge. 174pp.
Zahner, R., Saucier, J.R. and Myers, R.K., 1989. "Tree-ring model
interprets growth decline in natural stands of loblolly pine
in the southeastern United States." Canadian Journal of
Forest Research. 19:612-621.
Zonneveld, I.S. and R.T.T. Forman, 1990. Changing Landscapes: An
Ecological Perspective. Springer-Verlag, New York. 281pp.
98
-------
APPENDIX A
FUTURE WORK
Additional Climatologies
1. Ozone or other pollution episodes could, over time, impact
forest growth and productivity. Ozone climatologies are
needed for use with pollution indicator species monitoring
data.
2. Climatologies of other temperature related disturbances:
a. heat waves (evaporative stress and heat shock)
b. frequency, intensity and duration of cold winter
temperatures (winterkill and tree-ring chronologies)
c. warm August temperatures (tree-ring chronologies)
3. Climatologies of other precipitation variables:
a. Fall precipitation greater than or equal to 1.25 cm
(correlated with growth)
b. Spring precipitation greater than or equal to 1.25 cm
(correlated with growth)
4. Statistical definition of drought areas not necessarily based
on climate divisions. This would help to target special
forest health studies and is of interest to tree-ring
scientists.
5. Climatology of wet-spell or flooding events. The extreme
positive end of the PDSI scale could be considered as an
indicator of wide-spread soil saturation. It would not be
adequate for local or short-term flooding events. If this is
critical, alternative methods of estimation will need to be
identified.
6. Climatology of spring or fall freeze duration (within and
between 24-hour periods) and pre-freeze conditions (winter
hardiness).
7. A fire weather climatology. Because of the influence of
forest management, this is a difficult issue, but is critical
for climate trend/climate change assessment research. Much
work has been done by USDA Forest Service scientists already,
so this may only involve determining how best to apply the
information in an EMAP/FHM setting.
8. Expansion of the spring hard freeze and snowfall climatologies
to late fall events.
99
-------
Program Design
1. Investigation of the implications of the proximity of weather
data for FHM data analysis: i.e., are the stations we have
selected adequate in time and space? No comparison of climate
site descriptors to sample site descriptors has been made.
Selection criteria is based only on distance from sampling
site and quality of record. If interpolation is needed, which
methods are the most appropriate? This may vary with the
climate variable to be interpolated.
2. Examination and resolution of any differences between EMAP
monitoring and climate CIS designs.
3. Exploration of potential interactions between climate and air
pollution and deposition activities.
Database Development
1. Extraction from data record of "modeling" of snow and ice
disturbance events. A model is needed because of inaccuracies
in the snow/ice record for most NWS cooperative observer
locations. Ice storms in particular can result in substantial
areas of crown damage, defoliation and mortality.
2. Acquisition of digitized tropical cyclone paths. Since these
paths are remotely sensed, this information may help to fill-
in severe storm observations in sparsely populated areas.
Since the resolution of the paths is known, error bars can be
placed around the central path to establish potential impact
area estimates.
3. Calculation of PDSI by station. Once this database has been
developed, it can be manipulated in a variety of ways,
including support for tree-ring chronology studies and the
spatial drought climatology described above.
4. Acquisition of Canadian weather data to supplement data-sparse
northern border areas. This has been started, but has not
progressed very far.
5. Acquisition of a lightning strike database. Data are
available from networks operated by Bureau of Land Management
(BLM), National Severe Storms Laboratory (NSSL) and the State
University of New York at Albany (SUNYA).
6. Acquisition of the Robinson quality controlled snowfall
database from NCDC. These data would be used to better define
heavy (wet) snowfall events and to then generate more
meaningful return period climatologies.
100
-------
Applied Research
1. Investigate the application or integration of remotely sensed
weather data to quantitative summaries used by landscape
ecologists that also use remotely sensed land surface data.
2. Explicitly examine the way in which disturbance climatologies
have been used and may be used in tree-ring chronology
studies. This would be done specifically with trend detection
and source analysis in mind.
3. Develop and complete definition of disturbance phenomena in
three-dimensional (3-D) space (size, intensity and duration)
for given disturbance frequencies. For instance, what does a
100 year drought look like in terms of intensity, size and
impact area and duration (where area is defined by percent of
total area involved)? Perform an historical study to
determine spatial and temporal variability of the 3-D space.
Investigate ways of detecting trends or changes in the
character of the 3-D space. Develop a 3-D time series such as
Figure 30 summarizing regional characteristics. This study
would also relate to spatial pattern analysis of drought
described earlier.
4. Explore the use of extreme value distributions to estimate
return periods beyond the cooperative (30-year) data record.
This has been done in the literature for temperature and
precipitation variables, but not necessarily for all those
identified as critical to forest health.
5. Explore the development of interactive/multiple disturbance
climatologies such as moisture stress and Southern Pine Beetle
infestation and damage severity or similar relationships for
Gypsy moth in the northeast.
101
-------
APPENDIX B
EMAP BIBLIOGRAPHIC RESOURCE LIST
compiled 11/90
Anderson, Eric A. NOAA-ARS Cooperative Snow Research Project:
Watershed Hvdroclimatology and Data for Water Years 1960-1974.
Washington: U.S. Government Printing Office, 1977. *
Atlas of Monthly and Seasonal Temperature Departures from the Long-
Term Mean (1895-1983) for the Contiguous United States: Fall.
Asheville, NC: NOAA National Climatic Data Center, 1984.
Air Weather Service Climatic Briefs: North America. Scott Air Force
Base, IL: Air Force Environmental Technical Applications
Center, 1984. *
Baron, William R. "Frost-free Record Reconstruction for Eastern
Massachusetts". Journal of Climate and Applied Meteorology.
23, no. 2 (Feb. 1984), 317-319. *
Baron, William, et al. Loner-time Series Temperature and
Precipitation Records of Maine. 1808-1978. Orono, ME:
University of Maine at Orono, 1980.
Bates, R.E. Climate at CRREL (Cold Regions Research & Engineering
Laboratory) Hanover. New Hampshire. Hanover, NH: United States
Army Corps of Engineers, Cold Regions Research and Engineering
Lab, 1984. *
Bates, Roy E. Snow Cover and Meteroloqv at Allaqash. Maine. 1977-
1980. Hanover, NH: United States Army Corps of Engineers, Cold
Regions Research and Engineering Laboratory, 1983. *
Bates, Roy E. Winter Thermal Structure. Ice Conditions, and Climate
of Lake Champlain. Hanover, NH: United States Army Corps of
Engineers, Cold Regions Research and Engineering Laboratory,
1980. *
Bates, Roy E. and Brown, Mary-Lynn. Lake Champlain Ice Formation
and Ice Free Dates and Predictions from Meteorological
Indicators. Hanover, NH: United States Army Corps of
Engineers, Cold Regions Research and Engineering Lab, 1979.*
Bilello, M.A. Review of Snow Conditions and Winter Climate Near
Burlington and Underhill. Vermont. Hanover, NH: United States
Army Cold Regions Research and Engineering Laboratory, 1980.*
102
-------
Bogart, Dean B. Floods of August to October 1955. New England to
North Carolina. Washington: United States Government Printing
Office, 1960.
Brumbach, Joseph J. The Climate of Connecticut, np. 1965. Series:
State Geological and Natural History Survey of Connecticut
Bulletin 99.
Bulletin of the New England Weather Service. Boston, Mass: New
England Weather Service, 1892-.
Burlington IAP VT. Revised Uniform Summary of Surface Weather
Observations. Parts A-F. Scott Air Force Base, IL: Air Force
Environmental Technical Applications Center, 1987. *
Brunswick Naval Air Station. Maine. Revised Uniform Summary of
Surface Weather Observations. Parts A-F. Scott Airforce Base,
IL: Air Force Environmental Technical Applications Center,
1987. *
Climatolocrical Data. New England. Asheville, NC: National Oceanic
and Atmospheric Administration. Environmental Data and
Information Service, National Climatic Center, 1979.
Dana, Terry and Easter, R.C. "Statistical Summary and Analyses of
Event Precipitation Chemistry From the MAP3S Network, 1976-
1983." Atmospheric Environment. 21, no. 1 (1987), 113-128. *
Dethier, Bernard E. The Climate of the Northeast Precipitation
Probabilities. Ithaca, NY: New York State College of
Agriculture, 1965.
Dow Air Force Base. Banqor. Maine. Revised Uniform Summary of
Surface Weather Observations. Parts A-F. Scott Airforce Base,
IL: Air Force Environmental Technical Applications Center,
1987. *
Fay, Richard. New England Coastal Fog. New York, NY: United States
Weather Bureau Eastern Region, 1967.
Federer, C.A. Annual Cycles of Soil and Water Temperatures at
Hubbard Brook. Upper Darby, PA: United States Forest Service.
Northeastern Forest Experiment Station, 1973. *
Fobes, Charles B. Climatic Divisons of Maine. Orono, ME: University
Press, 1946.
Fort Devens Army Air Field. Massachusetts Limited Surface
Observations Climatic Summary. Parts A-F. Scott Air Force
Base, IL: Air Force Environmental Technical Applications
Center, 1985. *
103
-------
Glidden, D.E. "Great Windstorm of 2 April 1973 on Cannon Mountain,
New Hampshire." Weatherwise. 27, no. 4 (Aug. 1974), 156-161.*
Gosslee, David G. Weekly Precipitation and Temperature in
Connecticut, storrs, CT: University of Connecticut, College of
Agriculture Experiment Station, 1961.
Govoni, John W. Comparison of Winter Climatic Data for Three New
Hampshire Sites. Hanover, NH: United States Army Corps of
Engineers, Cold Regions Reasearch and Engineering Laboratory,
1986.
Gow, Anthony J. and Govoni, John W. Ice Growth on Post Pond (NH) .
1973-1982. Hanover, NH: United States Army Corps of
engineers, Cold Regions Research and Engineering Laboratory,
1983. *
Hamburg, Steven P. et al. "Historical Decline of Red Spruce
Populations and Climatic Warming." Nature. 331, No. 6155 (Feb.
4, 1988), 428-431. *
Hedin, Lars O. et al. "Decrease in Precipitation Acidity Resulting
from Decreasing SO SUB 4 Super 2 Super- Concentration."
Nature. 325, no. 6101 (Jan. 15, 1987), 244-246.
Hunter, Bruce W. and Meade, Daniel B. Precipitation in Connecticut.
1951-1980. Hartford, CT: Natural Resources Center, Connecticut
Dept. of Environmental Protection, 1983.
Hurley, Charles W. "Recent Temperature and Snowfall Trends." News
Bulletin (Mount Washington Observatory, Gorham, NH), 10, no.
3 (Sept. 1969), 13-15. *
Impact of Tropical Storm "Acmes" on Chesapeake Bay. Vol.1. Summary.
Prepared by Cheseapeake Research Consortium Inc. Baltimore,
MD: United States Army Corps of Engineers, Baltimore District,
1975. (vol. 2 also avail.) *
Karl, Thomas R. Atlas of Monthly and Seasonal Precipitation
Departures from Normal (1895-1985) for the Contiguous United
States : Fall. Asheville, NC: NOAA Natioanl Climatic Data
Center, 1985.
Karl, Thomas R. and Knight, Richard W. Atlas of Monthly Palmer
Hydroloaical Drought Indices (1895-1930) for the Contiguous
United States. Asheville, NC: NOAA. National Climatic Data
Center, [1985].
Karl, Thomas R. and Knight, Richard W. Atlas of Monthly Palmer
Moisture Anomaly Indices (1895-1930) for the Contiguous United
States. Asheville, NC: NOAA. National Climatic Data Center,
1985
104
-------
Karl, Thomas R. et al. Probabilities and Precipitation Required to
End/Ameliorate Droughts. Asheville, NC: NOAA. National
Climatic Data Center, 1986.
Karl, Thomas R. Statewide Average Climatic History; Connecticut.
1895-1982. Asheville, NC: NOAA. National Climatic Data Center,
1983. *
Karl, Thomas R. Statewide Average Climatic History: Maine. 1895-
1982. Asheville, NC: NOAA. National Climatic Data Center,
1983. *
Karl, Thomas R. Statewide Average Climatic History; Massachusetts.
1895-1982. Asheville, NC: NOAA. National Climatic Data Center,
1983. *
Karl, Thomas R. Statewide Average Climatic History: New Hampshire.
1895-1982. Asheville, NC: NOAA. National climatic Data Center,
1983. *
Karl, Thomas R. Statewide Average Climatic History: Vermont. 1895-
1982. Asheville, NC: NOAA. NAtional Climatic Data Center,
1983. *
Krenz, Maria E. et al. Institutional Directory. Climate-related
Impacts Network. Boulder, CO: National Center for Atmospheric
Research, Environmental and Societal Impacts Group, 1989.
Landin, Michael G. and Bosart, Lance F. "Diurnal Variability of
Precipitation in the Northeastern United States." Monthly
Weather Review. 113, no. 6 (June 1985), 989-1014. *
Lautzenheiser, Robert E. Climatological Summary; Greenville. Maine.
Boston, MA: National Weather Service. Weather Service Office,
1971. * (NOAA report no.71121615).
Lautzenheiser, Robert E. Climatological Summary: Lewiston. Maine.
Boston, MA: National Weather Service. Weather Service Office,
1971. * (NOAA report no.71121616).
Lautzenheiser, Robert E. Snowfall. Snowfall Frequencies, and Snow
Cover Data for New England. Silver Spring, MD: U.S. Department
of Commerce , Environmental Science Services Administration,
Environmental Data Service, 1969.
Lee, Richard. "Lattitude, Elevation, and Mean Temperature in the
Northeast." Professional Geographer. 21, no. 2 (July 1969),
227-231.
Local Climatological Data. Boston. Massachusetts. Monthly Summary.
Asheville, NC: NOAA. National Climatic Data Center, monthly.
105
-------
Loring Air Force Base. Maine. Revised Uniform Summary of the
Surface Weather Observations. Parts A-F. Scott Air Force Base,
IL: Air Force Environmental Technical Applications Center,
1987. *
Miller, J.F. et al. Probable Maximum Precipitation for the Upper
Deerfield River Drainage Massachusetts/Vermont. Springfield,
VA: NOAA. National Weather Service. Office of Hydrology, Water
Management Information Division, 1984.
Mohnen, V.A. Exposure of Forests to Air Pollutants. Clouds.
Precipitation and Climatic Variables. Research Triangle Park,
NC: U.S. Environmental Protection Agency, Atmospheric Research
and Exposure Assesment Laboratory, (in cooperation with State
University of NY at Albany), 1989. * (EPA report no. EPA-R-
813934; EPA/600/3-89/003).
Monthly Normals of Temperature. Precipitation, and Heating and
Cooling Degree Days 1951-1980. Asheville, NC: NOAA. National
Climatic Data Center, 1982.
Morrill, R.A. et al. Maine Coastal Storm and Flood of February 2,
1976: Report Prepared Jointly by the U.S. Geological Survey
and the National Oceanic and Atmospheric Administration.
Washington: U.S. Government Printing Office, 1978.
North American Droughts. Boulder, CO: Westview Press, 1978.
Otis Air National Gaurd Base. Massachusetts. Revised Uniform
Summary of Surface Weather Observations. Parts A-F. Scott Air
Force Base, IL: Air Force Environmental Technical Applications
Center, 1987. *
Palley, Patricia A. and Miller, David R. "Climate Data Sources in
Connecticut." (Storrs, Connecticut University) Acrricultual
Experiment Station Bulletin. 461, 1982. *
Pannuto, Robert J. "Review of Mt. Washington Observation in
Relation to Winter Weather at Boston." News Bulletin (Mount
Washington Observatory, Gorham, NH), 16, no.l (March 1975),
17-21.
Pearson, F.J and Fisher, D.W. Chemical Composition of Atmospheric
Precipitation in the Northeastern United States. Washington:
U.S. Government Printing Office, 1971. (Sudoc no. 119.3:1535-
P).
Proceedings. Sertlina Forest. NY; Conference on the Drought in
•hhe Northeastern United States. 1967. Pub. NY: New York
University Press, 1968.
106
-------
Rao, G.A. Drought Probability Maps. Geneva: World Meteorological
Organization, 1987.
Revised Uniform Summary of Surface Weather Observations. Pease Air
Force Base. New Hampshire. Scott Air Force Base, IL: Air Force
Environmental Technical Applications Center, 1985.
Richardson, William S. et al. Tide Climatology for Boston.
Massachusetts. Boston, MA: NOAA National Weather Service,
Forecast Office, 1982. *
Sanders, Frederick and Bosart, Lance F. "Mesoscale Structure in the
Megalopolitan Snowstorm, 11-12 February 1982. Pt. 2, Doppler
Radar Study of the New England Snowband." Journal of the
Atmospheric Sciences. 42, no. 13 (July 1985), 1398-1407. *
Shenton, Edward H. and Horton, Donald B. Literature Review of the
Marine Environment Data for Eastport. Maine, rev. ed.
Portland: ME: Research Institute of the Gulf of Maine, 1973.
Significant Snowstorms. Northeastern United States. Bedford, MA:
Weather Services Corp., 1978.
Silver, Andrew L. and Brown, Wendell S. Great Bay Estuarine Field
Program; 1975 Data Report. Pt. 2. Temperature. Salinity and
Density. Durham, NH: University of New Hampshire, Durham;
Dept. of Earth Sciences, 1979. *
Stensland, Gary J. Precipitation Chemistry Trends in the
Northeastern United States. In Proceedings of the 12th
Rochester International Conference on Environmental Toxicity.
NY: Plenum Press, 1980. *
Suckling, Phillip W. and Kimsey, Mary B. "Seasonal Snowfall in
Northern New England: Recent Trends and Variability."
Climatological Bulletin. 22, no. 2 (Aug. 1988), 14-25. *
Summary of Meteorological Observations. Surface (SMOS). Brunswick.
Maine. Asheville, NC: Naval Oceanography Command Detachment,
1988. *
Summary of Meteorological Observations. Surface (SMOS) for South
Weymouth. Massachusetts. Asheville, NC: Naval Oceanography
Command Detachment, 1988. *
Summary of Meteorological Observations. Surface (SMOS). South
Weymouth. Massachusetts. Asheville, NC: Naval Oceanography
Command Detachment, 1984. *
107
-------
Summary of Synoptic Meteorological Observatons; North American
Coastal Marine Areas -Revised. Atlantic and Gulf Coasts.
Volume 2. Area 8 - St. Lawrence River. Area 9 - Gulf of St.
Lawrence. Area 10 -Cape Breton Island S.E.. Area 11 - Halifax.
Area 12 - Boston. Area 13 - Quonset Point. Area 14 - New York.
Washington, DC: Naval Weather Service Command, 1975.
Surface Observation Climatic Summaries (SOCS) for Hanscom Air Force
Base. Bedford. Massachusetts. Scott Air Force Base, IL: Air
Force Environmental Technical Applications Center, 1989.
Thorp, John M. and Scott, Bryan C. "Preliminary Calculations of
Average Storm Duration and Seasonal Precipitation Rates for
the Northeast Sector of the United States." Atmospheric
Chemistry. 16, no. 7 (1982), 1763-1774. *
U.S. Drought 1988; a Climate Assessment. Rockville, MD: U.S. Dept.
of Commerce. National Oceanic and Atmospheric Administration,
[1988] .
United States Army- Corps of Engineers. New York District. Flood
Plain Information. Dog River. Northfield. Vermont. New York:
The Corps, 1975.
United States Environmental Data Service. Weather Atlas of the
United States. Originally titled: Climatic Atlas of the United
States. Detroit: Gayle Research Co., 1975.
United States National Oceanic and Atmospheric Administration.
Climatic Atlas of the United States. Asheville, NC: NOAA
Natioanl Climatic Center, 1977.
United States Weather Bureau. Climates of the States. 1959-1969.
Washington: U.S. Dept. of Commerce, Weather Bureau, 1969.
Westover Air Force Base. Chicopee Falls. Massachusetts Revised
Uniform Summary of Surface Weather Observations. Parts A-F.
Scott Air Force Base, IL: Air Force Environmental Technical
Application Center, 1981. *
Westover Air Force Base. Massachusetts Limited Surface Observations
Climatic Summary. Parts A. C-F. Scott Air Force Base, IL: Air
Force Environmental Technical Applications Center, 1986. *
Wind. Ceiling, and Visibility Data at Selected Airports. Vol.l. New
England Region. United States Federal Aviation Administration,
Dept. of Transportation, 1980. n.p.
108
-------
Note: * = abstract available on request
Climate Librarian
U.S. Environmental Protection Agency
Mail Drop MD 80
Research Triangle Park, NC 27709
FTS 629-0912
or
919-541-0912
109
------- |