United States
Environmental Protection
Agency
Atmospheric Research and
Exposure Assessment Laboratory
Research Triangle Park NC 27711
Research and Development
ฉEPA Project Summary
EPA/600/S3-90/026 Aug. 1990
Strategies for the Development of
Climate Scenarios for Impact
Assessment: Phase 1 Final
Report
Peter J. Robinson and Peter L. Finkelstein
In order to create a strategy for the
development of climate scenarios for
use in impact assessment, potential
techniques of development were re-
viewed and the information needs of
potential users assessed. Available
techniques were assessed through
literature reviews and consultations
with scenario development experts.
Techniques were divided into ten
modules, groups of techniques with
similar methodologies, input require-
ments and output formats. Three
modules involve approaches which
model atmospheric processes, four
concern analysis of past climate
records, and three concentrate on
methods for linking the other two.
Each module can provide a scientifi-
cally well-founded piece of needed
information, and a series of modules
used together will produce-the sce-
nario. User needs were assessed in
consultation with selected individuals
who had experience in the use of
scenarios. The major needs were
revealed to be for general descriptive
statistics of the major climatic ele-
ments, for information about climatic
anomalies, notably drought, for statis-
tics on the frequency and probability
of events exceeding particular
thresholds, and for general informa-
tion about stormlness. The results of
the two sets of assessments are
combined to provide a scenario de-
velopment strategy. An iterative ap-
proach is recommended. Project
areas Incorporating both scenario
development efforts and fundamental
research are identified for the first
three iterations.
This Project Summary was
developed by EPA's Atmospheric
Research and Exposure Assessment
Laboratory, Research Triangle Park,
NC, to announce key findings of the
research project that is fully docu-
mented In a separate report of the
same title (see Project Report order-
ing information at back).
Introduction
The overall mission of the U.S.
Environmental Protection Agency (EPA)
Global Climate Change Research Plan is
the assessment of the potential impacts
of climate change on the environment,
with particular emphasis on impacts in
ecology, water resources, and air quality.
A key element within this mission is the
development of climate scenarios
oriented toward the needs of those
assessing these impacts. The goal for
the current report is to provide recom-
mendations and priorities for the devel-
opment of techniques to produce impact
oriented scenarios.
-------
To meet this goal, the specific objec-
tives are as follows: a) Identify all tech-
niques potentially available for the devel-
opment of scenarios; b) Identify user
needs in areas where EPA Is responsible
for Impact assessment; o) Assess appro-
priate techniques for Impact-oriented
scenario development, Including
assessment of current knowledge, level
of effort needed, speed of potential
development, needs of users, and suit-
ability for transformation into routine
scenario production techniques; d)
Provide recommendations to EPA for the
priorities for research, development, and
production of impact-oriented climate
scenarios.
A scenario Is defined as: A suite of
possible future climates, developed by
using sound scientific principles, each
being Internally consistent, but none
having a specific probability of occur-
rence attached.
This general definition applies to all
elements of climate for any time in the
future. However, to meet the EPA mis-
sion there are several requirements of
scenarios:
They must be tailored to user needs.
Any scenarios produced must
emphasize those climatic elements,
time scales, space scales, and geo-
graphic regions of prime concern to
those undertaking the impact
assessment. In practical but general
terms, regional scenarios, covering a
specific geographical region, must be
produced.
They must be scientifically sound.
The current state of the science is
such that many assumptions must be
Included when any projection of future
climate Is made. Established scien-
tific principles should be used when-
ever possible, so that the assump-
tions are reduced to a minimum. All
assumptions should be tested using
accepted scientific methods, their
influence on the results evaluated,
and their presence made explicit in
any documentation associated with
the scenario.
They must be Internally consistent.
Any scenario which is produced must
be based on the established physical
and chemical laws controlling atmo-
spheric and surface processes. Thus
a scenario must not contravene these
laws, which control not only the theo-
retical limits of variability of a single
climatic element, but also the relation-
ships between two or more elements
and their temporal and spatial variabil-
ity-
They must I provide a range of future
conditions. A scenario is regarded
here as a single estimate of the future
climatic conditions. Since no deter-
ministic forecast of the future is pos-
sible and no single scenario can be
regarded as the "best estimate" of the
future, a set of scenarios creating a
range of possible futures must be
produced. These may be developed
from variations of the assumptions
within a single scenario development
method or by the use of alternative
methods. |
The concern that climate change may
have an impact on human affairs is rather
recent and there is a relatively small, but
rapidly increasing, body of experience in
development of scenarios. Some scenar-
ios have been created primarily to assist
in understanding climatic change, but the
majority have |been oriented towards the
assessment of potential impacts. These
include some general purpose scenarios
designed to treat a variety of impacts and
some more closely tailored to specific
ones. However, no comprehensive
review of possible approaches has
included assessment of the strengths,
weaknesses, and suitability for particular
tasks of the various potential techniques.
Such a review must be undertaken if
appropriate, efficient, and effective
scenarios are to be developed.
The users of these scenarios are
those responsible for assessing the
potential impacts of future climate
changes. They also have little experi-
ence. Assessments are commonly un-
dertaken using a model where statistics
derived for the present climate are the
input. This information is combined with
information about the impact, and manipu-
lated in various ways to yield an output
relating climate variations to the specific
impact. However, it is often difficult to
disaggregate the model to identify those
climatic elements and modes of variation
which are most important, and therefore
those elements whose emphasis in future
scenarios is rpost important. Thus, users
have had varying degrees of success
with the presently available scenarios.
The initial experience, however, provides
insights into the real information needs of
users, experience which is assessed
here as an important element in the
production of user-oriented scenarios.
Combining both the user needs and
the scenario development possibilities,
potentially fruitful lines of research into
methods of scenario development can bo
identified. By including consideration of
the development effort required, probabil-
ity of success, and the potential for trans-
forming the method from a research tool
to an operational technique, it is possible
to develop priorities and recommenda-
tions for scenario research.
This review provides a theoretical
framework upon which practical experi-
ence of scenario development must bis
built. Current knowledge, along with the
value of the experience likely to accrue
from creating and using scenarios, dic-
tates that an iterative approach to devel-
opment be ussed. Scenarios are devel-
oped by climalologists in cooperation with
users and tested by those users.
Refinements can be incorporated as a
result of both the user evaluation and the
ongoing scenario research. Over several
iterations scientifically sound, readily
usable, and clearly documented scenar-
ios can be produced for routine use in
impact assessment.
Research Needs
Programmatic Needs
In developing such a strategy, there
are two major considerations:
1. The need to develop some scenarios
in the near-term.
As part of the overall scheme of the
EPA Global Climate Change Research
Plan, and as the explicit strategy adopted
here, an iterative approach to scenario
development must be used. As such, it Is
necessary to produce some scenarios In
the near-term which can be provided to
those assessing impacts. Although
these must contain the best and most
pertinent information possible, an equal
focus is on assessing the user's
"reaction" to the nature and quality of the
scenario which is provided. This informa-
tion can then be used in the next iteration
to produce a new set of scenarios which
are both more scientifically sound and
more appropriate for the user.
-------
2. The need to support research
designed to develop, in the long-term,
more scientifically sound scenarios.
New methods of scenario development
must be explored in an explicitly research
mode. A variety of efforts, involving sev-
eral modules, and with varying levels of
immediate relevance to scenario devel-
opment, must be fostered. Both user
needs and scientific credibility must be
considered. It is anticipated that the
results of this type of research, as they
become available, will be incorporated
into the subsequent generations of
scenarios. Although EPA has a specific,
mandated role in the production of
assessments of impacts of potential
climatic changes, andjspecific research
strengths because of its traditional mis-
sion, it is not operating in isolation. A
considerable body of research, from a
variety of perspectives, is being under-
taken or sponsored by several interna-
tional bodies, government agencies and
private groups. EPA must be cognizant
of this work, both to avoid unnecessary
duplication and to seize opportunities to
use legitimate research results, where
appropriate, in the further development of
scientifically sound scenarios.
Scenario Development
Methods
Three basic types of scenario meth-
ods werฉ identified: Process Models,
Empirical Methods and Linkage
Techniques. Thฎ first two arise because
of the fundamental division in the type of
climatic information available for scenario
production. The process models attempt
to use the underlying physical laws of
atmospheric processes to deduce climate
under changed conditions. The empirical
approaches use the records of past
climate to indicate possible future condi-
tions. Both approaches have strengths
and weaknesses (Table 1). A consensus
is emerging that process models must
form the basis of any scenario because
they provide the only means for explicitly
estimating climate in changed atmo-
spheric conditions. At present, however,
such models cannot provide the spatial
and temporal detail needed for the provi-
sion of user-oriented scenarios. Hence,
they must be linked to empirical methods
if useful scenarios are to be produced
(Table 2). Although the importance of this
linkage may decrease as the process
models become more sophisticated,
detailed, and reliable, it is clear that they
will be needed well into the foreseeable
future. Methods of establishing these
links provide the third major group of sce-
nario modules.
The process models strive to use the
laws governing atmospheric processes to
understand the present conditions and
estimate future ones when boundary
conditions are changed. There are
numerous possible approaches to pro-
cess modeling. The major division is
based on spatial scale. Different model-
ing techniques, having different input
requirements and output products, are
used for each scale.
Process models include both
"physical" and "chemical" models. The
former are emphasized here, being more
directly related to the elements normally
regarded as climatic. Nevertheless,
chemical models are designed for air
quality assessments, and thus people
using them constitute one group of users
for whom scenarios must be designed.
Nevertheless, chemical activity in the
atmosphere can have a major influence
on the climate itself, and there are impor-
tant links between chemistry and climate
at all scales.
Within the empirical techniques cate-
gory are analyses of past conditions,
whether the instrumental record of the
past few decades, historical sources for
the past few centuries, or the paleocli-
matic reconstructions of past millennia.
The strength of these approaches is that
many of the techniques are welt estab-
lished and can provide a great amount of
detail derived from actual, known occur-
rences. The prime drawbacks are that the
physical causes of future climate condi-
tions may be different from those of the
past, and so the future conditions may be
outside the past range of observed situa-
tions.
The above discussion indicates that
few of the modules, either process or
empirical, can stand alone as scenario
production techniques. There must be
linkage between them. The most desir-
able linkage is one whereby the output of
one module feeds directly into the subse-
quent one. For such links the modules
must have appropriate types of outputs
and inputs. In many cases this mainly
involves a matching of temporal and spa-
tial scales. Consideration of the potential
for such links should be encouraged dur-
ing module development. At present,
however, few direct links can be estab-
lished and other linkage techniques must
be used.
Table 1. The Major Strengths and Weaknesses of the Process and Empirical Approaches to
Scenario Development
Strength
Weakness
Process Models
Explicitly incorporate many
atmospheric processes
Directly include changed
atmospheric composition
Empirical Methods
Provide great temporal and spatial
resolution for many regions
Many elements available
Coarse temporal and spatial
resolution
Few elements estimated with any
confidence
Cannot readily consider changed
atmospheric composition
Processes deduced only by
inference
-------
Table 2, Tho Potential Scenario Development Modules
Global Models
General Circulation Models
Global Chemical Models
Regional Models
Numerical Weather Prediction Models
Regional Climata Models
Regional Chemistry Models
Local Models
Mesoscale Meteorological Models
Urban Chemistry Models
Surface Boundary Models
Analogue Techniques
Instrumental Analogues
Proxy, Analogues
Spatial Shifts
|
Circulation Analysis
Circulation Indices
TelecOnnections
Climatological Statistics
Mean, Statistics
Event Statistics
Temporal Statistics
Synoptic \Analysis
Statistical Synoptics
Spatial Synoptics
Linkage Techniques
Adjustment Methods
Statistical Adjustments
Spatial Adjustment
Transfer Function Methods
Statistical Transfers
Process Transfers
Synthetic Methods
There Is relatively little experience In the
creation of linkage modules and no
methods are well established. Indeed,
this Is an area of research which has itself
been stimulated by the need for scenario
development. Nevertheless, several
lines of approach can be suggested within
the modular context. They must, in most
cases, be speculative, and Indicate gen-
eral lines of approach rather than refer to
already well-established techniques.
User Needs Assessment
The second major piece of information
required for scenario development is
specification of the climatic information
needs of those undertaking Impact
assessments. The impacts involve a
variety of disciplines, including water
resources, ecology, air quality, and agri-
cultural production. Therefore, a wide
variety of needs could be anticipated.
However, for this analysis the require-
ment Is for the Identification of the major
types of needs, expressed In a form
which allows them to be linked with the
possible scenario development
approaches Identified In the preceding
section. Although some Indications of
such needs can be obtained from the
published literature it was deemed most
advisable to assess them in direct coop-
eration with people having experience
with impact assessment.
We asked for a priority list of climatic
elements that the users would like to have
in future scenarios. Table 3 gives a list of
those variables, with the percentage of
the total number of responses that asked
for that variable. Most respondents listed
a number of factors, some only a few.
Other variables, not listed in Table 3, that
were mentioned more than once included
mixing depth, jevaporation, growing sea-
son, storms, and glaciation.
Other user needs are summarized to
Identify major needs for climatic informa-
tion (Table 4).
A time scale of one day will, with the
exception of the unusual events, satisfy
most of the user needs. The daily values
must, however, be capable of producing a
time average allowing for simpler scenar-
ios for those who require them. Shorter
time scales arid unusual statistics may be
needed by only a small minority of users.
Spatial scale requests were much
more variable^ but it seems that a goal of
a 100-km grid would satisfy most users.
However, variables such as corrective
rainfall and orographlc effects will cer-
tainly not be represented In this grid, and
are needed in many important cases, so
some accommodation will be necessary
for these cases.
The response to questions about the
actual climatic information needed allows
the division of the information needs into
four broad categories (Table 4). In large
measure thesa reflect current experience
with scenario use. The first of these can
at this time be specified in some detail,
the others, in many cases, will require
extensive research to be able to ba
produced.
Conclusions and
Recommendations
This project assessed the information
needs of potential users of climate sce-
narios, the methods available to develop
scenarios, and strategies for research to
develop scientifically credible impact-
oriented scenarios.
User needs were ascertained through
consultation with people having some
experience with scenario use. The major
needs are:
-------
Table 3. Climate Element Needs of Impact
Assessors
Table 4. Summary of Major User Needs Identified from
Discussions with Scenario Users
Variable
Temperature
Precipitation
Wind
Radiation
Water Vapor
Clouds
Snow
Pressure
Percent of
Respondents
90
31
52
43
48
29
20
10
Simple descriptive statistics for indi-
vidual climatic elements.
Climate anomaly information, espe-
cially concerning drought.
Information about the frequency and
probability of occurrence of various
threshold values.
Synoptic information, especially con-
cerning storms.
Potential scenario development
methods were identified through literature
reviews and consultations with people
having experience in scenario
development. It was recognized that
currently available scenarios are primitive
and may be misleading, but that
techniques are already available to
produce more scientifically sound ones.
From these investigations, it is
recommended that a modular approach to
scenario development be adopted. Each
module consists of a set of related anal-
ysis techniques which are themselves
well established and tested, and which
can be used to provide pieces of informa-
tion required for scenarios. Linkage and
combination of modules allows the
development of a complete scenario.
Ten modules were identified. Three
involve process models which use the
Time Scale
Daily values - not necessarily a sequence of daily weather
Space Scale
100-km x 100-km grid-specific needs highly variable
Information Needed
1. Simple descriptive statistics for individual
elements (means and variance, totals, extremes, etc)
2. Climatic anomaly information, especially concerning
drought (magnitude and persistence of events)
3. Threshold values
(probability of values significant for particular impact)
4. Synoptic information, especially for storms
(sequences of events and combinations of elements)
laws of chemistry and physics to provide
understanding of the processes acting to
control climate, and thus can be used
directly to investigate climatic changes
resulting from the greenhouse effect.
Included with these are the General
Circulation Models which provide the main
basis for development of scenarios. Four
modules incorporate empirical techniques
for scenario development, which use the
records of past climatic conditions to
provide indications of the nature of
climatic variations on the local time and
space scales needed for most impact
assessment. The final three modules are
linkage techniques designed to allow
combination of the other modules to
provide the required scenarios.
The techniques in each of these
modules are at various stages of devel-
opment. Some can be used directly for
scenario production while for others a
major research effort is needed before
they can be used. For almost all of them,
their suitability as scenario production
techniques has yet to be tested.
Because of the various levels of
development of the modules, and given
the overall lack of experience with sce-
nario development, it is recommended
that an iterative approach to scenario
development be adopted. For each phase
a set of scenarios are produced and used
for the assessment of a particular impact.
Thereafter the scenario is assessed for
its scientific credibility and its value for
impact assessment. Appropriate modifi-
cations and refinements are made and a
new scenario produced to start the next
iteration.
Three phases of iteration are recom-
mended here, covering the short-term (1 -
2 years), the medium-term (3-4 years),
and the long-term (more than 4 years).
Within each phase, three types of pro-
jects are recommended:
Scenario development
The actual production of scenarios
which can be provided to the users for
impact assessment.
Module research
Investigations aimed at refining the
techniques within modules to produce
more soundly based scenarios.
Ancillary activities
Projects designed to ensure that any
scenarios produced are responsive to
user needs and can be routinely and
easily used in impact assessment.
-------
-------
-------
Peter J. Robinson is with University of North Carolina, Chapel Hill, NC,
27599. :
Peter L. Flnkelsteln is the EPA Project Officer (see below).
The complete report, entitled "Strategies for\ the Development of
Climate Scenarios for Impact Assessment," (Order No. PB 90-192
022/AS; Cost: $17.00, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703/487-4650
The EPA Project Officer can be contacted at: ;
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States
Environmental Protection
Agency
Center for Environmental Research
Information :
Cincinnati Of-145268
Official Business
Penalty for Private Use $300
EPA/600/S3-90/026
------- |