CLIMATE MODELS AND THE VALIDATION AND PRESENTATION OF
GREENHOUSE CHANGE THEORY
by
James S. Risbey
University of Melbourne
1983
M.Sc., Massachusetts Institute of Technology
1987
B.Sc.(Hons),
Submitted to the
Center for Technology, Policy, and Industrial Development
in Partial Fulfillment of the Requirements
for the Degree of
MASTER OF SCIENCE
IN TECHNOLOGY AND POLICY
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
February
©
Signature
1990
Massachusetts Institute of Technology 1990
of Author
Center for Technology, Policy and Industrial Development
tmospheic, and Planetary Sciences
Department of Earth,
February 14, 1990
Certified
byJudith Kildow
Thesis Supervisor
Certified by
Peter
Thesis
Accepted
by
Stone
Supervisor
Richard de Neufville
Chairman, Technology and Policy Program
K-A
ACKNOWLEDGEMENTS
I would like to thank my advisors, Peter Stone and Judith Kildow for
their longstanding support and guidance, and for providing the necessary
direction and emphasis to ensure completion of the thesis. Thanks are also due
to them for giving attention to the thesis, often under short notice, while I was
I am grateful to Mark Handel and Chantal
attempting to meet final deadlines.
Rivest who have been great resources and officemates, as well as suppliers of
Bill Clark
the necessary quantity of fine chocolate for working on a thesis.
and Tad Homer-Dixon were also important sources of resources and feedback.
and other students in the International
Peter Cebon, Peter Poole,
Environmental Issues Study Group have provided helpful stimulus in various
I thank my family for their support
ways throughout the course of the work.
Linda Murrow, Jane McNabb, Tracey Stanelum, and Daniel
from afar.
Leibowitz also provided invaluable support.
Funding was generously provided by NASA Goddard Institute for Space
Studies (GISS) for this research under grant 'NASA /G-g NSG 5113', and I am
GISS director James Hansen in particular was most
indebted to them.
understanding in continuing to provide financial support while I registered
in the Technology and Policy Program, which was a digression of sorts from
conventional straight science projects carried out in the department.
A number of people provided data and data support for this project, and
They include Bill Gutowski, W.C. Wang and
I am very grateful to them all.
David Salstein of AER, who provided most of the NCAR, GFDL, and GISS GCM data.
Reto Reudy of NASA GISS provided energy transport data, flux data, and
Richard Wetherald of GFDL provided
topographic data for the GISS GCM.
Robert Black of MIT provided topographic
additional data for the GFDL GCM.
data for the NCAR GCM. I thank NCAR, GFDL, and GISS GCM modelling groups
for allowing me to use output from their models.
CONTENTS
6
ABSTRACT
-7
LIST OF FIGURES
LIST OF TABLES
10
GLOSSARY OF PRINCIPAL SYMBOLS
11
1.
1.1 Intent,
2.
12
INTRODUCTION
and thesis
background,
definition
20
GREENHOUSE THEORY
2.1 Greenhouse
effect
20
2.2 Greenhouse
change
21
23
2.3 Climate models and climate
3.
experiments
2.4 Greenhouse
change
modelling
2.5 Greenhouse
change
controversies
2.6 Greenhouse
problems
24
26
33
VALIDATION OF CLIMATE MODELS
34
3.1 Approach
34
validations
3.1.1
Internal
3.1.2
Perturbed-run
3.1.3
Control-run
validations
validations
3.2 NCAR, GFDL, GISS model validations
3.3 Energy
4.
12
validations
transport
38
40
43
46
hierarchies
4.2 One-dimensional
36
46
MODEL DESCRIPTIONS
4.1 Model
35
energy
4.3 NCAR, GFDL, GISS models
balance climate model
47
50
5.
DATA
53
5.1 Climate model output
53
5.2 Observational
5.3 Surface
5.4 Sea
energy
56
derivation
temperature
5.4.1
Lapse rate data
5.4.2
Topographic
5.4.3
Sea level
56
57
data
57
temperatures
5.5 Satellite data
58
5.6 Sea ice data
58
5.7 Journal literature
6.
56
data
air temperature
level
55
data
transport
and media
59
accounts
VALIDATION OF MODEL FUNDAMENTALS: ENERGY TRANSPORT
61
6.1 Energy transport and its role in climate
61
6.2 Procedure to calculate energy transport in the models
6.3 Comparison of direct and indirect transport calculation
62
method
66
6.4 Results
7.
65
6.4.1
Annual radiative fluxes at the top of the atmosphere
6.4.2
Energy
6.4.3
Atmospheric
energy
6.4.4
Temperature
structure
66
68
transports
71
transports
73
MODEL CLIMATE SENSITIVITY
76
7.1 Introduction
76
7.2 Climate sensitivity to the energy transport
77
7.3 Equivalent
79
EBM parameterization
7.3.1
Parameterization
7.3.2
Mean annual
79
technique
meridional distribution
of solar radiation
80
and solar constant
7.3.3
Iceline latitudes
7.3.4
Outgoing
7.3.5
Albedos
and sea level temperatures
longwave
7.4 Equivalent EBM results
7.5 Discussion
7.6 GCM validation implications
radiation
_81
82
83
_85
89
90
5
8.
GREENHOUSE CHANGE VALIDATION FRAMEWORK
96
8.1 Introduction
96
outline
8.3 Validation
components
103
8.3.1
Uncertainty
105
8.3.2
Credibility
107
8.3.3
Useability
110
8.4 Greenhouse
9.
97
8.2 Framework
change
characteristics
112
GREENHOUSE CHANGE VALIDATION PRESENTATIONS
116
9.1 Introduction
116
9.1.1
Approach
118
9.1.2
Background
119
9.2 Discussion of scientific papers on or implicated in detection of
greenhouse
change
9.3 Characterization
9.4 Science
of the media coverage
presentation
9.4.1
General
9.4.2
Scientific
9.5 The
121
standards
guidelines
science-media
standards pre and post presentation
interface
9.6 Conclusions
130
131
131
133
138
139
10. SUMMARY AND CONCLUSIONS
142
FIGURES
147
REFERENCES
178
CLIMATE MODELS AND THE VALIDATION AND PRESENTATION OF
GREENHOUSE CHANGE THEORY
by
JAMES S. RISBEY
Submitted to the
Center for Technology, Policy, and Industrial Development
in Partial Fulfillment of the Requirements for the Degree of
Master of Science in Technology and Policy
ABSTRACT
We identified problems in validating greenhouse change theory
We focused on GCMs, noting the heavy reliance
science-policy context.
As a case study
3-D GCMs for developing greenhouse change scenarios.
validation, we analysed simulation of energy transports in the three
GCMs used in greenhouse climate studies of NCAR, GFDL, and GISS.
in a
upon
GCM
major
Comparison of energy transports in the GISS GCM calculated directly
from model motions and indirectly from the net flux of radiation at the top of
The
the atmosphere indicated that the indirect technique is satisfactory.
atmospheric energy transports calculated using the indirect technique were
found to be similar in all GCMs, but too efficient, being much larger than
For the total energy transports significant differences were
observations.
obtained between the three GCMs in the NH, with NCAR and GFDL less than
We used a 1-D EBM to
observations and GISS greater than observations.
demonstrate that the climate is potentially quite sensitive to the energy
To test implications of the differences in energy transport between
transport.
GCMs, we parameterized the EBM with output from the GCMs and observational
Using an increase of solar constant by 2% as a
data, separately in each case.
proxy for CO 2 doubling in the EBMs, similar sensitivities were obtained
between equivalent EBM representations for each GCM and with the GCM
The differences in energy transport between GCMs
doubled CO 2 sensitivities.
were apparently compensated for in the equivalent EBM representation,
though the sources of compensation are not yet clear.
We described a framework for validating GCMs and greenhouse change,
comparing and contrasting the approach and requirements of science and
We noted potential mismatches in operation and
policy communities.
requirements between the communities, and the current inability to
We
incorporate uncertainties in critical evaluation of greenhouse change.
considered communication of validation information from science to policy by
Instances of
analysing science journal literature and its media interpretation.
to a
partly
attributed
and
identified
were
media
the
in
misinterpretation
journal
the
in
audience
a
broader
to
appropriate
context
failure to provide
We argued that the addition of context is required when science
literature.
journal literature has clear policy implications and is undergoing scrutiny by
the media.
Thesis Supervisors: Dr. Judith Kildow and Dr. Peter Stone
Titles: Professor of Technology and Policy, and Professor of Meteorology
LIST OF FIGURES
5.1
Zonal mean 850mb - 950mb lapse rate from Oort (1983) data
147
5.2
GISS topography on the GISS grid
148
5.3
NCAR CCM1 R15 surface geopotential height
149
5.4
NCAR (N) and GISS (G) zonal mean topography
150
5.5
NCAR (N), GFDL (P), and GISS (G) control run surface
temperatures (solid lines) and sea level temperatures
(dashed lines)
151
6.1
Annual mean zonal mean total meridional (northward) energy _152
transport (T), atmospheric energy transport (A), and ocean
energy transport (0) for Carissimo et al. (1985) observations
(dashed lines), and calculated directly from the GISS control
run model motions (solid lines)
6.2
Annual mean zonal mean total northward energy transport
calculated directly from the GISS model motions (D), and
indirectly from the net flux of radiation at the top of the
atmosphere in the GISS model (I) (control run)
153
6.3.16.3.6
Annual mean zonal mean shortwave flux at the top of the
atmosphere (SWTP), longwave flux at the top of the atmosphere
(LWTP), shortwave flux at the surface (SWSF), longwave flux at
the surface (LWSF), sensible heat flux at the surface (SHSF),
and latent heat flux at the surface (LHSF) for the NCAR (N1,N2),
GFDL (P1,P2), and GISS (G1,G2) models. The numbers '1' and '2'
refer to equilibrium results for 1 x CO 2 and 2 x CO 2 simulations
154
6.3.7
Annual mean zonal mean net radiative flux at the top of the
atmosphere from NCAR (N), GFDL (P), and GISS (G) model
control runs, and from Stephens et al. (1981) observations (0)
156
6.4.16.4.3
Annual mean zonal mean total northward energy transport for
NCAR, GFDL, and GISS models (control run). The curves
labelled 'N' are for integrations starting at the north pole, the
curves labelled 'S' are for integrations starting at the south
pole, and the curves labelled 'C' are corrected curves. The
dashed curve is from Carissimo et al. (1985) observations (0)
157
6.4.4
Corrected annual mean zonal mean total northward energy
transport for NCAR (N), GFDL (P), and GISS (G) models (control
run), and for Carissimo et al. (1985) observations (0)
_157
6.5
Annual mean zonal mean northward atmospheric energy
transport for NCAR (N), GFDL (P), and GISS (G) models (control
run), and for Carissimo et al. (1985) observations (0)
_158
7.1
Iceline latitude (xs) versus ratio of the solar constant to the
current solar constant (q/qo) for the 1-D EBM
159
7.2
Meridional energy transport flux as a function of the sine of
latitude (x) for the 1-D EBM
160
7.3.17.3.3
Maximum energy transport flux, second component of
outgoing longwave radiation (I2), and iceline latitude (xs) as
functions of the dynamical transport efficiency (D) for the
1-D EBM
161
7.4
Iceline latitude (xs) for NCAR (ncarnh, ncarsh, ncargl), GFDL
(gfdlnh, gfdlsh, gfdlgl), and GISS (gissnh, gisssh, gissgl)
models (control run), and from Alexander and Mobley (1976)
observations (obsnnh, obsnsh, obsngl).
In each case, 'nh' is
for the Northern Hemisphere, and 'sh' is for the Southern
Hemisphere.
The global (gl) values are averages of the
hemispheric values in each case
162
7.5
As in fig. 7.4, but for the sea level temperature at the iceline
latitude.
Temperature observations are from Sellers (1965)
162
7.6.17.6.6
Annual mean zonal mean planetary albedo (ALBEDO), net
flux of longwave radiation at the top of the atmosphere (I),
and sea level temperature (T) for NCAR (N), GFDL (P), and GISS
(G) models (control run), and from Stephens et al. (1981)
observations (0 for Albedo and I) and Sellers (1965)
observations (0 for T). The curves are plotted separately for
the Southern Hemisphere (SH) and Northern Hemisphere (NH)
as a function of the sine of latitude (x)
163
7.7.17.7.2
As in fig. 7.4, but for parameterized values for A and B from the
temperature - outgoing longwave radiation parameterization
I = A + BT in the equivalent EBMs.
164
7.8.17.8.2
Parameterized outgoing longwave radiation (I) as a function of _165
the sine of latitude in Southern and Northern Hemispheres (SH
and NH) for NCAR (N), GFDL (P), and GISS (G) models, and for
observations (0). I is calculated via I=A+BT with the
parameterized A and B values and with the actual zonal mean
temperatures for each case (not with the two mode Legendre
representation of the temperature)
7.9.17.9.3
As in fig. 7.4, but for parameterized values for the albedo
components ao, a2, and 8 in the equivalent EBMs
7.10.1Parameterized albedo as a function of the sine of latitude (x) in
7.10.2 Southern and Northern Hemispheres (SH and NH) for NCAR (N),
GFDL (P), and GISS (G) models, and for observations (0)
7.11
Outgoing longwave radiation at the iceline latitude (Is) from
each of the twelve equivalent EBMs
166
167
168
7.12
Dynamical transport efficiency (D)
equivalent EBMs
from each of the twelve
168
from
169
Outgoing longwave radiation components
7.13.17.13.2 each of the twelve equivalent EBMs
Temperature components
7.14.17.14.2 equivalent EBMs
(I)
and (12)
(TO) and (T 2 ) from each of the twelve
170
Sea level temperature from the original NCAR, GFDL, and GISS _171
7.15.17.15.8 GCMs and observations (solid lines), and calculated via the
equation T(x)=To+T 2 P 2 (x), where the To and T2 have been taken
from the equivalent EBMs (dashed lines). The results are
plotted as a function of the sine of latitude (x) for each
hemisphere
7.16
Energy transport flux maximum from the equivalent EBMs
(solid bars) and from the original GCMs and Carissimo et al.
For the latter the global
observations (cross hatched bars).
'gl' values are averages of the hemispheric values
172
7.17
Iceline latitudes (xs) as input to the equivalent EBMs in each
case (solid bars), and as calculated in the equivalent EBMs
when the solar constant was increased by 2% in each case
(cross hatched bars)
173
7.18
Climate sensitivity calculated from the equivalent EBMs
174
8.1
Greenhouse
change
validation
framework
175
10
LIST OF TABLES
5.1
Latitudinal spacing of the zonal mean data in the NCAR, GFDL,
and GISS models, in Carissimo et al. (1985) observations, and in
Sellers (1965) observations (NO and S*)
6.1
Sea level temperatures
at 550 and 25* in each hemisphere
_
60
_
75
_
95
derived from GCMs and Sellers (1965) observations. AT is the
temperature difference per degree of latitude over this range
of latitudes
7.1
In each
Inputs and outputs from the twelve equivalent EBMs.
case the first four letters denote the model or observations, and
the last two letters denote the hemisphere (Northern
Hemisphere or Southern Hemisphere) or average of the
Results are given for both the current
hemispheres (Global).
The
solar constant and for the solar constant increased by 2%.
climate sensitivity calculated for each case is also shown
GLOSSARY OF PRINCIPAL SYMBOLS
symbols of physical
The principal
and mathematical
quantities used in
Some symbols formed from principal symbols by
this report are listed below.
adding subscripts, and others that are used in only one place are not listed.
Some quantities are denoted by two different symbols for clarity of notation in
Usage should be unambiguous in any given section.
different sections.
Symbol
Name or Definition
a
Radius of the earth
A
Longwave
-
temperature
parameter
B
Longwave
-
temperature
parameter
D
Dynamical
transport
F
Meridional
energy
FTA
Net downward flux of radiation at the top of the atmosphere
Hn
Insolation
I
Outgoing longwave radiation at the top of the atmosphere
Pn
Legendre
q
Solar
R
Radius of the earth
S
Meridional
T
Sea
TT
Total
x
Sine of latitude
z
Geopotential
a
Albedo
efficiency
transport
flux
component
polynomial
constant
distribution
of solar radiation
level temperature
meridional
Climate
Lapse
energy
transport
height
sensitivity
rate
parameter
6
Albedo
0
Latitude
a
Stefan
<p
Longitude
- Boltzmann
constant
12
1
definition
thesis
and
background
Intent,
1.1
INTRODUCTION
theory, its validation
The subject material for this thesis is greenhousel
here
to
to
refer
the
of whether
question
in
changes
with
associated
is used
theory
the
We consider the problems associated with the
and
science
greenhouse
support policy
to
enough
from
obtained
the information
is credible
models
Greenhouse
presentation.
of knowledge
body
greenhouse effect.
atmospheric
climate
and subsequent
models,
in climate
Long
decisions.
the
domain of the scientist, greenhouse theory is now also part of the purview of
the politician,
the policy
analyst,
The
examines
thesis
of
problems
involves
the
growing
links
between
theory
greenhouse
validating
link between
basic
in developing
greenhouse
in
and
science
policy
about
of thinking
for the purpose
a framework
and
scientific
the
communication
validation
in
process
science
of validation information
etc.)
is uncertain,
from
given to assessing how well
scientists
arenas
supports
theory,
what is knowable,
Attention
to policymakers.
scientific information
and policy
(what
what undermines theory, what is known, what is unknown,
what
change
greenhouse
in validating greenhouse change theory in science and policy.
issues involved
A
the
will present
We
domains.
used
proletarian.
and the
These links will be highlighted through a consideration of
greenhouse policy.
the
models
climate
and explores
scenarios,
owner,
the property
on validation
will be
is presented
to the policy community.
The
impacts
associated
with
current
greenhouse
small part of the greenhouse issue is considered.
change
policy measures
implemented
in
from
We examine the theory of
the point of view of what matters
involving public expenditures
response
to
require some justification to act.
greenhouse
are
In the thesis however, only a
potentially serious, and warrant much attention.
greenhouse
predictions
for policy.
and lifestyle changes
predictions,
They will need
tw
then
If
are to be
policymakers
know if the theory
is
1The term 'greenhouse' is objectionable to some, primarily on the grounds that the physics associated
with the workings of a glass greenhouse are not very analogous to those associated with the atmospheric
The term is widespread however, and the analogy does have some merit (see
greenhouse effect.
discussion in Bohren, 1987), so its usage will be continued here.
13
from
predictions
the
whether
is
that
'valid';
theory
the
are
and
credible
A climate model is valid to the extent that it is plausible and produces
At some other levels of
results at some levels of aggregation that are useful.
aggregation the results will not be very useful because the model is not
designed to resolve certain scales or processes, or contains assumptions that do
useful. 2
not
we
as
set
implicitly
In a validation
under certain conditions.
hold well
to
seek
are
range
of
any
is
there
whether
determine
study, bounds
3
aggregation of output results that can be considered credible and useful.
thesis,
this
In
analysis
largely
confined
be
will
the
to
of
issue
credibility, and we will not consider actual policies that might be implemented
The credibility of greenhouse theory with
in response to greenhouse change.
regard to policy has not been directly addressed since beginnings were made
in U.S. National Academy of Science reports in 1979 and 1982 (NRC, 1979/1982).
that
Given
greenhouse
of credibility
assessment
of credibility,
varying degrees
accord
scientists
different
predictions
with
by non-scientists
in
No attempt is made to answer the
the policy domain is bound to be difficult.
That assessment largely
question of credibility definitively for policy makers.
boils
to
down
collectively
greenhouse
judgement,
the problems
and
science
value
by interpreting
on
focus is
a
policy
issue
that
available
be
must
made
be
knowledge through
of assessing
arenas,
can
and
credibility
and
on
identifying
used
by
science
individually
value systems.
of greenhouse
and
characteristics
policy
and
The
theory
in
of
the
communities
to
establish a basis for examining policy credibility.
Greenhouse
of uncertainty,
scientists
warming is a complex issue characterized by a high degree
making policymakers particularly sensitive to the way that
present information
relevant to establishing
credibility
in the issue.
Scientific inputs initiate the policy process, alerting society to the existence of
a problem (potential greenhouse warming) and the need for policy responses.
We take a practical approach (considering what is useful) rather than a philosophical approach (i.e.
can a model or theory ever be validated) to the problems of validating theory, since this is more
immediately applicable to the policy making process.
3
This definition is general enough to apply to validation in science or policy arenas, though in science
it may occasionally be irrelevant, since we are sometimes only interested in understanding processes
in models and do not care about the validity of the results otherwise.
2
1
Introduction
14
Once a scientific issue takes on policy dimensions
research
further scientific
is stimulated and the results continue to filter through to the policy
The presentation of these results shapes the policy process by
bolstering
or
We
response.
the perception
reducing
will
carry
to
attempt
been
have
of the
problem
presenting
in
which
scientists
greenhouse
change
theory to the policy community.
manner
assessment
validation
problems
in
encountered
assessing
the
a
of the
information
in
The first is to
Thus far, essentially two tasks have been constructed.
highlight
either
case for
and the
out a preliminary
arena.
credibility
of
greenhouse
and policy arenas, and to identify links to facilitate closer
To this end, the second task
coupling of information between the two arenas.
involves assessing how well scientists are doing at translating knowledge on
validation of greenhouse theory from the scientific domain to the public
theory in scientific
domain.
This
consideration
is
still
far
too
of fundamentals in
broad,
so
the hope
we reduce
both
problems
that narrow but basic
to
a
approaches
to the problem serve some utility.
Assessing the credibility
of greenhouse theory from
the point of view
of policy entails analysing the nature of predictions from the theory, since it
is the predictions that are the ultimate input of the science to the policy
For greenhouse theory, the 'prediction inputs' to policy take the form
4
of 'how big a temperature rise or precipitation change, how fast, and where?'
By and large, these predictions come from three-dimensional general
5 As
circulation climate models (hereafter simply 'climate models' or 'GCM's').
process.
More correctly, it is probably the impacts on social systems associated with these predictions that is
meaningful to policy makers, and not so much the predictions themselves. In other words, 'so what if
In practice however, the impacts are incredibly
the temperature rises five degrees on average?'
In addition,
difficult to determine accurately, and add another layer of uncertainty to the problem.
the predictions from the models are by no means totally uniform, so that it is difficult to agree on a
climatic scenario on which to study impacts. For these and other reasons, policymakers still find it
useful to focus on the raw climatic predictions.
5
The terminology is a little confusing, since 'GCM' is used to refer to both general circulation weather
models used in weather forecasting, and general circulation climate models used in climate studies.
Though the two types of models have many similarities (indeed, some climate models are modified
The differences stem from the need to consider
weather models), there are also crucial differences.
different time scales and therefore different physical processes in one type that are not important in
the other type. In this work, the usage of GCM as climate model is always intended. Sometimes climate
models are further classified as atmospheric (AGCM) and oceanic (OGCM), or as a coupled oceanatmosphere GCM.
4
1
Introduction
15
noted
in the
thesis, in a philosophical
sense the predictions
(and levels
of
confidence associated with them) are not purely the result of the 3-D GCMs.
Input to the 3-D models and interpretation of the results is aided by a host of
simpler
changes
is
it
however,
past
of
observations
to
necessary
a
use
GCM
climatic
about future
quantitative predictions
In order to make reasonable
behaviour.
climatic
and
reasoning,
physical
models,
to
incorporate
anything like the full degree of complexity of the climate system. 6
(1983)
that
notes
of
use
assessment
comprehensive
necessary
for
a
mechanisms
feedback
Hence,
climate change.
distribution of greenhouse
Congressional
of various
influence
of the
is
of climate, as well as for simulating the latitudinal
upon the sensitivity
geographical
GCM's
three-dimensional
Manabe
and
much of the
on greenhouse change has been by GCM
testimony in the U.S.
modelling groups on the results of their model predictions. 7
Representatives
of various modelling groups considered in this work have all testified before
U.S.
GCM's
in
determining
greenhouse
three
major
Research (NCAR),
NASA
Goddard
dimensional
climate
will
focus
on GCM's
Institute
of
data
Space
for
was
Modelling
in
In particular, we will analyse data from
the
National
Center
for Atmospheric
the Geophysical Fluid Dynamics Laboratory (GFDL),
models,
intercomparison.
models
we
predictions,
assessing the credibility of theory.
the
Because of the central role of
Congress on greenhouse change predictons.
Studies
available
made
groups
(GISS).
at
the
in
United
8
a
and the
For these fully three
form
Kingdom
that
facilitated
Meteorological
6
Note that the raw output of an individual 3-D climate model is generally not considered to be a hard
forecast or prediction. Rather, the output is the result of a sensitivity study, say to doubling of carbon
dioxide in the atmosphere. Results of sensitivity studies effectively become predictions or projections
when people start to have confidence in certain aspects of the results. Confidence may come through
consistency with theoretical principles or observations, or confirmation through other sources (other
3-D models or simpler models). This is typical of the way that science works, and should not be
The forecast does not consist of every tiny detail yielded by the model
considered unusual.
experiment, but usually just of the broader scale features, or those features that seem to be grounded
In a weather forecast model, the
on plausible physical mechanisms, or appear- as an overall trend.
patterns yielded on a particular day are intended to be predictions for that day. Climate however, is
really only meaningful in an average sense, and model patterns for individual years are not intended to
be forecasts for those particular years.
7
Readers may well be familiar with the recent testimony of Hansen (1989) (Director of the NASA GISS
climate group) which received media attention (New York Times, 1989d) after being altered on fairly
blatant ideological grounds by the Office of Management and Budget.
8
Running a 3-D climate model requires large resources in financial, technological, and intellectual
capital, and there are currently few 3-D climate models worldwide. The best known climate modelling
groups are listed here. To aid the reader in identifying the groups in subsequent discussion, members
of the groups who are prominently featured as authors on most of the groups papers are listed as well:
NCAR (Washington), GFDL (Manabe), GISS (Hansen), UKMO (Mitchell), and OSU (Schlesinger). UKMO is
the United Kingdom Meteorological Office, and OSU is the Oregon State University.
1
Introduction
16
Office
and
(UKMO)
Oregon
University
State
greenhouse climate experiments,
also
performed
in
are not directly included
but their results
For UKMO, output from the CO 2 experiments was not available at
analysis here.
The OSU model has severly limited
the time thesis research was undertaken.
vertical
have
(OSU)
(2
resolution
levels)
is
and
therefore
well
not
suited
for direct
comparison with the other 3-D models.
and
intents
purposes just
impossible to verify
as
complex
real
system. 9
climatic
to verify
(however, attempts
we
manageable,
To keep this
the
to
simulation
process
a single
to
validate,
has
and until now
of climate,
major models.
for the three
scrutinized
consider
albeit
a
We have chosen to validate the energy transport because it
fundamental one.
is fundamental
will
isolated
even
aspects of model performance have not been extensive to date).
study
It is
all aspects of GCM performance in simulating the earth's
it might change
climate and how
the
as
are for all
since the GCM's
Further refinement of scope is necessary,
The
from
results
not been
a validation
of
energy transports do not yield definitive answers to questions of credibility of
the models, but constitute an important case study from which to explore this
issue.
The role of energy transport can be understood in terms of the earth's
annual
radiative
energy
The
budget.
earth
receives
significantly
more
solar (short wave) energy in the tropics than in the polar
annually
averaged
regions.
The emitted longwave energy is more uniform over the globe, and is
unable to completely offset the equator to pole gradient in the solar heating,
resulting in
a strong
equator to pole net heating
the fundamental
heating
gradient
is
oceanic
general
circulation
energy
transport.
radiative
Since
and dynamical
deficiencies are occuring.
driving
(Ramanathan,
the
energy
processes,
force
1988),
transport
analysis
This
gradient.
for
and
the atmospheric
results
reflects
radiative
the
a
poleward
action
of both
in
of it may indicate
We will also use a one-dimensional
and
where model
energy balance
climate model to study how sensitive the climate is to the energy transport.
9
0f course, the GCM's are gross simplifications of the real climatic system, but they capture many of
the essential processes and feedbacks, and possibly even some of the intrinsic internal behaviour. As
While we
such, they are far too complex and sophisticated for us to fathom all their interconnections.
can perform sensitivity experiments with GCM's, and examine important processes in isolation with
simpler models, the GCM's remain black boxes to some degree.
1
Introduction
17
the
earth,
which
temperature
energy
in
turn
gradient,
transport
which
is thus
not modelled
through
climate.
component
an important
the
ice-albedo
the meridional
Validation
in the overall
of the
verification
greenhouse change projections.
role in
a fundamental
If the transports play
are
climate
regional
influences
of GCMs used for developing
yet
the
influences
The energy transport also determines
mechanism.10
feedback
iceline position on
an important role in determining the
The transport plays
as
well
we
suspect,
are
what
the climate,
determining
the
implications
for
We will consider this
credibility of the models and greenhouse predictions?
and outline the process by which credibility is generally established.
question,
There are no sufficient tests that can be performed to validate the models, so
that confidence
theory
and
is established
this point is crucial for the policy
Understanding
observations.
a gestalt of
by considering
in the predictions
analysis of greenhouse theory if any sense is to be made of it at all by the nonthe
Otherwise,
climatologist.
debate
validation
greenhouse
degenerate
can
into a series of apparent refutations and confirmations.
The
task
second
address
is to
of greenhouse
knowledge on validation
the
of translating
problem
domain to
the scientific
theory from
available
the public domain, in so far as this is relevant to the formulation of policy
responses
many
to greenhouse
levels.
testimony
channels
reaching
press,
radio,
This translation
is between
of communication
television
interviews,
magazines.
All
and
and
occurs
through
One of the more
to politicians.
of scientists
goes on at
of knowledge
most direct forms of translation
of the
One
Congressional
change.
and the media in
scientists
through
popular
fora
articles
have
in
various
newspapers
and
shortcomings,
and the lack of available time and space is common to all of
those listed above.
greenhouse
theory
literature.
Keeping
communication
far
By far the most sophisticated
(and
with
also
the
our focus
root
on
source)
source of information on
is
fundamentals,
the
we
scientific
have
journal
chosen
to
10
Ice-albedo feedback begins when an initial warming causes sea ice and snow to melt. This shrinking
Incrementally more
of the snow and sea ice margins results in a lower albedo in these regions.
incoming radiation can then be absorbed in the formerly ice- and snow-covered areas, thereby
resulting in further warming, more snow and ice melt, and so on until equilibrium is established. For
cooling, the reverse occurs.
Ice grows and increases the albedo and energy loss, leading to further
cooling and growth of ice.
1
Introduction
18
analyse
this
medium
for
the
translation
of
greenhouse
theory
(and
the
problems of validating theory) to the policy level.
The
interpreted
coverage
in
disseminating
also
interpretation
shapes
the
of the journal
that media interpretation
linking science
with
knowledge
receipt
articles
of the
policy
theory
and
and
the
theory
knowledge
determining
controversies
publicly.
of
theory
presentation.
greenhouse journal
and in
been
has
Press coverage of journals is
and resultant
the credibility of greenhouse
greenhouse
of
of
theory
greenhouse
in the press throughout its history.
instrumental
assesses
on
literature
journal
science
literature
Press
through
We believe
is important in
how the policy
community
theory.
Thus,
we
will consider
relating
to
are
presented
in
the
are then
interpreted
by
the
journal
literature,
media.
The New York Times will be used as the principle media source, because
it is well archived
discussion
research
of
results
articles
and is among the most sophisticated of the presses in its
environmental
relevant
various patterns
attempt
how the journal
it
how
to classify
them
context contributes
to
Misinterpretation
validating
to
and types
issues.
greenhouse
of misinterpretation
the occurence
theory
can
media
the
is
frequent,
be identified.
the failure
and show how
in
to
We
provide
and
will
appropriate
Though
of misinterpretation.
of
instances
of misinterpretation of the validation of theory probably result from errors by
the media as well as scientists, we primarily will be concerned with the role of
scientists.
above, it is hoped that we
Through the sequence of analyses described
can identify what the major problems associated with validation of greenhouse
theory
are,
and
the
how
problems
are
being
dealt
community in presentation to the policy community.
have
policy
scientists
think
framework
about
when
the
validation
communicating
attempt at this last problem would try to account
the
by
science
A longer range goal is to
problem
validation
with
in
relationship
to
the
Any serious
information.
for the factors motivating
scientists in framing their results, which is too big a task for this thesis.
In subsequent
the role of climate
climate models.
chapters
we will briefly outline
greenhouse
theory and
models, before reviewing previous work on validation of
We will then describe
the three climate models
considered
1
Introduction
19
here, the output used
from the models,
and the observational
data used
to
The results for validation of the models will be
with the models.
Implications of errors
presented, and implications of the results considered.
in the energy transport on climate sensitivity will be examined by
compare
Taking
parameterizing 1-D energy balance models with output from the GCMs.
the validation issue in a wider context, chapter eight will present a scienceThe penultimate chapter will consider how well
policy validation framework.
greenhouse theory is presented,
and conclusions will follow.
1
Introduction
20
2 GREENHOUSE THEORY
2.1
Greenhouse
Perhaps
way
good
been discussed
has
to
in the
widely
through
a
We note that greenhouse
scientific
without close attention to language definition.
is
theory
greenhouse
introduce
definition of parts of the theory.
and
breakdown
theory
a
effect
literature
the press
and
This has resulted in a blurring
of distinction between key parts of the theory.
Following
'greenhouse
effect'
physics
(primarily
radiatively
whereby
at the earth's
than
the
change'
to
we
will
to
refer
use
absorption
gases
warm surface of the earth.
in the infrared
dioxide and ozone)
different
blackbody
absorb
by
of
temperature
the longwave
radiative
anthropogenic
gases
a
and clouds, maintains
of
the
considerably
radiation
The
planet. 11
emitted
by
the
This energy is reemitted back to the surface, and
This yields a
also back into space at much colder atmospheric temperatures.
nothing
parts
various trace
in the lower troposphere
surface and
equivalent
active trace
net trapping of
terms
the
The greenhouse effect refers to the well known and basic
water vapour, carbon
temperature
greater
'greenhouse
and
greenhouse theory.
radiative
communication)
(personal
Handel
energy
or greenhouse
in this definition,
effect.
is
there
Note that
time dependent.
nor even anything
The above description of the greenhouse effect can be set in quantitive
The effective radiating temperature of the earth,
terms quite easily as follows:
Te, is determined by the requirement that infrared emission from the planet
balance the absorbed
solar radiation thus:
47cR 2aTe = nR 2(1-A)q,
where R is the radius of the earth, a
(2.1)
the
constant,
Stefan-Boltzmann
A
the
albedo of the earth, and qo the flux of solar radiation at the mean earth-sun
distance.
Choosing realistic values of A=0.3
and qo=1367W/m
2
(following
11The references to earth and its primary greenhouse gases could easily be removed to allow the
definition of greenhouse effect to apply to other planets as well.
2
Greenhouse theory
21
The mean surface temperature of the earth
Hansen et al. 1981) yields Te=255K.
The 33K difference between this and the blackbody value of
Greenhouse
255K is due to the greenhouse effect of trace gases and clouds.
gases cause the infrared opacity of the atmosphere to increase, and the mean
To achieve radiative balance, the
radiating level to be above the surface.
is about 288K.
temperature of the surface and atmosphere must increase until the emission of
1
radiation from the planet again equals the absorbed solar radiation. 2
Jean-Baptiste
mathematician
The
the
described
Fourier
greenhouse
Fourier essentially understood that the earth's
atmosphere is largely transparent to solar radiation, while it retains some of
Early investigations of this
the longwave radiation emitted from the surface.
basic greenhouse effect were concerned with explaining why the temperature
effect in 1827 (Fourier, 1827).
at the surface of the earth is as warm as it is.
the greenhouse
effect
is
as
as
important
Ramanathan (1988)
in
solar energy
notes that
maintaining
the
temperature of the planet.
change
Greenhouse
2.2
'Greenhouse
change'
refers
an
to
alteration
of the greenhouse
effect
to the time dependant build up of greenhouse gases (carbon dioxide,
chlorofluorocarbons, methane, etc.) in the atmosphere, that perturbs the
related
climatic
the
system
terms
from
greenhouse
greenhouse
climatoligical
effect
is
global
effect
well
field, whereas
radiative
and
equilibrium. 1 3
greenhouse
understood
and
greenhouse change
change
widely
The distinction between
is
since
the
in
the
understood,
and
useful,
acknowledged
is less well
1
still disputed in various details within the climate community. 4
Hansen et al. (1981) describe a useful analogy to the surface temperature resulting from the
greenhouse effect in terms of the depth of water in a leaky bucket with constant inflow rate. If the
holes in the bucket are reduced slightly in size (increased concentration of greenhouse gases), the
water depth (infrared opacity) and water pressure (tropospheric temperature) will increase until the
flow rate out of the holes (the radiation leaving the planet) again equals the inflow rate (the radiation
entering the planet).
13
Greenhouse change could be defined more broadly to refer to changes in trace gas concentrations on
paleoclimatological time scales, leading to warming or cooling. In this paper it will be used mostly in
the sense of recent changes (plus and minus one hundred years or so from the present), and so refers to
the largely anthropogenically induced greenhouse change.
14
1t does not enhance the policy debate to lump the greenhouse effect in with greenhouse change,
thereby attributing unnecessary uncertainty to a part of theory that has been widely accepted.
12
2
Greenhouse theory
22
(unpublished)
Tyndall
15
Chamberlin
and
(1896)
and later Arrhenius
were the first to apply the notion of greenhouse change by considering
In particular, they
changes in atmospheric carbon dioxide concentration.
(1897)
were
carbon
invoking
dioxide
for understanding
mechanisms
The
variations
dioxide
in carbon
changes
epochs.
glacial
past
to
related
past
in how
interested
is
climate
greenhouse
now
theory
variations
of the
the
of
were
change
of climate
one
as
viewed
concentration
major
(Ramanathan,
past
1988).
With the
work
of Arrhenius
greenhouse
(1908)
began
change
to be
viewed as a possible consequence of human activity as well as being related to
George Callendar circa 1940 was the first to
longer term natural variations.
Callendar
suggest that human activity had already led to greenhouse change.
effectively
synthesized
emissions,
atmospheric
information
on
CO 2
CO 2 's radiative characteristics, and
fossil
CO 2 concentration,
consumption
fuel
and
He combined
marine and land surface temperatures between 1880 and 1935.
these observations with simplified models of air-sea exchange of CO 2 and
surface
radiation
energy
balance
climate
models
to
a
develop
theory
of
greenhouse change, the essence of which is yet to be disproved (Ramanathan,
The theory contends that increased atmospheric CO 2 concentration can
1988).
be attributed to CO 2 released by fossil fuel combustion, since the exchange of
C 02 between the air and the intermediate ocean waters is not fast enough to
absorb the excess CO 2 from industrial activities. Further, CO 2 strongly absorbs
radiation in the 12-17pm region, and hence the increased CO 2 would radiatively
heat the surface.
The increased CO 2 heating is then invoked by Callendar to
help explain the observed global warming. 16
Subsequently,
short term global surface air temperature
1970) levelled off somewhat, and probably
changes
(1940-
a decline in
helped contribute to
attention to Callendar's
theory.
Following the work of Plass, Revelle, Suess,
the
1950's,
greenhouse
and
Keeling
in
theory
underwent
refinement with the paper by Manabe and Wetherald (1967).
a
significant
They utilized a 1-
15
The reference to Tyndall's involvement here is speculative (Handel, personal communication).
It is interesting to note that Callendar is not distressed by the possibility of warming, since like
Arrhenius (1908), he views the warming as protection against pending ice ages.
16
2
Greenhouse theory
23
D
model to demonstrate
radiative-convective
strongly
are so
coupled by
forcing
greenhouse
relevant
convective
heat and
moisture
the
surface
governs
that
and troposphere
that the surface
transport that the
warming
radiative perturbation of the troposphere as well as that of the surface.
D radiative-convective models opened the way for the development
the
is
The 1of 3-D
climate models, and still form an important tool for interpreting the 3-D model
results.
For a comprehensive
of
description
greenhouse
theory,
the reader
is
We shall proceed with a basic
referred to Ramanathan et al. (1987).
description of a 3-D climate model, and model CO 2 equilibrium experiments.
The particular NCAR, GFDL, and GISS climate models will be described in the
chapter on models.
2.3
will
Climate
models
and
climate
Before discussing the role of climate models in greenhouse theory, we
briefly outline the approach taken in constructing a 3-D model of the
According to the traditional approach (outlined in Schneider
and Dickinson, 1974), a climate model attempts to account for the basic factors
governing climate (solar radiation, ocean circulation, ice cover, vegetation,
winds, etc.) by relating them by the equations expressing conservation of
climate system.
mass,
momentum,
and
energy,
together
with
the
thermodynamical
and
chemical laws governing the change in material composition of the land, sea,
The set of coupled non-linear three-dimensional partial differential
and air.
equations yielded by the theory are solved subject to the external input of
solar radiation,
and for a given initial
state of the earth-atmosphere
system.
In principle, the time dependency of the equations allows for modelling of the
evolution of the dynamical and thermodynamical state of the atmosphere, land
The equations are solved on a three-dimensional grid representing
and ocean.
Some models use a combined grid and spectral
Typical GCM
computational efficiency and accuracy.
latitude, longitude, and height.
representation
for
resolutions are at best a few degrees of latitude and longitude horizontally, and
Processes occuring on scales larger then
a few kilometres or so vertically.
these
are
explicitly
treated,
whereas
phenomena
occuring
on smaller
2
scales
Greenhouse theory
24
such
as
cumulus
convection,
condensation,
transport,
and
cloud
be parameterized. 17
formation must
The task of modelling climate
enormous
turbulent
number of operations.
on a computer involves performing
Following
Stone
an
communication),
(personal
consider solving for say 10 primary variables over a 3-D grid of 100 latitude
points x 100 longitude points x 10 vertical levels.
This yields 106 unknowns at
For 15 minute time steps over a typical simulation period of say
each time step.
Large computers
30 years, the number of unknowns to solve for is about 1012.
are required for such an undertaking.
Schneider
appropriate
and Dickinson
spatial
for
technique
parameterization
and
resolution
"the technique
note that
(1974)
representation,
use
in
the
time
step
of these elements,
along
factors, determines the model."
with
the choice
The use of 3-D
models is a relatively new
2.4
(1975)
among the
For a good description of a single 3-D climate model, Hansen
in further details.
experiments
chemical
a good overview of climate modelling for those interested
Dickinson provides
(1983)
and
of physical
The article by Schneider and
first of the 3-D model papers in the literature.
al.
and
and the particular
addition to greenhouse science, with Manabe and Wetherald
et
size,
approximate
of
construction
theories of climate is a large part of the art of 'modeling',
choice
of selecting
is
comprehensive
and
includes
description
of
sensitivity
as well.
Greenhouse
change
modelling
experiments
To model the response of the climate system to greenhouse forcing, the
amount of CO 2 and other trace gases is simply increased in the radiation
scheme
of the model.
atmosphere,
which
in
This
turn
reduces the amount of energy
alters
the temperature
and then
leaving the
other climate
variables in the model such as winds, cloud cover, and precipitation patterns.
All that modellers require to generate the response of the system to an initial
perturbation
is
fundamental
equations,
17Parameterization
the
ability
to
in this
calculate
the
case absorption
physical
forcing
of radiation
terms
by gases
in
the
in the
is short for parametric representation.
2
Greenhouse theory
25
for conservation
equation
performed
a transient
as
energy
infancy however.
a transient
(Rind
experiment,
system to the forcing.
the climate
et al.
demonstrating
Transient
Ideally,
1988).
the gradual
are
experiments
this
is
response
of
still in their
represents the first published results of
Hansen et al. (1988)
experiment.
of
consistency
For
selected
of
from
results
across
comparison
the more traditional
modelling
we
groups,
equilibrium
GCM CO 2
have
response
In the equilibrium CO 2 response experiments, the model
experiments to study.
is first run for a number of years (typically 10 or 20 years of model time) with
The last ten years
boundary conditions and forcing factors as they are today.
or so of this control run provides an important test of how well the climate
model
simulates the
response
experiment
real climate
is to introduce
conditions or forcing factors.
doubling
instantaneous
The next step in the equilibrium
system.
a perturbation
In the experiments
carbon
of atmospheric
of the boundary
of one
considered here, this is an
dioxide
concentration.
18
Note
simulation of current climate in the control run is important for
experiment, since the doubling of CO 2 induces a perturbation on the
that accurate
the
If the current climate is not correct, then the accuracy of the
current climate.
perturbed climate
linearities
are
simulation may be called
important
and
cannot
into question in so far as non-
be linearized
with
sufficient
Because of the importance of the control run in the CO 2 sensitivity
accuracy.
experiment,
later analysis of transports in the major 3-D models will be for results from the
control
run
After
simulations.
instantaneously
doubling
number of years for the perturbed run.
CO 2 , the model is again run for a
The difference between the perturbed
run and the control run is a measure of the sensitivity of the model to the
The commonly quoted
prescribed change between two equilibrium conditions.
2 to 5K temperature change for a doubling of CO 2 in the atmosphere is based on
the results of these equilibrium experiments with GCM's. 1 9
18
The change of carbon dioxide in the models is usually taken as a surrogate for a change in all of the
greenhouse gases (Kellogg and Zhao, 1988).
19
The global surface air temperature changes resulting from these experiments for the models
considered in this work are: NCAR:3.5K; GFDL:4.OK; and GISS:4.2K, (and for UKMO:5.2K).
2
Greenhouse theory
26
of
doubling
CO 2
of
use
make
frequently
change
the
advocating
Arguments
for
the
GCM
predictions
temperature
Current
in the atmosphere.
greenhouse
to
responses
policy
need
GCM predictions
for
a
are rather
The greenhouse
sobering, and are quoted to stress the gravity of the problem.
policy arguments usually note that if the temperature response is as predicted
by the GCM's,
change will be virtually without
then the rate of temperature
precedent, and may have major implications for ecosystems and social systems.
We note then, that some arguments for greenhouse policy responses are
on the results of GCM model simulations, and so
predicated to some degree
investigation of the credibility of the models is a policy-relevant pursuit.
2.5
change
Greenhouse
controversies
From a policy perspective, a lack of consensus in a scientific community
on a particular theory weakens
seriously,
and
of scientific
part
are
disputes
Since
implemented.
policies
will be taken
the likelihood that the theory
development and will always exist to some degree, policymakers must ascertain
whether existing disputes are significant or relevant to the range of possible
policy options or not.
With this in mind, we turn to a discussion of greenhouse
disputes.
theory
In
greenhouse
theory,
the
effect
greenhouse
undisputed.
is virtually
With respect to greenhouse change, the important parts of the theory are the
determination of the greenhouse forcing, and the climatic response to the
forcing.
trace gas concentration
about.
of the radiative forcings associated with changes in
The magnitudes
However,
are reasonably
greenhouse
change
well understood,
is disputed
and not much argued
as to whether the climatic
change resulting from the greenhouse forcing will be as large as the models
predict, and as to whether the change is yet detectable.
Criticizm
crudity
of the
(ocean response,
models
in
neglecting
or
results
from
some simple models.
conflicting results, the work of Idso (1980)
surface
various
oversimplifying
clouds and moist dynamics, vegetation, hydrology,
ocassionally on conflicting
using
is based on the
of the magnitude of the model predictions
energy
balance
and Newell
considerations
stands
processes
etc.),
In regard
and
to
and Dopplick (1979)
out.
2
These
works
Greenhouse theory
27
less temperature
an order of magnitude
predicted
doubling
response to CO 2
These results have been explained by Ramanathan (1981) on
than the GCM's.
the basis that the "important tropospheric CO 2 radiative heating is ignored by
both Newell and Dopplick (1979) and Idso (1980)."
Probably the most potent criticizms related to global scale shortcomings
in the GCM's involve the representation of clouds and the role of cloud
For the current climate, Ramanathan et al. (1989) find that clouds
feedbacks.
play a major role in determining the radiative balance of the atmosphere, and
That is, clouds presently cool the
presently constitute a negative feedback.
than
they
radiation).
it
heat
The
question
reflecting
greenhouse
their
(through
crucial
in
effect
albedo
their
(through
planet
regarding
radiation)
shortwave
change
greenhouse
longwave
trapping
in
effect
more
is how
the
For the current
cloud mix would change in response to greenhouse forcing.
GCM's employing predictive cloud schemes, cloud changes provide a
significant positive feedback.
small increase
For the GISS model this occurs as a result of a
greenhouse
(enhancing cloud
in mean cloud height
effects) 2 0 ,
and a small decrease in cloud cover (decreasing their albedo effect) (Hansen et
Since current cloud schemes in the models are quite crude and cloud
al. 1984).
contain
significant uncertainty.
the predictions,
to
significantly
contribute
feedbacks
Uncertainty
alone
the
model
predictions
implies nothing
about
the
direction of potential error in the feedback, though there are some indications
in this regard, which we will consider.
feedback can occur as a result of changes in cloud
The 3-D models consider (albeit crudely)
and height.
Changes in cloud
composition, coverage,
As mentioned, the
changes in cloud coverage and height, but not composition.
changes in coverage and height in. the models are such as to lead to positive
As regards cloud composition, it is thought that the greater
feedback.
availability
of
water
vapour
for
condensation
in
a
greenhouse
warmed
atmosphere could lead to a systematic increase in cloud liquid water contents,
though the relevant physics are not well understood (Somerville and Remer,
1984).
In the 1-D
radiative-convective
model of Somerville
and Remer, the
20
As cloud top height is increased, the cloud top temperature becomes progressively lower than the
surface temperature, reducing the emission of longwave radiation from cloud top to space, and thus
enhancing the radiative heating of the troposphere and surface.
2
Greenhouse theory
28
of increased
cirrus)
is to increase
Mitchell
(1988)
In the
there
Remer,
temperature
of climate
is
limited observational
liquid
water
Somerville
content
with
above 273K.
et al.
by Twomey
mechanism was proposed
Another negative feedback
Twomey pointed out that an increase in tropospheric
(1984).
two.
and Remer may
data employed by
cloud
in
change
little
factor of
a
by
sensitivity
water content depends on factors other
correct, cloud liquid
than temperature.
and
decrease
notes that although the results of Somerville
be qualitatively
than thin
of clouds more than the greenhouse
the albedo effect
possible
and
other
(for clouds
This yields a negative feedback for cloud optical thickness
effect of clouds.
changes,
liquid water content
cloud
result
particulates will
lead to an increase in the number density of cloud condensation nuclei, which
A variant of this feedback involving
can enhance cloud optical depth.
biogenic processes was proposed by Charlson et al. (1987).
Over the oceans, the
number density of cloud condensation nuclei is influenced by the emission of
An increase in surface temperature
dimethyl sulfide from marine organisms.
could lead to an increase
in dimethyl
processes
are
certainly
plausible,
and thus also to an
notes that while the above
Ramanathan (1988)
increase in cloud optical depth.
microphysical
sulfide emissions,
the
of these
extent
global
demonstrated.
processes has not yet been satisfactorily
Some of the cloud physics not included in the model cloud schemes could
provide significant additional positive feedback also (Ramanathan and Handel,
communication).
personal
extent
and
longer
For instance,
allowed
lived than
vertical
heat
cirrus clouds may be of larger areal
for
transport
which
in the models,
than
the
microphysics
are
more
of
cloud
concerned
with
formation.
Cirrus shields would probably provide an additional positive cloud
feedback due to the dominance of their greenhouse effect.
In a
(1979)
1979 assessment
allowed
of cloud
a +1 0 C error in CO 2
uncertainties in high cloud effects.
temperature,
feedback effects on
doubling
response
temperature
NRC
due
to
Writing prior to such work as Somerville
and Remer (1984) and Twomey et al. (1984) they were unable to find evidence
for an appreciable negative feedback due to changes in low and middle cloud
2
Greenhouse theory
29
0
albedos or other causes, and allowed only 0.5 C as an additional margin for
2
error on the low side due to cloud feedback uncertainty. 1
The point of this discussion of cloud feedbacks is to illuminate the
controversy and complexity, and to recognize that criticizm of the magnitude
of the warmings produced by the models is not unfounded, but doesn't
necessarily
invalidate
the predictions
either.
Other areas in which GCM's are criticized is over the representation of
oceans, ond over the validity of equilibrium response as opposed to transient
Schneider and Thompson (1981) attemped to weigh the
seriousness of deliberately neglecting heat capacity of the deep oceans in most
models and of using a fixed-CO 2 increase instead of a time evolving scenario of
C 02 increase. The actual mixed layer of the oceans is generally deeper in high
response experiments.
latitudes than in low latitudes, and would therefore tend to heat up more slowly
Thus, although albedo/temperature processes would cause
in high latitudes.
high latitudes to warm more in equilibrium, mid-latitudes and some high
latitude
regions
would
not necessarily
warm up
as
fast as
lower latitudes
Just how important
during the transition period towards the new equilibrium.
actually
this process is depends on how rapidly the CO 2 concentration
If CO 2 increases rapidly, neglecting the transient would
increases with time.
be a more serious error than if CO 2
Schneider (1984)
increases
notes that atmospheric
more
addition,
In
slowly.
with
water vapour increases
surface
temperature increase, accounting for perhaps 25 to 40% of the equilibrium
During the transient warming
surface temperature response to CO 2 increase.
period, oceanic response may well lag that of the atmosphere, thereby
temporarily
reducing
occur after equilibrium
the
water
vapour
increment
that
would
eventually
was reached.
The transient warming response would also be affected by the rates at
which water from the ocean mixed layer mixes with water below it, and by
changes in the rate in response to climatic change. In the Hansen et al. (1988)
transient response experiment, the uptake of heat perturbations by the ocean
Bryan and
beneath the mixed layer is approximated as vertical diffusion.
21
0
Their overall estimate considering all sources of uncertainty was 3*C±1.5 C.
2
Greenhouse theory
30
(1985)
model.
During the greenhouse warming episode, their model yields a partial
collapse
of the
up
twice
investigate
thermohaline
ocean
much
as
as
heat
conclude
that
"an enhanced
feedback
for
greenhouse
circulation
thermohaline
circulation 2 2 , allowing the ocean to take
predicted
be
would
found in the numerical
simple
diffusive
and
Spelman
Bryan
the
However,
a
would produce
of heat
sequestering
warming.
by
conditions.
climatic
normal
under
approximation
CO 2 perturbations
in a coupled ocean-atmosphere
Spelman
a negative
would also affect
experiment
strong as
carbon cycle, possibly producing a climatic feedback as
the global
of the
collapse
partial
Broecker
that caused by an enhanced uptake of heat from the atmosphere."
outlines ocean circulation processes not included in the climate
(1987)
models
He
with simple mixed layer oceans, that could lead to abrupt climate changes.
against
warns
that the
assuming
climatic
response to
increased
greenhouse
gases will be a relatively smooth gradual warming over a period of about 100
Stouffer
In initial experiments with a coupled ocean-atmosphere model,
years.
et al. (1989) report at least one surprise in the model simulation.
the upwelling
of cold
response to greenhouse
related
in the
water off the coast of Antarctica
model
in
forcing, leading to cooling in that region.
Though transient response experiments
the uncertainty
This relates to
to the crude
are in the offing now, much of
of the oceans
representation
in climate
models will exist for decades to come, and we should not expect to see quick
improvement over the model simulations discussed here.
of controversy
other major area
The
relevant to greenhouse
The detection problem is
over whether greenhouse change is yet detectable.
incredibly
difficult
since
currently
available
would
estimates
rough
the
have
the
greenhouse
of
to
variations,
quantify
volcanoes,
climatic
etc.). 2 3
variations
due to
response
climate
signal
perhaps
only
times
just
In addition, it is
beginning to emerge from the background of climatic noise.
difficult
theory is
other forcing
factors
(solar
Thus, it is difficult to provide unambiguous
answers to the detection question at this stage, though it is expected that it will
become easier with time as the greenhouse signal emerges
more clearly from
22
The thermohaline circulation is the circulation that is driven by the buoyancy flux in the ocean.
Hansen et al. (1981) have attempted this exercise, but have been criticized in Clark (1982) for using
up all available degrees of freedom in fitting observations.
23
2
Greenhouse theory
31
For a more indepth discussion of greenhouse detection
the background noise.
and Ramanathan (1988).
problems, refer to Clark (1982)
Detection
temperature
efforts
signal,
and
particularly
at
surface
the
mainly
concentrated
have
controversies
(air)
for
on
the
average
the global
24
The major recent compilations of
over the last one hundred or so years.
global surface air temperature of Jones et al. (1986) and Hansen and Lebedeff
This warming has
show a warming of about 0.5K for the last century.
been called into question mostly on the grounds of possible contamination by
urban heat island effects from growing population centres over the period.
(1987)
Both the Jones et al. and Hansen and Lebedeff works have included attempts to
screen out data contaminated by urban effects, and to remove any remaining
Analysis of this processed data for urban
urban heat island effect bias.
warming contamination has been carried out mostly for the U.S. portion of the
globe (which represents perhaps an upper limit to the degree of
Karl and Jones (1989) find the urban bias to be a substantial
contamination).
portion of the overall trend for the U.S. in the Jones et al. and Hansen and
Lebedeff data sets (but this trend is still small relative to the global trend).
However, subsequent reworking of the U.S. data by Hansen et al. (1989) (where
the 'et al.' includes Karl) found that the urban bias was not as significant as
determined by Karl and Jones. The Karl and Jones work was in error due to an
incorrect mapping of the U.S. temperature data in one of the data sets (Hansen,
personal
communication).
Though the detection
attention,
give
problem
is difficult,
of a greenhouse
since unambiguous detection
added impetus
policy iniatives
to the credibility
as well.
believe that greenhouse
climate
change is here,
of
signal would
and therefore to
of greenhouse theory,
It is probably
the centre
it has been
reasonable to surmize that if people
they are more likely to believe the
model predictions for how further changes will take place, and more likely to
Conversely, if a clear refutation of any detection of
press for policy responses.
expected
greenhouse
climate
signals
could
be given,
would
this
undermine
credibility of the theory, and tend to dampen greenhouse policy initiatives.
24
The global scale is the best place to look for greenhouse signals, since it is the least noisy.
2
Greenhouse theory
32
Among climatologists, Hansen (1988)
greenhouse
sufficiently
cause
detection
large
and effect
literature
journal
that
25
in
we can
relationship
and
press
that
claiming
"the
warming
global
greenhouse
of
coverage
that
Recent
effect."
literature
that belief in detection
signal has for credibility
a
greenhouse
26
and
Because
of a greenhouse
or lack thereof
and policy
of the theory
now
reacted
has
responded directly and indirectly to the above statement of Hansen.
of the importance
is
of confidence
with a high degree
ascribe
to the
is perhaps the most outspoken on
responses,
we
will pay
particular attention to this issue in the chapter on presentation of theory.
The greenhouse detection issue is also impacted by a problem regarding
the nature
system.
of the
and policy
system,
climate
responses
to
in the
changes
The problem stems from the fact that the time scales for response of
the climate system to greenhouse forcing are relatively long - of order decades
1985), so that by the time greenhouse change can
to centuries (Hansen et al.,
be detected in a manner convincing to all, the earth will already be committed
to a further greenhouse warming by the trace gases currently present in the
atmosphere.
Thus, Ramanathan et al.
(1989) note that "even if the trace gas
concentrations were to stop increasing today, the planet will still warm by 0.82.4K.
(The three-fold
predictions.)
If, on
range is the currently perceived uncertainty
the other hand,
the committed
continue
unabated,
years."
The policy leverage
actually quite significant.
warming
available
Hansen
the
currently
can
observed
double within
in model
growth
rates
the next
50
in reducing the committed warming
is
rates
of
(1989)
notes that if the growth
greenhouse gases, CFC 11 and CFC 12 had not been reduced in 1978 as a result of
the aerosol CFC ban in some countries, the decadal greenhouse forcing
produced by CFC's would currently be as large as that due to CO 2 . As it stands,
the CFC contribution to the decadal forcing is about half that due to CO2.
25
Among climatologists, Hansen is also somewhat unique in having directed a concerted, sustained,
and productive effort to theoretical, modelling, and observational aspects of greenhouse theory.
Readers should note the possibility for bias here in the author's institution's cooperation with the
NASA GISS group, but we feel that the above comment is reasonable nonetheless.
26
See Kerr (1989) for one version of the controversy surrounding Hansen's statements.
2
Greenhouse theory
33
2.6
Greenhouse
problems
In this work, we will not consider those aspects of the greenhouse issue
that
might
under
fall
problem(s)'
refers
ecosystems
and
communication),
loosely
social
greenhouse change.
the
term
to
systems
'greenhouse
difficulties
due
to
'Greenhouse
problem(s)'.
resulting
climate
from
alteration
of
on
by
brought
change
The plural form follows a distinction by Clark (personal
that the resulting
problems
from
greenhouse
change
will be
very different for different people in different places.
2
Greenhouse theory
34
VALIDATION OF CLIMATE MODELS
3
Approach
3.1
is crucial
climate models
of 3-D
Validation
greenhouse
to validating
change predictions because of the central role of the 3-D models in generating
the
employed
Here we present a brief review of strategies
predictions.
validate the 3-D climate models.
Before proceeding however, note the point
that " 'model validation' is a misnomer because it is
made by Keepin (1986)
The most one can do is to
actually impossible to show that a model is valid.
Thus model
show that it is not invalid, and even this can be a formidable task.
condition
only be verified by
thorough
circumstantial
systematic,
evidence
et
(Hansen
model
with
dynamical processes.
such
as
variances
validation
that
so
al.,
basic
1983)
or
saved
observations
of
and
climate
opportunities
by
the models.
it can
to
has
to
generate
neither
necessary
With the exception of the GISS
climate
for
been
models
the
diagnostics
models
to
internal
ensure
have
been
not
compare
model
consistency
of
Mean model fields are saved, but other basic statistics
of model variables have not generally been saved or
published from model runs in order to examine climate variability
27
notes
Schneider (1987)
evidence."
have been missed.
calculated
systematically
performance
circumstantial
however,
general
nor
sufficient
of a climate model cannot be proved conclusively;
that "the accuracy
In
not
but
necessary
a model".
the results from
for 'believing'
a
establishing
to
equivalent
is
validation
to
The reasons for a less than comprehensive
as simulated
approach
to
model validation among most of the modelling groups are not clear. It may be
a low priority because of resource or funding constraints, or perhaps even
because it appears
to be less interesting than more direct model development
projects.
Validation of 3-D climate models is made difficult by the fact that we do
not have detailed knowledge of anything but the present climate state and its
27Steps in this direction have now been taken with the publication by Rind et al. (1989) for the GISS
model.
3
Validation of climate models
35
seasonal changes to compare with the models.
Knowledge of past climates is
lacking in detailed information, and future climatic states have not yet evolved
weather
forecast
models.
In
of validating
Contrast this with the luxury
predictions.
to test the model
weather
the
forecasting,
initial
of the
state
atmosphere is determined as well as possible, then the model is run out for a
Subsequent to the forecast,
week or two.
of the actual
detailed knowledge
evolution of the atmosphere over the period of the forecast can be collected,
and
direct
a
validation
of the
transient
is
forecast
analogs of various kinds.
The variety
a
such as this
satisfactory direct verification of 3-D climate model performance
is not possible at the present time,
While
possible.
involving
indirect methods can be used
We shall discuss some of these presently.
tests that are left once
of validation
be grouped into three
with the transient climatic response is ruled out can
categories (following Wigley and Santer, 1988).
a direct comparison
These are, validations of the
of the models, validations
internal physics and subgrid scale parameterizations
against present climate with the model in the control run mode, and validation
We will touch on each
against other climate states in the perturbed-run mode.
of these strategies, before
Validations
in each
transports.
validation of model energy
considering
in order to
are necessary
of the categories
gain some
confidence in the credibility of the 3-D models, though validations in a single
category alone are not sufficient to demonstrate the veracity of the models.
As
we proceed, we will rediscover Keepin's point that it is possible to devise many
testable
necessary
conditions
to
validate
testable
sufficient
conditions.
This
is
greenhouse
an
important
but
theory,
point
to
no
real
grasp,
and
implications of this for policy will be discussed in later chapters.
3.1.1
Internal
For
internal
parameterization
validations
validations,
the accuracy
each
of
internal
process
with observations,
is tested separately by comparison
or
or by
comparison with the results of more detailed models of these processes. Prime
candidates for internal validation ate the radiative transfer schemes, and
subgrid scale details such as cloud and land surface processes parameterizatons
(Wigley and Santer, 1988).
3
Validation of climate models
36
recent
A
been
has
effort
organizational
towards
directed
intercomparison of radiative codes in climate models (ICRCCM) (Luther et al.,
The relative accuracy of various radiation codes can be assessed via
1988).
ICRCCM,
but
prevented
by
Outstanding
a lack
uncertainties
compare
to
is
with.
to the treatment of
water vapour uncertainties
of the major role played
by water-vapour
sensitivity.
feedback in climate
The
The
schemes
the
of
data
schemes relate
and methane.
important because
accuracy
observational
in the radiation
water vapour, carbon dioxide,
are particularly
absolute
quality
high
of
the
of
determination
limitation of internal
validations
in isolation is that they cannot
guarantee that the complex interactions of many individual model components
To investigate these interactions it is necessary to
are properly treated.
consider the response in the control run and perturbed run modes.
Perturbed-run
3.1.2
validations
The most obvious perturbed-run
is to compare
validation technique
the
This
equilibrium model results of one 3-D model with another.
28
is not a sufficient test since the model physics across models is similar , and
Obtaining agreement in the perturbed
agreement could simply indicate this.
perturbed-run
run
difficult than
is more
in
the control-run
however,
agreement
since
of
tuning 2 9
of the
some features in the control run can be partly explained by
models about the same current climate. We will discuss some of the features of
the perturbed-run comparisons for the NCAR, GFDL, and GISS models in a later
section.
Another
the
models
analogs
follows:
in
is
approach to validating the doubled
to
compare
the past
"Perhaps
the
century.
there
is
CO 2 equilibrium results in
resulting
climate
response
with
warm
Schneider
(1984)
explains
the
rationale
something
characteristic
about
the
year
as
regional
28The conservation equations are the same for each climate model, though the parameterization
schemes vary somewhat from model to model. The physics and numerics in the NCAR and GFDL model
are relatively similar, but quite different from the GISS model.
29
Tuning refers to the adjustment of empirical parameters that govern model-simulated processes
which are not calculated by integrating conservation laws.
3
Validation of climate models
37
climatic patterns in 'warm years', and that the distribution of regional climatic
anomalies in the warmest few years of the past century might be repeated
long-term warming were to be created in the future by CO 2 increases."
such
as
by
that
Pittock
and
have noted
(1982)
Salinger
if
Studies
some broadscale
A problem with this approach however is
agreement with such an approach.
that "a CO 2 induced warming will not necessarily result in similar seasonal or
regional
patterns
increase
from human
years
of warmer,
drier climates,
and
external
forcing
century
could
well be
internally
looking
by
since
warm
individual
rather
caused
CO 2
than
showed that inferences of cause
Analogously, Lorenz (1979)
those described
or
is an
can be very different
climatic variations
for longer time 'forced'
and effect
wetter,
activities
of the twentieth
externally forced.
from
cooler,
variations"
climatic
at shorter period 'free'
1984).
(Schneider,
chosen from
the paleoclimatological
of
reconstruction
record.
Hansen et al.
patterns
in
climate
regional
can be
of warmer periods
analogies
warm years,
Instead of choosing
the
note that
(1981)
(5000-9000
altithermal
years ago, when the earth was a degree or two celsius warmer than at present)
Manabe and Bryan (1985)
show some similarity to 3-D climate model results.
used a coupled ocean-atmosphere climate model with simplified geography and
various CO 2 perturbations from 1/2 to 8 times the present value to consider
implications.
paleoclimatic
signature
obtained
hypothesis
for
from
the
the
climatic
They
found
model
appears
in
changes
the
exception: the tropical sea surface temperature
significant
increase
with
increasing
"in
that
to
be
general,
consistent
Cenozoic
with
the
climatic
with
a
the
CO 2
following
in the model has a small but
atmospheric
CO 2
concentration,
while
tropical sea surface temperature as deduced from the isotopic record appears to
have no systematic trend during the Tertiary."
Paleoclimatic
validation
have
studies
numerous
problems
associated
with them, not least of which is the problem of separating different forcings.
In addition,
the time scales of the relevant
Anthropogenically
100 years,
induced
greenhouse forcing occurs
while longer period forcings
time scales of 10000 years or longer.
series
analysis
of various
forcings may well be different.
climatic
such as
over
a period of order
orbital variations occur on
This is significant, since detailed timeindicators
has
shown
3
that
the climatic
Validation of climate models
38
response
lags
behind
the
forcing,
and
that
dependent and non-linear (Imbrie
and Imbrie,
warming
time
on
contempory
paleoclimatological
greenhouse
forcing,
the
response
1980).
scales
frequency
Though the causes for
may
Kellogg and Zhao
is
be
different
(1988)
note
from
that
the
warmer
climatic periods do have one important feature in common with a greenhouse
warmed earth.
This
is that "the equator-to-pole
temperature
difference
was,
and will be in the future, less than the present norm, and the large scale
circulation
patterns
may
therefore
respond
similarly
to
this
change
in
forcing."
3.1.3
Control-run
validations
The basic rationale behind control run validation is that if a model fails
to simulate important features of the observed climate in the control run mode,
then it will also fail to simulate these features in a perturbed mode.
This may
not be fatal since it is the difference between the runs that counts.
However,
non-linearities
such
as
the
ice-albedo
feedback
mechanism
problems for the perturbed run if the control run is in error.
non-linearities
cannot
be
linearized
with
sufficient
could
cause
That is, if the
accuracy,
then
it
is
important to have the simulation of the control climate at least approximately
correct.
Validation of the current climate simulation
advantage
that
climate.
Features
current climate.
surface
it
can
be
compared
directly
in the control run has the
with
observations
of the
in the model can be tuned to match observations
real
of the
For example, the surface albedos can be tuned to ensure that
temperatures
are
in reasonable
agreement
with
the
reflecting more or less radiation from the surface as desired.
observations
by
The models are
just too complex however for modellers to be able to tune all variables to match
observations, so validation of the control run is still a useful exercise.
In later
consideration of the control run simulation of energy transport in the models,
we will compare the transports with observations, and between models.
Different climates can still be considered in the control run by looking
temporally
at the simulation of seasonal climate features,
simulation
of regional
average
climates
in the model,
3
or spatially
and comparing
at the
with
Validation of climate models
39
The rationale for this approach is that a good seasonal model
observations.
forced by time varying solar insolation may also be a good greenhouse forced
model.
Stone et al. (1977) were the first to test this approach on a 3-D GCM,
(a model distinct from the
using the GISS model of Somerville et al. (1974)
July and January model simulations were compared with
current GISS model).
data on seasonal
other and with climatological
each
Stone et al.
changes.
accurately the "northward displacement of the
found that the model simulated
mid-latitude jets, the low-latitude Hadley cells, the tropical rain belt, the trade
winds, and the ITCZ in July compared to January, the reversal of the Indian
monsoon,
the weakening
and
of the
zonal
meridional
Manabe and Stouffer (1980)
decline of eddy ectivity in the summer."
and the
circulations
also used
a 3-D climate model to compare the regional pattern of seasonal temperature
range
a 68m deep passive thermodynamic mixed layer ocean, which
study employed
allows
tuning
the
of
average
range
temperature
seasonal
processes
and is not
the model
exceptions,
of
in the model results
distribution of response
and
seasonal
Manabe
tuned.
"that with
and Stouffer found
of
variation
geographical
scale characteristics
observed
the
some
atmospheric
points out that "it is necessary that a valid
Schneider (1984)
temperature."
from internal model
in reproducing the large
succeeds
(though
Nonetheless, the
observations suggest that this choice of depth is reasonable).
geographic
Their
simulation with observed values.
from a seasonal solar forcing
model reproduce seasonal climatic features, but that this may not be sufficient
to engender much confidence that the model will also have a valid response to
a different forcing, like CO 2 increase.
encouraging
on
the
'natural
experiment'
the GFDL results were
Nevertheless,
of
seasonal
and
forcing
response,
'passed' this necessary conditions test."
having successfully
Another useful test of a model's control climate is examining how the
variability
simulated by the model compares with the variability
the observed climate.
control
run lends
simulation
of
perturbed
run.
determining
the
climatic
displayed by
The ability to reproduce the observed variability in the
confidence
to
a models'
variability
associated
Knowledge
of
impacts
on
ability
with
climatic
ecosystems,
extremes that produce the most damage.
a
to provide
changed
variability
since
it is
is
a reasonable
climate
in
the
important
for
often the
climatic
Validation of climate variability has
been carried out for the GISS model by Rind et al. (1989) for temperature and
3
Validation of climate models
40
precipitation.
observed,
They find
especially
in
that
late
"the modeled
summer,
variability
possibly
due
is often
to
the
larger
crude
than
ground
hydrology" in the model.
Validations of the control run with current climate and between models
have
been carried out considering most of the climatically
relevant
variables
(temperature, pressure, precipitation, soil moisture, cloud cover, etc.).
Some of
the results of these comparisons will be presented in the next section as they
relate to the NCAR, GFDL, and GISS models.
3.2
NCAR, GFDL, GISS model validations
Direct comparisons of the NCAR, GFDL, and GISS model control runs with
observations
have
been
made
by
the
modelling
groups
in
presentation
of
The papers relevant to the versions of the models considered in this
results.
NCAR (Washington and Meehl, 1984), GFDL (Manabe and Wetherald,
work are:
Without considering the results in detail
1987), and GISS (Hansen et al., 1983).
here, Hansen et al. find that "it is verified that the major features of global
climate can be realistically simulated with a resolution as coarse as 1000 km."
Washington
and
Meehl
find
and many observed
simulation
"good
agreement
between
A
particular
quantities."
model is that "sea ice forms equatorward
the
of its observed position partly as a
On a broad scale,
reasonable agreement is obtained by the GFDL model.
The UKMO
model is not considered in this work, but Wilson and Mitchell (1987)
conclude
with the present level of
that "the model simulation of the observed climate
atmospheric
model
of their
deficiency
result of the lack of poleward heat transport by the ocean."
similarly
present
CO 2 (323 parts per million by volume (ppmv)) is fairly good."
Moving from the broad scale down to the regional
scale, as would be
expected, model results do not compare as favourably with observations or one
another.
observations
"although
averages
compares NCAR, GFDL, and GISS model results with
Grotch (1988)
of
surface
the models
over large
temperature
often agree
areas,
spatial extent is reduced,
examined."
air
well
substantial
particularly
Wigley and Santer (1988)
and
when
precipitation
comparing
disagreements
and
seasonal
become
when detailed regional
finds
that
or annual
apparent
as the
distributions
are
consider regional scale simulations from
3 Validation of climate models
41
For pressures over Greenland
the GISS and other models.
and the seasonal
cycles of the latitude and intensity of the Iceland low and Azores high, they
find that model simulations are "generally poor."
at the regional scale is important for at least two
From the point of view of validation, errors in the smaller scales in
reasons.
the control run are important since the smaller scale processes could alter the
energy budget in such a way as to influence the simulation of broad scale
Obtaining accuracy
Accurate simulation of the regional scales would
features in a perturbed run.
also
useful
very
be
and
studies
impact
climate
for
concomitant
policy
responses.
the control run
and doubled CO 2 run in the NCAR, GFDL, and GISS models relative to one
0
They note global temperature increases of 3.50C to 4.2 C for these
another.
models,
and
11%.
to
by
the
Furthermore,
recent
the
simulations
precipitation
smaller
For
scales,
distributions of the CO 2 -
find that "the geographical
warming obtained
not quantitatively.
between
differences
of 7.1
increases
precipitation
Schlesinger and Mitchell
induced
compare
and Mitchell
Schlesinger
qualitatively
agree
and soil
moisture
but
changes
Kellogg
do not agree quantitatively and even show qualitative differences."
and Zhao (1988) examine the soil moisture response to a doubling of CO 2 in the
above three models plus those of OSU and UKMO. For North America, they note
considerable differences in the results for soil moisture distribution changes,
The areas of agreement are
but are able to note areas of agreement also.
These studies
"consistent with the results of studies of past warmer periods."
highlight the fact that surface moisture results in the GCM's are critically
dependent on the specification of soil properties and the parameterization of
hydrological processes in the models (Gates, 1989).
Gutowski
et al.
(1988)
have
examined
surface energy
fluxes
for the
control run and a doubled CO 2 run in the versions of the NCAR, GFDL, and GISS
They find that the surface energy balance is
models considered in this work.
dominated
by
longwave
radiation.
fluxes of longwave radiation
Specifically,
the upward
and downward
are the largest components (though
they act in
For the global average control climate, individual surface
agree to within 25W/m 2 , but this difference is a "large fraction of the
opposite directions).
fluxes
3
Validation of climate models
42
net longwave
radiation
at the
(-60W/m 2 ), and it is as large as any of
surface
the global average seasonal changes in these fluxes".
find
that
intermodel
note
They
changes.
are
discrepancies
apparent
that this
only
For CO 2 doubling, they
small
fractions
should
agreement
the
flux
viewed
with
of
be
caution "because flux changes associated with doubled CO 2 are about the same
size as the differences between models in their control climate: 25W/m 2 ".
difference found for the global
when
flux changes
twice the
are up to
surface flux discrepancies
intermodel
changes,
regional
average, and often disagree over the sign of
Gutowski et al. note that such differences
CO 2 doubles.
(like the
appear to be closely asociated with differences in model hydrology
soil moisture
results).
In summary,
frequently
in
for GCM validations,
to
related
differences
in
the
climates
are
parameterization
of
in control
differences
for
schemes
For the three GCM's under consideration
physical processes.
differences
For
the
representation
outlined in the chapter on models.
mode, model discrepancies
of differrent
physical
here, the major
will
processes
be
In validation of GCM's in the perturbed run
in the control
are frequently related to differences
simulations of the respective models (and thus indirectly also to the different
parameterization
schemes
employed).
GCM's simulate current climate much better on the larger scale than the
regional scale, most obviously because of the limited resolution employed by
Simulation of the doubled CO 2 climate is also more reliable on
the models.
larger scales for the same reason, and additionally (as noted by Mitchell et al.,
1987) because "the regional response of climate models to small perturbations
is shown to be highly dependent on the unperturbed simulation."
Most of the work on validating the 3-D climate models has been at the
level of specific validations of one aspect or another of model performance.
Overall
assessments
of models
U.S.
problems are less common.
1979,
and
1982
have
and
their validity
in
regard
to
greenhouse
National Research Council reports in 1975,
addressed this question.
Their conclusions
based on
earlier versions of the GFDL and GISS models and on simpler climate models
show a general
confidence
in the ability of the models to simulate broader
aspects of greenhouse change.
NRC (1975)
focused
on the current
climate
3 Validation of climate models
43
simulation by the models, and note that "the several
atmospheric
and oceanic
GCM's have now reached the point where reasonably accurate simulations of
the global distribution of many important climatic elements are possible, and
NRC
into a single dynamical system is now feasible."
(1979) concluded that "the predictions of C0 2 -induced climate changes made
with the various models examined are basically consistent and mutually
The differences in model results are relatively small and may be
supporting.
where their coupling
in
differences
by
for
accounted
and
characteristics
model
simplifying
conclude that "comparisons of simulated time means
of a number of climatic variables with observations show that modem climate
models provide a reasonably satisfactory simulation of the present large-scale
global climate and its seasonal changes", and that "model-derived estimates of
NRC (1982)
assumptions."
changes,
temperature
averaged
globally
perhaps changes
and
averaged
along
latitude circles, appear to have some predictive reliability for a prescribed CO 2
In the NRC reports, ultimate confidence in the general
perturbation."
predictions from the 3-D models is based on the fact that 3-D models verify
results from simpler models (radiative-convective and heat balance models)
that can be understood in purely physical terms.
3.3
Energy
The
transport
poleward
validations
energy
transport
on
contributions
receives
earth
from
In this work we
the atmosphere, the ocean, and drifting sea ice (Bryan, 1982).
are interested in examining the annual total meridional energy transport in
the
models
major climate
and
Validations
observations.
of climate
model
energy transports have not been extensive to date.
Observational
efforts
aimed
at calculating
energy
transports
have used
Indirect
a variety of direct and indirect methods, and combinations of the two.
methods can be used to calculate the total energy transport from radiative
fluxes at the top of the atmosphere measured from satellite observations (e.g.
This technique has also been employed by
Oort and Vonder Haar, 1976).
Carissimo
et al.
Further discussion
(1985),
whose
observational
data is chosen
of this data will be given in
for this work.
the chapter on data.
To
calculate the contributions by atmosphere and ocean to the total transport, one
component can be calculated directly from observations, and the other
3
Validation of climate models
44
component
inferred
calculation
of
as
a residual.
ocean
transport
Since
is
less
the observational
reliable
than
for
Carissimo et al. calculate the ocean transport as a residual.
of ocean transport have been made (e.g. Wunsch,
Atlantic Ocean.
data
for direct
the
atmosphere,
Direct calculations
1980),
but mostly for the
Indirect methods to calculate the ocean transport have been
based on surface heat flux considerations (e.g. Hastenrath,
1980).
Miller et al. (1983) have used essentially the indirect method to calculate
ocean
transport,
but
atmospheric GCM.
calculate
based
on
surface
ocean
transports.
transport, their results compare reasonably
considerable
flux
data
generated
by
an
Miller et al. used a one year simulation of the GISS GCM to
the model
Vonder Haar
heat
(1976),
Hastenrath
differences
(1980),
between
For the annual
well
and
global
heat
with the results of Oort and
other studies,
observations,
oceanic
though there
particularly
in
the
are
Southern
Hemisphere.
Covey (1988)
compares atmospheric
GCM simulations and observations.
and others.
The GCM's discussed include the GISS model
However, the versions of the NCAR, and GFDL models discussed in
the paper are not the same
ones
as considered
consider do not include ocean transports.
transport (atmosphere
the
and oceanic energy transports from
models
atmosphere.
and
in
plus
ocean)
satellite
here, since
Covey finds that the total energy
appears to be
observations
the versions we
approximately
of irradiances
at
the same in
the
top of the
In addition, Covey finds that "in the models most of this transport
takes place in the atmosphere
observations
indicate that
whereas the combined satellite and radiosonde
half or more of the transport
takes place
in the
oceans.
Previous satellite based studies have found that the oceans are more
effective
in
transporting
energy
in
low
latitudes,
while
the
atmosphere
is
perhaps more effective in middle and high latitudes (at least in the Northern
Hemisphere).
ocean
Covey may be a little too enthusiastic in donwplaying the role of
transports
in
the models,
and overplaying
the
atmospheric
transport.
Some of the numbers cited in comparison between models and observations in
Covey's paper are plainly
incorrect,
or the comparisons
inappropriate. 3 0
We
30
1n particular, p156, paragraph 2. The value given by Covey for the GISS model is the total energy
transport, not the atmospheric transport as implied by the context and the comparison with the
atmospheric transport in Carissimo et al. (1985). The same may also be true of the GFDL value quoted.
3
Validation of climate models
45
treat
Covey's
findings
with
some
skepticism
pending
the
results of further
studies.
3
Validation of climate models
46
4
4.1
Model
MODEL DESCRIPTIONS
hierarchies
The 3-D GCM's are most useful when used in conjunction with simpler
as
models so that particular features or processes can be studied in isolation
A hierarchy of models with dimensions from O-D to 3-D has
well as in context.
models
in time
and resolution
processes,
surface
dynamics,
While the 3-D
to study climate.
been developed
and
or relying
the
simpler
simplest models
convective
(EBM's)
their parameterization.
on
The
The
type.
the
calculate
vertical
averaged
(0-D)
into
incorporated
and will be described subsequently.
globally
1-D radiative-convective
averaged)
and
processes
lapse rate.
temperature
EBM's
of
redistribution
North et al. (1981)
it
relatively
is
by
energy
processes
Saltzmann (1978).
to
the
atmospheric
have reviewed EBM's.
profile
by
the vertical (usually
of the radiative
explicit modelling
which
a predetermined
re-establishes
The RC models have been reviewed by Ramanathan
have a vertically
easy
This necessitates
(RC) models compute
a 'convective-adjustment'
Two-dimensional
surface
incoming
A 1-D latitudinally varying model is used in this work
and oceanic transport.
The
between
Because of the simplicity with which
radiation.
horizontal
of the
models
balance
energy
generalize the models to allow for horizontal variations.
parameterization
or radiative-
balance
a balance
as
temperature
surface
are
dependences
of the energy
are usually
horizontally
solar and outgoing infra-red
with
space,
radiation,
include only some of these processes explicitly, neglecting the others,
models
(1978).
include
resolved atmosphere.
In 3-D
(atmosphere, or oceans
deal explicitly
statistical dynamical models (SDM's)
and dynamics
GCM's,
or both)
in a zonally
The
the 3-D
averaged
nature
is incorporated,
framework,
been
SDM's have
and Coakley
reviewed
of the climatic
by
system
and an attempt is made
represent most physical processes believed to be important.
and
to
The NCAR, GFDL,
and GISS 3-D models will be described subsequently.
4
Model descriptions
47
4.2
One-dimensional
energy
One-dimensional
been
have
latitude
useful
of temperature
variation
it
models
dependent
energy
is
climate
balance
balance
for
assumed
models
varying
the
coupling
examining
and temperature
that
model
albedo.
rate
which
at
of
function
between
dependent
the
a
as
latitudinal
In the latitude
enters
heat
each
infinitesimal latitude belt during the year is exactly balanced by the loss rate.
For the ith latitude strip
(net
transport
horizontal
out); + (infrared out); = (solar absorbed);
(4.1)
We employ a 1-D EBM in this work to study the sensitivity of the climate to the
We will use the simple EBM formulation given by
energy transport.
(1975),
involves
which
approximating
the
earth's
temperature
North
by
structure
the first two Legendre polynomials in a series expansion.
To set up the EBM it is necessary to assume that all energetic fluxes can
be parameterized in terms of the temperature at the earth's surface.
the characteristic
temperature
by the temperature
of the earth as
a radiating body
of the atmosphere far above the surface,
can be empirically related to that at the surface.
to
I,
space,
empirically
is
taken
as the prime
dependent
is determined
this temperature
The flux of infrared radiation
variable,
and
is represented
by
(4.2)
I=A+BT
where
I is in W/m 2 , T is
A=213.04W/m
Stone (1980).
of latitude.
Although
2
the surface
(sea level)
temperature
(*C)
and
and B=1.63Wm-2oC-1.
The values of A and B are from Wang and
Latitudinal dependence
is represented by x, where x is the sine
I is a function of x.
The absorbed solar heating is given by the form (q/4)S(x)a(x,xs), where
q is the solar constant, S(x) is the mean annual meridional distribution of solar
radiation (which is normalised so that its integral from 0 to 1 is unity), and the
absorption
system.
coefficient
a(x,xs) is one minus the albedo of the earth atmosphere
The quantity xs is the sine of the latitude of the ice or snow line.
4
The
Model descriptions
48
function
S(x),
which
uniformly approximated
is
determined
within
from
astronomical
calculations,
2% by
(4.3)
S(x) = 1 + S2 P2 (x)
with S2=-0.482,
and P2 (x) is the second Legendre polynomial, (3x 2 -l)/2.
a parameterization
is
for the zonal earth-atmosphere
albedo,
a,
We use
from Wang and
Stone (1980)
a = ao+a2P2(x)+8u(x-xs)
where u(x-xs) = {
1 x>x5
,
0 x<xs
(4.4)
ao=0.316, a2=0.146 and 5=0.186.
The horizontal
transport
of energy (meridional divergence
of the flux)
is given by the amount of heat per unit time leaving a latitudinal strip.
flux
divergence
is
modelled
by
-V
.(D
V T) where
D
The
is a thermal diffusion
D is calculated from the iceline condition for the current climate.
coefficient.
Following Wang and Stone (1980) we set Ts=-13.0*C (temperature at the iceline)
at xs=0.95 for roughly the current value of the solar constant (1365W/m 2 ).
yields D=0.355.
This
In terms of x, the energy balance equation for the unknown, I
may then be written
-
from
1x2) D.. (x) + I(x) = (q/4)S(x)a(x,xs)
dx
dx
which the transport can be derived.
expansion
in Legendre
the eigenfunctions
d-x2)
polynomials,
(4.5)
Equation 4.5 can
which have the advantage that
of the spherical
diffusion
annual
model
they are
operator
(4.6)
P(x) = -n(n+1)Pn(x)
For a mean
be solved by
with symmetric
hemispheres,
the boundary
condition that the gradient of I(x) must vanish at pole and equator yields I(x)
in the form
4
Model descriptions
49
I(x) = I
(4.7)
InPn(x)
n even
Substituting 4.7 into 4.5 yields
(4.8)
[n(n+1)D+1]In = (q/4)Hn(xs)
where
Hn(xs) = (2n+1) f
For
the
two
1
(4.9)
S(x)a(x,xs)Pn(x) dx
0
mode
the
approximation,
solution
of the
temperature
structure in equation 4.8 is reduced to the solution of three coupled algebraic
equations
Io = (q/4)Ho(xs)
(4.10)
6DI 2 = (q/4)H2(xs) - 12
(4.11)
Is = Io +;2-2(3x s2- 1)
(4.12)
The unknowns in these equations are the two coefficients of I in the truncated
For the parameterizations we have chosen, HO is a
expansion, 1o and I2, and xs.
third order polynomial,
and H 2 is a fifth order polynomial as follows:
Ho(xs) = [1/2 5 S2] xs 3 + [
- 1/2 S S 2 ] xs - 1/5 a 2 S2 + 1 - ao - 8
(4.13)
3
H 2 (xs) = [9/4 8S 2] xs 5 + [5/2( 8 - 5 S 2 ) ] Xs + [ 5/4 5 ( S 2 - 2) ] xs -
(4.14)
2/7 a2S2 + S 2 (1 - ao - 6) - a2
Stone (1978b)
notes
that 4.10 corresponds
absorption of solar radiation
is balanced
to the statement
by the global
that the
global
emission of infrared
radiation; 4.11 to the statement that the poleward diffusion of heat compensates
the net radiative heating in low latitudes and the net radiative cooling in high
latitudes; and 4.12 is the definition of the position of the edge of the polar ice
These equations can be combined into a single seventh-order polynomial
cap.
equation for x.
4
Model descriptions
50
Is = (q/4) [ Ho(xs) + H 2 (xs)(3x 2 1)
2(6D+1)
xs(q) which describes how the extent of the
This equation defines a function
polar ice cap
changes
as
(4.15)
the
solar constant
changes.
obtained by North (1975)
result for
our
The result is similar to that
is illustrated in figure 7.1.
parameterizations
The
with a different albedo parameterization.
The result
has been discussed extensively in the literature (e.g. North et al., 1981)
4.3
NCAR, GFDL, GISS models
In
this
section
we
will
briefly
outline
similarities
and
differences
between the versions of the NCAR, GFDL, and GISS models considered in this
Original
work.
of the
descriptions
models
are
for
available
NCAR
in
Washington and Meehl (1984), GFDL in Manabe and Wetherald (1987) and GISS
in Hansen et al. (1983,
Major characteristics of the models have been
1984).
outlined adequately by Schlesinger and Mitchell (1985,
al. (1988),
provides a nice summary of much of the information that will
Mitchell (1985)
All
Table 4.4 in Schlesinger and
and we shall follow their descriptions.
be presented
1987) and Gutowski et
below.
three
atmospheric
GCM's
integrate
the
primitive
equations
in
spherical coordinates for the entire globe, and all three represent the vertical
derivative
in
these
and
other
equations
by
finite
differences
with
nine
vertical sigma (pressure coordinate) layers between the surface and the top of
the
atmosphere.
The
sigma
are
layers
unevenly
to
spaced
give
highest
The GISS model represents the
resolution near the surface and the tropopause.
horizontal derivatives by finite differences with an 8* latitude by 100 longitude
resolution.
the
The NCAR and GFDL models employ the spectral method to evaluate
horizontal
derivatives,
but
the
advection
terms
are
transform method on a 4.50 latitude by 7.50 longitude grid.
evaluated
by
the
In all three models
the physics terms such as the diabatic heating are also computed on the grids
of the models.
horizontal
The global distribution of land and ocean is realistic within the
resolutions
of the
models,
as
is the
topography,
earth's
albeit
smoothed.
4
Model descriptions
51
Each of the atmospheric GCM's is coupled to a mixed layer ocean. In the
NCAR and GFDL models the depth of the mixed layer is constant, and there is no
heat transport. In the GISS model, the depth of the mixed layer is prescribed to
vary seasonally on the basis of observations, and the oceanic heat transport is
The oceanic heat transport used is
prescribed geographically and seasonally.
deduced
by
temperatures
The
simulated
requiring a match of the model's
of large-scale
calculations
snow
and
condensation
sea surface
method).
(the so-called 'Q-flux'
with observations
control
are
essentially
The calculations of sea ice, soil moisture, and
surface temperature are essentially the same in the NCAR and GFDL models and
Fractional sea ice cover can occur within a grid
different in the GISS model.
the same in all three models.
box in the GISS model, but not in the NCAR and GFDL models. The GISS model
has two layers within the ice, wheras the NCAR and GFDL models have a single
This is also the case for the number of soil layers used for the soil
moisture and soil temperature calculations. The NCAR and GFDL models have a
prescribed field capacity of 15cm everywhere, whereas the GISS model uses a
layer.
geographical
layers
distribution
with values
that range
from
2
to
65cm
for both
together.
The treatment of terrestrial and solar radiation is similar in all three
models in that they include the principal absorbers and emitters, H2 0, CO 2 and
03. The NCAR and GISS models also include absorption of solar radiation by 02,
and the GISS model also includes aerosols and solar absorption by NO 2 , as well
The solar
as the effects of N2 0 and CH 4 in the longwave radiation calculation.
calculation
essentially
and Hansen (1974)
follows the method of Lacis
in all
three GCM's, although different band models are used in the longwave in each
2
GCM. The solar constant in the GISS and NCAR models is 1367 and 1370W/M
The
respectively, which is within the current observational uncertainty.
2
GFDL model uses a value of 1467W/m .
constant
in
investigate
the GFDL
the
transport in the current
is not known to
model
implications
The rationale for this choice of solar
of
climate.
this
choice
for
however
us,
the
The GISS model
we
simulation
includes
later
will
of
energy
the diurnal
and
seasonal cycles of solar radiation, whereas only the seasonal cycle is included
in the NCAR and GFDL models. In all models, the albedos of snow-free land and
open ocean vary geographically,
but do not change when the climate changes,
4
Model descriptions
52
except in the GISS model where the ocean albedo depends on the surface wind.
Snow
Each model's surface albedo changes as snow and sea ice cover change.
and sea ice albedos are constants in the NCAR model, whereas these albedos
depend on factors such as depth and underlying vegetation in the GFDL and
GISS models.
Clouds form in the models as a result of either large-scale condensation
The GISS model does not allow clouds at pressures less
or cumulus convection.
The NCAR model allows clouds to form in all layers except the
than 100mb.
The GFDL model allows clouds to form in all layers.
lowest and two highest.
Clouds in all three models
by diagnostic
are produced
and are not
relations
advected, so that clouds can appear and disappear from one time step to the
from large-scale
next.
Clouds
relative
humidity criterion
condensation
occur
a
models,
The GISS convective cloud
grid box layers undergoing convective adjustment.
includes
and
albedos
depths,
when
apear when a relative humidity requirement is satisfied for
convective clouds
parameterization
model
GFDL and NCAR
In both the
is met.
in each
Cloud optical
a convective mass flux dependence.
emissivities
are
functions
of
location
cloud
and
the
number of layers a cloud encompasses.
In
summary,
it
appears
that
the
parameterizations
physical
and
numerics are similar in the NCAR and GFDL models, while the GISS model
differs from the others in both these characteristics.
The physics in the GISS
model appears to be more realistic than that in the other models, having been
ab
developed
initio for climate
studies.
One
of the principal differences
relevant to the transport validations considered here is the inclusion of ocean
heat transport in the GISS model, but not in the other models.
The horizontal
resolution is coarser in the GISS model than in the other models however.
important
for energy transport considerations
Also
is the large value of the solar
constant employed in the GFDL model (1467W/m 2 , which is 7% larger than the
values
models).
chosen
to
be consistent
with
observations
in
GISS
the
and NCAR
Only the GISS model includes the diurnal as well as the seasonal cycle.
The other major difference between models is in the parameterization of moist
convection.
4
Model descriptions
53
5
5.1
Climate
DATA
model output
The principal data sources used on this study are from 3-D GCMs and
The data used are primarily those relevant
observations of the Earth's climate.
to calculating energy transports, and to parameterizing the 1-D EBM.
The climate model data 3 1 used in this study consists of output from the
control run (1xCO 2 ) and doubled CO 2 runs in equilibrium experiments of the
Most of the model output
NCAR, GFDL, and GISS models described previously.
has
been
via
supplied
for
integration
(AER)
Research
Environmental
each
Gutowski
Bill
of
the model
is
experiments
as
The length of
GISS:
follows.
unaccelerated solar cycles for 1C0 2 , 35 unaccelerated cycles for 2CO 2 .
NCAR: For the 1C0
unaccelerated solar cycles.
2
and
Atmospheric
1987).
Climate Group,
(AER
at
associates
and
45
GFDL: 49
simulation, 12 solar cycles with
calendar accelerated by 9 (40.6 days per solar cycle), then 4 solar cycles with
of 3 (121.7
acceleration
For the
cycles.
solar
unaccelerated
output
All
obtained
from
only
simulation,
2CO 2
the 4
solar
11 unaccelerated
days per solar cycle), finally
(3x)
accelerated
and
11
cycles.
from
the
final
the
models
of
years
has been
an
supplied
experiment's
averages
as temporal
For
run.
NCAR,
the
averaging period for the output is the final 3 years, while for GFDL and GISS it
is the final 10 years.
and
1xCO 2
Washington and Meehl (1984) have noted that the 2xCO 2
simulations for the NCAR model had not reached quasi-equilibrium
by the end of these experiments, because of the lack of computing resources
available
termination
at
the
time.
of their
Washington
experiment
by
and
Meehl
claiming
that
justify
partly
the
3-D
the
model
early
should
approach equilibrium with a 50m ocean at the same rate as their experiments
with a 1-D radiative-convective
(personal
communication)
notes
(RC) model coupled to a 50m ocean.
that
this
justification
is
invalid
Stone
however,
since the RC model has half the sensitivity of the GCM, and therefore only half
31
Since the use of the word 'data' to refer to output from computer models is considered inappropriate
Where
by some people (see discussion in Bryce, 1989), its use will be kept to a minimum here.
possible, we will use the word 'output' instead.
54
the response time.
Schlesinger and Mitchell (1987)
note that the NCAR 1C0 2
experiment was cooling at 0.4K per year, and the 2CO 2 experiment was cooling
at 0.21K per year.
published, the 1C0
respectively,
2
After
the Washington
and Meehl
(1984)
results were
and 2CO 2 experiments were extended by 5 years and 2 years,
and showed a smaller secular cooling trend in both experiments.
The NCAR output supplied for this study is the same as that considered in the
Washington and Meehl (1984)
The
implications
considered
of the lack
of equilibrium
in the
NCAR
model will
be
derived from
the
in later analysis.
Model
seasonal
paper; that is, prior to the integration extension.
analysis here uses
mean output
annual means,
supplied for NCAR
which were
and GISS
and the monthly
mean
output supplied for GFDL.
All model output supplied by AER was formatted on the respective model
grids.
For the transport calculations we are only interested in the zonal means
(average
around
a
latitude
circle),
converted to zonal mean form.
is shown in table 5.1.
and
so
all
the
relevant
flux
data
was
The latitude spacing for the respective models
The flux data included the net shortwave and longwave
radiation at the top of the atmosphere and at the surface, as well as the surface
For the 'indirect' method we use to calculate
fluxes of sensible and latent heat.
model energy transports, only the flux data at the top of the atmosphere
needed.
indirect method
The
To test the validity
transports.
motions
the
However,
in
temperatures
of the energy
the model
chapter
on
energy
one would like to
transport
atmosphere
from model
and
(if
ocean
The relevant data to calculate the transport directly was not saved
included).
from
and
in the
of the indirect method,
with direct calculations
compare results
derived
be described
will
is
NCAR
the
and
GFDL model
atmosphere
and
ocean
saved from the GISS model run.
zonal mean
energy
transports
runs,
from
nor was
energy
the transport
transports
were
calculated.
calculated
and
Reto Reudy of GISS supplied annual mean
the GISS model control
run to allow
a
comparison with the indirect method to be made.
Gridded model output of the surface air temperature,
precipitation,
soil moisture was also supplied from each of the models by AER.
and
The surface
air temperature was used in conjunction with lapse rate data to calculate sea
5 Data
55
The sea level temperatures
for each of the models.
level temperatures
were
used in parameterizing 1-D EBMs to be 'equivalent' to the 3-D GCMs.
5.2
from Carissimo et al. (1985).
using the
transport
data,
total
plus ocean)
satellite
from
radiation at the top of the atmosphere.
storage
oceanic
and
atmospheric
and
transport
energy
observations
atmosphere,
of the
top
at the
radiation
(atmosphere
method
indirect
They estimated the poleward energy transports in
of the net
satellite measurements
Estimation of the
taken
system for the annual mean and four seasons based on
the atmosphere-ocean
atmospheric
climate was
transport data for the current
energy
Observational
data
transport
energy
Observational
was made
flux
net
the
of
data.
of
Note that our model derived estimates
for the total energy transport follow the same technique used to generate the
Carissimo et al. obtain the atmospheric transports from in situ
observations.
The Carissimo et al.
observations and infer the oceanic transport as a residual.
data is zonally averaged with latitude spacing every 100 as shown in table 5.1.
Errors in the
Carissimo et al.
transport
due to
data occur principally
Carissimo et al.
uncertainties in the satellite measurements of radiative fluxes.
apply various schemes to correct the radiative flux data to ensure global
For the curves corrected by the different techniques, some
spread exists at mid-latitudes, yielding an error estimate in the transports on
the order of 0.5x10 15 W in mid-latitudes. To gain an estimate of the contribution
radiative balance.
by
errors
Carissimo
estimates
budget,
cannot
that
et
al.
consider
corrected
Carissimo
et al.
systematic
take as
(1982)
correction.
energy
balance,
independent
satellite
For the case of the annual
radiation
by
independent
for various
The
global
requiring
difference
the
that are in global balance.
computed by Hastenrath
same
be
spread
between
the transport
estimates
samples,
satellite
in
the
but
Hastenrath
curves
using the
curves
is
comparable with the spread in the Carissimo et al. curves corrected for global
Therefore, for the Carissimo et al. data, the "total uncertainty in
imbalance.
the
transport
of energy
by
the
atmosphere
and
ocean
amounts
to
about
1x101 5 W in mid-latitudes and somewhat less in the tropics and high latitudes."
5 Data
56
5.3
data used
air temperature
surface
The
data
temperature
air
Surface
was
observations
represent
to
The data is averaged annually and zonally with
taken from Sellers (1965).
latitude spacing every 100 as shown in table 5.1. The 1-D EBM employed later in
this study
Sellers
uses sea level temperatures,
surface
observations
temperature
air
not surface
temperatures,
the
(like
were converted to sea level temperatures
temperatures)
model
and so the
surface
air
using lapse rate data
and topographic data as described below.
5.4
Sea
level
derivation
temperature
We are interested in sea level air temperatures
since these make more
physical sense as a driver of meridional energy transports than do surface air
The 1-D EBMs are parameterized in terms of sea level air
temperatures.
temperatures.
Sea
level
where the topography
departs
rather
a
is
temperature
from
arbitrary
sea level,
quantity
since the
in
choice
any location
of
how to
For instance, one could
extrapolate back to sea level is somewhat arbitrary.
use the moist or dry adiabiatic lapse rates, or standard atmosphere lapse rates,
or observed lapse rates at similar latitudes, or other such measures of what the
sea level temperature might be in the absence of the existing topography.
Observed lapse rates are probably as reasonable a choice as any, and so we use
observations
of the
lapse
rate
in
the
lower
atmosphere
extrapolate temperatures from surface air to sea level.
(850-950mb)
to
The lapse rate data and
topographic data is described below.
5.4.1
Lapse rate data
To calculate the observed lapse rate between 850 and 950 mb we used
The Oort data
data on temperatures and geopotential heights from Oort (1983).
consists of zonal averages at 50 latitude spacing between the equator and 70*,
with an additional observation at 800.
The Oort data is temporally averaged
Lapse rates
also, being an annual mean averaged over the years 1963 to 1973.
(y) were calculated as y = -dT/dz, where T is temperature and z is geopotential
5 Data
57
height.
lapse
The
The resultant 850mb - 950mb lapse rates are shown in figure 5.1.
rate over 850-950mb varies quite a bit with latitude (justifying the
additional work of using climatologically observed lapse rates rather than just
picking a single value as representative of all latitudes), and is lowest in north
A value for
polar regions, indicating the high degree of static stability there.
the lapse rate has not been calculated at 800 in the southern hemisphere (SH),
since there is no 850mb or 950mb data there due to the presence of the
In interpolating lapse rate values to model latitudes and
Antarctic continent.
to the Sellers temperature latitudes, for latitudes south of 70*S the value of the
lapse rate at 70*S was used. For latitudes north of 80*N, the lapse rate was set to
zero.
5.4.2 Topographic
Topographic
data
from the lapse
data for calculating sea level temperatures
For the
was available for each of the models.
GISS model the topography was supplied by Reto Reudy on the GISS grid, and is
The major continental mountain ranges and Antarctica
plotted in figure 5.2.
rates and surface temperatures
For the spectral NCAR
and Greenland are clearly discernable in the figure.
and GFDL models the equivalent gridded topography was supplied by Robert
Zonal average topography was calculated
Black, and is shown in figure 5.3.
from the gridded data and is shown in figure 5.4. The curve labelled 'G' depicts
the GISS zonal mean topography, and the curve labelled 'N' depicts the NCAR
and
GFDL
zonal
mean
topography.
Despite
the
differences
in
resolution
between the GISS model and the other models, the zonal mean topography is
relatively
5.4.3 Sea
Sea
similar.
level
level
temperatures
temperatures
were
calculated
from
the
topographic data, and surface air temperatures via T,,s.c = Ta,
lapse
-
yz.
rate
data,
Figure 5.5
shows the derived annual sea level temperature for each of the models (plotted
as dashed lines) and the corresponding surface temperatures (plotted as solid
The sea level temperatures depart most from the surface temperatures
at latitudes corresponding to the presence of Antarctica, the Himalayas, and
lines).
Greenland.
5 Data
58
5.5
data
Satellite
the 1-D EBM with observational
As part of parameterizing
data, it is
necessary to have observations of the outgoing longwave flux of radiation
the top of the atmosphere,
at
system.
and of the albedo of the earth-atmosphere
in zonal
Such observations have been made from satellite data and tabulated
Note that the Stephens et al. data was also
mean form in Stephens et al. (1978).
used as a source of data in producing estimates of the total meridional energy
transport in Carissimo et al. (1985).
The Stephens et al. data was produced from a composite of 48 monthly
mean
budget
radiation
from this data set.
Stephens et al.
zonally
Stephens
averaged
temperatures
The
synchronous
We
maps.
et al.
used
calculated
mean values
annual
in
The latitude spacing of the annual mean
data
is the
same
as
that
for the
Sellers
shown in table 5.1.
in the Stephens et al.
satellites employed
orbit
sampling
at near one
local
time
study were
during
all in sunhours.
daylight
Stephens et al. discuss sources of error and uncertainties in the satellite data.
To give an idea of the size of errors in the data Stephens et al. note that the
annual global average outgoing longwave flux is 234±7W/m 2 , the albedo is
top of the atmosphere
is
A value for the latitude of the sea ice limit, xs, in each hemisphere
is
0.30±0.01,
and the
net flux of radiation
at the
9±10W/m 2 .
5.6
Sea
ice
data
required in parameterizing the 1-D EBMs.
The valus of xs is incorporated into
When parameterizing the EBMs
the representation of the albedo in the EBMs.
to the GCMs we estimated values for x, from the graphical output for sea ice
limits published
figure 7.4).
Washington
by
the model
groups
for the
appropriate
control
runs
(see
For NCAR the sea ice extent was plotted for DJF and JJA in
We determined x, from the summer map in
and Meehl (1984).
each hemisphere, since the bulk of the insolation in polar regions is received
during the summer months.
For GFDL, February and August sea ice maps were
5 Data
59
For
supplied by Richard Wetherald, and the same procedure was followed.
GISS, xs values were obtained from information given in Hansen et al. (1984).
For observations we used the sea ice maps compiled by Alexander and
Alexander and Mobley
Mobley (1976) to determine the appropriate xs values.
present monthly distributions of the main ice packs of the Arctic and
Antarctic on a 1* global
grid using data
digitized from U.S.
Fleet Weather
Facility ice charts and Navy atlases. The sea ice limit was again estimated for
In each case
each hemisphere on the basis of the summer month maps.
(models and observations), the estimate of xs was taken as an 'eyeballed' zonal
average from the global maps, and serves as an approximate estimate to within
a couple of degrees latitude.
5.7
media
Journal literature and
the
In
in the
theory
presentation
chapter
on
popular
literature,
accounts
theory
of
we use
and
conventional
of
the
sources
for
interpretation
literature
On the presentation of greenhouse theory, journal literature is selected
from the major climate related journals, or from well known journals
For a
publishing greenhouse theory articles: JGR, GRL, Science, Nature, etc..
data.
data
source on
have
selected
because
of it's
public interpretation
the
news
impact
press
(as
and because
of the greenhouse journal
opposed
to television
it is documented,
literature
or other
archived,
we
sources),
and readily
The news press is quite vast and variable in coverage, so we adopt
the New York Times as the major news source in order to obtain some sort of
standard. We do not limit ourselves exclusively to the Times, but focus on their
researched.
coverage since it is presumably among the most sophisticated of the presses in
its discussion of environmental science issues.
5 Data
60
TABLE 5.1
Latitudinal spacing of the zonal mean data in the NCAR, GFDL, and GISS models,
in Carissimo et al. (1985) observations, and in Sellers (1965) observations (*N
and *S).
NCAR
GFDL
86.60
82.19
77.76
73.32
68.88
64.44
59.99
55.55
51.11
46.66
42.22
37.77
33.33
28.89
24.44
20.00
15.55
11.11
6.67
2.22
GISS
Carissimo et al.
Sellers
90.0
82.2
74.3
66.5
58.7
50.9
43.0
35.2
27.4
19.6
11.7
3.9
5 Data
61
VALIDATION OF MODEL FUNDAMENTALS : ENERGY TRANSPORT
6
6.1
Energy
The
and
transport
its
drive
fundamental
role
for
in
climate
motions
atmospheric
differential
is
(Stone,
heating and the resulting gradients of entropy and temperature
The 3-D
In the vertical,
and horizontally.
viewed vertically
radiation
tends to
The resultant convection
in isolation.
produced by radiation acting
is preferentially
gradients would be
at the surface so that large vertical temperature
absorbed
1984).
solar heating can be
from the differential
fluxes of heat resulting
solar
In a similar way, the variations with latitude of the
reduce these gradients.
absorbed flux would lead to large horizontal temperature gradients if radiation
Again, fluid motion takes place that tends to reduce these
acted in isolation.
all make significant
and mean meridional circulations
eddies,
the
and
transport,
energy
transient eddies, stationary
In the atmosphere,
low latitudes to high latitudes.
the
fluid motions transport energy from
and oceanic
The atmospheric
gradients.
the
of
transport
by
heat
sensible
represents
currents
ocean
In
1984).
primarily as sensible heat, latent heat, and potential energy (Stone,
addition,
forms,
different
in
transported
is
energy
contributions to
a
substantial fraction of the total energy transport (Oort and Vonder Haar, 1976).
Since
temperature
fluid
the
and
gradients
role
a crucial
play
external
change
concentration.
of heat
the
between
(by
a
given
heat
dynamical
(Stone,
1984),
transports
how sensitive
in determining
a
such
to
as
horizontal
role
in
Similarly,
the
fundamental
system.
and
temperature
the climate
increased
structure
ocean
currents)
greenhouse
meridional
temperature
transport
gradient,
surface
temperature,
and
thus
enhances
climate
sensitivity by increasing the contribution of the albedo feedback process.
the opposing
oceanic
heat
poleward.
sense,
Spelman
transport
This
gas
shift of the margin of snow covered area responding
in
change
the
reduces
is to
system
Held and Suarez (1974) speculated that the poleward
increases the latitudinal
to
play
of the atmosphere-earth-ocean
determining the climate
feedbacks
consequently
the
they modify
heat,
transport
motions
to
and Manabe
shift
reduces the
the
(1984)
margins
ice-albedo
of
feedback
note that we
snow
cover
and therefore
also
and
In
expect
sea
ice
the climate
sensitivity.
6
Validation of model fundamentals: energy transport
62
we will attempt to determine how well the
following analysis,
In the
NCAR, GFDL, and GISS models simulate the energy transport, and consider the
implications of the results.
to calculate
The procedure
the method outlined
follows from
budget
the
calculate
to
Procedure
6.2
latitude between
requires
the incoming solar radiation
at the top of the atmosphere
be balanced
and
the
transport
energy
total
in Carissimo
balance
Energy
considerations.
the
in
transport
energy
et al.
that
(1985)
the
outgoing
models
in the models
energy
using
difference
at each
infrared
radiation
by the sum of the divergence of
energy that is transported across each latitude circle by fluid motion and the
storage of energy in the system.
The energy budget for a unit-width latitude
band that includes both the atmosphere and oceans can be written as
SA +
So
= FTA
(6.1)
:(TA+To)
is the rate of change of energy in the atmospheric part of the
where SA=
band, So= D
a
-
is the rate of change of energy in the oceanic part, FTA is the net
downward flux of radiation at the top of the atmosphere (net shortwave plus
net longwave), TA and To are the transports of energy across a latitudinal wall
in the atmosphere and oceans, respectively, a is the mean radius of the earth,
and 0 the latitude. In equation 6.1, the storage of energy in the land, snow and
ice has been neglected (see Oort and Vonder Haar, 1976 for further discussion).
Integration of equation 6.1 over a polar cap yields
TT(,$) a TA(0,
where (( )) = (the T
2t/2 27r
8
subscript
(6.2)
+T(0,) = -(FTA) + (SA) + (So)
12 ( )
0
may
acosO' d$ a dM', TT is the total meridional energy transport
sometimes
be
dropped),
and
$ is longitude.
The
integration extends around a latitude circle (acos~d$) and over latitude (adO).
6
Validation of model fundamentals: energy transport
63
In the
For the present analysis we calculate only the annual transports.
annual case, equation 6.2 can be simplified (Carissimo et al., 1985) by assuming
that SA and S0 are zero when taken over a 'normal' year or, in other words, that
In Carissimo et al. observations this
interannual variability can be neglected.
is probably a reasonable assumption because the observed radiative fluxes and
other
several
parameters
energy
Similarly
years.
composites
represent
in
the models,
the
of measurements
annual
mean
made
radiative
over
fluxes
represent averages over 10 years (GFDL and GISS) and 3 years (NCAR), so the
This reduces the expression for the total meridional
assumption is again made.
energy
to
transport
_t/2 27c
TT(9) = -j
or
J
90
TT(O) = -2xTa
2
FTA(',)
i
0
acos' d$ adO'
FTA(O') cos'
dO'
(6.3)
where the overbar denotes time average over a year.
Integration of equation 6.3 from pole to pole would result in zero if the
net radiation at the top of the atmosphere was in balance (in Carissimo et al.
In practice this is
observations, or in the model simulated radiative fluxes).
not completely true in either case, and a residual results reflecting the degree
of imbalance in determination of the annual net radiation at the top of the
atmosphere. 3 2 For each of the models we have integrated FTA from pole to pole
In the GISS model, Hansen et al., 1984
to gain an indication of the imbalance.
note that there is a global annual net flux into the top of the atmosphere of
2
7.5W/m 2 . Of this imbalance 5W/m goes into the surface where it is absorbed
by the specified ocean (which is an infinite sink of heat) and the rest is lost in
and computer
the atmosphere through conversion to kinetic energy
Actually, there are several sources of imbalance in the annual net radiation at the top of the
atmosphere. With regard to observations, there are errors in measurement. In models, there are errors
In addition, there are interannual variations that result in imbalances from year to
in calculation.
year, which presumably smooth out to balance on average over a number of years if the climate is in
32
radiative equilibrium.
The climate is not in radiative equilibrium however.
It is being perturbed by
the long term build up of greenhouse gases among other things. Even over longer period averages we
would not necessarily expect annual radiative balance (though for most external forcings the radiative
perturbation from equilibrium is small relative to the net fluxes of shortwave and longwave radiation).
6
Validation of model fundamentals: energy transport
64
truncation.
In the GISS control run experiment the solar radiation absorbed at
the
surface
ocean
is
by
multiplied
to
order
in
0.96
cancel
the
energy
The imbalance is equivalent to 0.04 times the solar radiation at the
imbalance.
surface, so we subtract this amount from FTA in the GISS transport calculations.
In the NCAR model no attempt was made to correct radiative imbalances in the
Note however that the GFDL
The GFDL model is in balance.
annual mean.
model uses an incorrect value of the solar constant, and is unlikely to be in
balance with the correct value of the solar constant (if the solar constant
That the GFDL model is in balance with a solar constant
alone was changed).
7% too large suggests the presence of other compensating errors or tuning.
To
simulated
principally adopted in Carissimo et al. (1985).
represents
total northward
the accumulated
determining
'constant
the
followed
have
we
transport
in
imbalance
energy
the
for
correct
model
the
correction'
method
In this method, if T1=TNp(-T)
transport TA + To at the south pole
pole as denoted by the N P
T1
The corrected transport
subscript), the flux correction per unit area is (42)
(from
an
integration
north
the
at
beginning
at latitude 0 is given by
T(0) = T* (0) - f
NP
or
Tcos9
2 2/2
(4a
)
d
cosO dO
(6.4)
T(O) = T* (0) - T 1(1-sin0)
NP
where the star denotes uncorrected estimates.
The integration of equation 6.4 was carried out numerically.
GISS
model
we
employed
the
expended
method where the transport at latitude
j
trapezoidal
numerical
For the
integration
is related to that at latitude j-1 by
1j = Tj-1 - 21a 2 [F TA(j)cosj + FTA(oj.1)cosOj.1]/
2
(6.5)
The NCAR and GFDL models do not have grid points at 90*, and so it is was more
convenient to use
a midpoint
integration scheme
midpoint scheme the transport at latitude
6
j
for these models.
For the
is related to that at latitude j-1 by
Validation of model fundamentals: energy transport
65
j= Tj-t - 2xa 2 FTA(Oj)cosoj (Yj - yj-1)
(6.6)
where the y's are the midpoints (in radians) of the NCAR and GFDL latitudes.
The results of the transport calculations will be presented in a following
section, but first the validity of the indirect method will be explored.
6.3
Comparison of
direct and indirect transport calculation
method
The method used to calculate the total meridional transport of energy in
the models from the radiative balance at the top of the atmosphere has been
A direct approach would entail calculating the
labelled the 'indirect method'.
energy transports directly from the model fluid motions in the atmosphere
Model output to calculate the transports directly was not saved or
and ocean.
not available from the NCAR and GFDL models. This data was however provided
by Reto Ruedy in zonal mean form for the GISS model. This enables at least one
test to be made of how well the indirectly calculated total energy transport
compares with that calculated directly.
The energy transport output supplied by Ruedy includes the annual
northward global ocean energy transport and the annual northward energy
The total atmospheric energy transport was
transport in the atmosphere.
calculated as the sum of the northward transports of sensible heat,
The total energy transport
geopotential energy, and latent heat in the model.
was calculated as the sum of the atmospheric and oceanic transport.
The GISS direct transports have been plotted against the Carissimo et al.
observations for the atmosphere, ocean and total energy transport in figure
The figure shows that the total energy transport agrees reasonably well
6.1.
between observations and the model, while the model transports relatively
more energy in the atmosphere than the oceans than in observations (as
noted by Covey, 1988). The ocean transport in the model is particularly small
relative to the Carissimo et al. observations in middle latitudes in the Southern
6
Validation of model fundamentals: energy transport
66
Miller et al. (1983) also find that the GISS model yields smaller
Hemisphere.
The discrepancy may be
transports between 400S and 60 0 S than observations.
errors as well
partly related to observational
as to model shortcomings.
transports in the
observations, the radiosonde data used to derive atmospheric
are tentative
and the atmospheric transports
Southern Hemisphere is sparse
as
any error in
Since the ocean transports are inferred as a residual,
a result.
For
the atmospheric transports would affect the ocean transports as well.
In a test of the indirect method, the GISS indirect total energy transport
calculated from equation 6.3 was plotted against the GISS
total energy transports as shown in figure 6.2.
directly calculated
In general, the match of the
magnitude and distribution of the flux calculated by the two methods is pretty
us
gives
and
good,
method
and
The match is not perfect however, and we note that the
reasonably accurate.
indirect
useful
is
method
indirect
the
that
confidence
slightly
the
overestimates
peak
of
the
transport
in
the
This is possibly a result of the fact that the numbers
Northern Hemisphere.
for the net radiation at the top of the atmosphere in the GISS model still do not
In
annual balance.
yield a global
any case,
the spread between
the curves
relative to the
gives an indication of the error in using the indirect method
The error related to use of the indirect method is a few tenths
direct method.
of a terraWatt at the peak of the transport curve, and very small elsewhere.
6.4
6.4.1
Results
Annual
radiative fluxes
at the top of the atmosphere
averaged vertical fluxes for each of the models (obtained
The annually
The figures
from AER Climate Group, 1987) are plotted in figures 6.3.1 to 6.3.6.
show the major vertical fluxes of energy at the surface
heat,
longwave,
and
shortwave)
and
top
of the
(sensible heat, latent
(longwave
atmosphere
shortwave) for the 1xCO 2 and 2xCO 2 equilibrium runs in the models.
et al.
(1988)
calculation
Gutowski
have already analysed the model surface fluxes in detail.
of the
annual meridional
energy
transport
we
only the vertical fluxes at the top of the atmosphere (figs.
Clearly, there are differences
so we might begin
are
6.3.1
and
interested
For
in
and 6.3.2).
between the model simulated vertical fluxes, and
to suspect that there will be differences
6
in their energy
Validation of model fundamentals: energy transport
67
transports (as outlined in the next section).
We reserve discussion on why the
models differ until after we have considered the transports.
- 6.3.6 is that the
One feature which can be noted from figures 6.3.1
the
between
differences
the
between
differences
1CO 2 flux and 2CO 2 flux for each model.
greenhouse
change
That is, the
Does this mean that the game
than the differences in flux between the models.
successful
the
for each model is smaller
change in fluxes induced by the CO 2 perturbation
is over for
than
greater
are
fluxes
model
different
Not necessarily.
modelling?
A
could be that one of the models might be right and the
trivial explanation
This explanation is not required at all however, since there is a
others wrong.
If the mean state in each of the
far more plausible view that can be taken.
models is off by a little bit (yielding different values for the fluxes), that does
not necessarily mean that the predictions are off for doubled CO 2 . If the
models
are linearized about a different mean state, they would have different
the
equations
describing
only
slightly
and could
still yield
consistent results for the predicted
changes.
Whether this is so or not depends on how nonlinear the system is.
If the
for
coefficients
different),
nonlinearity
is strong, small differences
(but
physics
the
or perturbations might lead
to large
differences between predicted states (though it depends on the time scale and
Manabe
of interest).
the phenomenon
Stouffer
and
sort of behaviour in a coupled atmosphere-ocean
demonstrate this
(1988)
climate model.
In an initial
run with the model the ocean surface fluxes were close to observations, but a
thermohaline
circulation did not develop in the North Atlantic ocean
and the
simulated SST in the region was significantly lower in northern latitudes than
the observed.
In another model run with the surface ocean fluxes changed a
little, a thermohaline
circulation
developed
in the
the simulated SSTs were closer to observations.
North Atlantic ocean,
and
The question of how important,
On the basis of the
and what role the nonlinearities play is still an open one.
Manabe and Stouffer results at least, we have some grounds to be suspicious of
the
consequences
of the
fluxes noted above.
differences
between
the
model
simulated
vertical
On the other hand, we do not have sufficient grounding to
dismiss the model predictions on the basis of the differences.
In calculating
flux (shortwave
the energy
transports
we use the
annual net radiative
plus longwave) at the top of the atmosphere
6
(as represented
Validation of model fundamentals: energy transport
68
in eqn. 6.3).
The net flux at the top of the atmosphere is shown in figure 6.3.7
for each
of the models
Stephens
et al.
(1981),
and
for observations.
and are
represented
The
observations
by the dashed
line.
are
from
At most
latitudes the observations lie between the ranges displayed by the three GCM
net flux curves.
low latitudes
For all curves the broad pattern shows a net downward flux at
and a net upward flux in high latitudes.
This pattern yields
poleward energy transports as we shall see in the following section.
6.4.2
Energy
transports
calculated for the control
The annual northward total energy transports
run for each model from equation 6.3 are shown as a function of latitude in
figures 6.4.1 to 6.4.4.
In figures 6.4.1 to 6.4.3, three curves are shown for each
of the NCAR, GFDL, and GISS energy transports.
The curves labelled 'S' and 'N'
south and north
poles respectively.
The curve labelled 'C' is the corrected transport calculated
from equation 6.4.
are
for integrations
beginning
from
the
The curve labelled '0' is Carissimo et al. (1985) observations.
For the GFDL model,
the difference between
negligible, indicating that the model is in annual
GISS model there is a small residual.
curves
N,
S, and C is
radiative balance. 3 3
For the
This might be expected since there is no a
priori reason why the constant correction technique used to balance the GISS
model at the surface should yield a successful energy balance at the top of the
atmosphere.
In the NCAR model there is a large residual at the poles.
This is not
unexpected, since the NCAR model control run is known not to have reached a
quasi equilibrium.
globally averaged
Schlesinger and. Mitchell (1987)
surface air temperature
year in the control run.
note that the NCAR model
was cooling at a rate of 0.4K per
Using the NCAR transport residual it is possible to
estimate an implied drift for the NCAR model to compare with the actual drift.
We perform this calculation crudely below.
33
1t also serves as a check on the implementation of the integration scheme used to calculate the
transport, since it would be quite a coincidence for the scheme to be in error and still yield a perfect
balance for one of the models.
6
Validation of model fundamentals: energy transport
69
The
NCAR
surface area is
energy transport
5.1x10 1 4 m 2 ,
residual
is
about
5x10 1 5 W.
9.8W/m 2 .
so the imbalance is
The earth's
The model drift is
of the model.
given by this number divided by the total heat capacity
The
We will
NCAR model has a 50m deep ocean, land surfaces, and atmosphere.
approximate the area of the earth's surface covered by the NCAR ocean as 70%,
and we will neglect the heat capacity of the land relative to the ocean and
ocean
the
Approximating
since it is small.
atmosphere
specific heat at 1 atm. pressure and 0*C (Cp=4217
as
water 3 4 with a
) and density
1000kg/m 3 ,
the ocean heat capacity is
JJ
x 50m x 1000kg/m 3 = 1.49x108 12
0.7 x 4217-kgK
m K
the atmosphere as dry air at constant pressure with specific
J
density 1.225kg/m 3 , and depth 8km, the atmosphere heat
heat CP=1004kgK'
Approximating
capacity
is
J
J
1004- x 8000mn x 1.225 kg/rn 3 = 9.8x10 6 m 2K
kgK
The total model heat capacity is approximated
as the sum of the ocean and
Dividing this value into the
atmosphere heat capacity, and is thus 1.6x108mK.
energy imbalance yields a drift of 6.2x10- 8 K/s, or approximately 2K per year
(cooling).
If
the
was
model
perfectly
then the drift calculated
approximations were reasonable,
same
the figure
as
consistent
energetically
Schlesinger
quoted by
and
Mitchell
and
the
above
above should be the
(0.4K).
The
drift
implied by the energy imbalance at the top of the atmosphere does not match
drift in
imbalance
in the NCAR
possibly
it is larger.
the model;
the actual
model
is partially
This
suggests that the
compensated
by
something
energy
else;
the numerics.
34
A single heat capacity is used for the ocean in liquid form, and no attempt is made to include
variations for sea ice or for the heat capacity of the polar ice sheets.
6
Validation of model fundamentals: energy transport
70
corrected
This
value is subject to
should
corrected
residual
the transport
Because
be
model
NCAR
model,
to
referring
discussion
subsequent
in
in the
the
than for the other models.
greater uncertainty
mind
in
born
is so large
the
transports.
Figure 6.4.4 shows the corrected model energy transports plotted against
The figure shows
one another and against the Carissimo et al. observations.
differences between the different model transports, particularly at the peaks
in the transport
curve.
The differences
the NH
peak in
the transport
at
between the models are large compared to the error inferred in the technique
Only the NCAR transport curve
for calculating the transport from figure 6.2.
is unreliable due to the global energy imbalance in that model.
While the
differences between models can be assumed to be real, the differences between
the models and observations are not as reliable due to the uncertainty in
observations at mid-latitudes of about lx101 5 W.
The difference between models
and observations is of this order or greater, so the differences between the
The
models and observations are probably of marginal significance or better.
model shows a relatively larger total energy transport than observations
in the Northern Hemisphere, while the NCAR and GFDL models show relatively
The result is not completely symmetric in
less transport than observations.
GISS
At the transport peak in the SH the GISS
the Southern Hemisphere however.
and NCAR curves are close to observations.
greater than
observations,
though not
The GISS transport is a little
by more than
the
uncertainty
in the
The NCAR SH transport peak is even less significantly different
The
from observations due to the additional uncertainty in the NCAR curve.
GFDL SH transport peak is smaller than observations by a margin greater than
observations.
the
uncertainty
in the observations.
Returning
to
discussion
of
the
more
significant
NH
transport
we note that the GISS model total energy transport includes model
The ocean transports follow
contributions from the atmosphere and ocean.
differences,
from the prescription of SSTs using the 'Q flux' method.
The NCAR and GFDL
models do not include ocean transports, so all the energy transport must take
It is likely that the atmosphere does not pick up all
place in the atmosphere.
the slack caused by the omission of ocean transport (Stone,
the smaller NCAR
and GFDL transports.
6
The implications
1984), leading to
of this will
be
Validation of model fundamentals: energy transport
discussed in the next chapter.
overextensive
run simulates
For now we note that the NCAR model control
sea ice relative to observations (Washington and
They attribute this as being most likely due to the lack of ocean
heat transport in the model. In the GISS model, the annual mean sea ice cover
There is 15% less sea ice in the
is under extensive relative to observations.
Meehl, 1984).
transport
energy
calculations
relative
but
slightly underextensive
conclusions
about
in
whether
the
Northern
transport
small
these
Hemisphere
at other times
both hemispheres
communication).
personal
(Wetherald,
in
observations
to
energy
GISS
is
too
In the GFDL model the simulated sea ice is
strong relative to observations.
overextensive
the
which
in
This is consistent with the
1984).
model than in observations (Hansen et al.,
35
winter,
of the year
It is difficult to draw solid
differences
are
consistent
with
the
since the energy transport and sea ice extent
This will be elucidated in the
by other processes as well.
GFDL energy transport curves,
are
influenced
following chapter when analysing the EBMs.
In terms of model validation, our major concern from figure 6.4.4 is
whether the differences between the model simulated energy transports are
The next chapter will be
important; if so why, and with what implications.
devoted
to
exploring
this
by
issue
considering
how
sensitive
the
climate
system is to the energy transports.
6.4.3
Atmospheric
transports
energy
Before proceeding to the next chapter, we present here the atmospheric
The results will be discussed in relation
component of the energy transports.
to the temperature structure displayed in the models to determine whether the
errors in each quantity appear to be consistent with one another.
The
atmospheric
shown in figure 6.5.
transports
energy
for
observations
are
by the dashed curve,
and
models
Observations are represented
and
For the NCAR (labelled N) and GFDL (labelled
P) models, the atmospheric energy transports are the same as the total energy
are from Carissimo et al. (1985).
The sea ice simulation in the GFDL model run corresponding to this data has not been described in
any published documents, and these observations were taken from model sea ice maps supplied by
Richard Wetherald.
35
6
Validation of model fundamentals: energy transport
72
transport.
ocean energy
are
transports
energy
from
atmospheric
model
from the
calculations
direct
the atmospheric
G),
model (labelled
For the GISS
have no
since these models
using the indirect method,
calculated
transports
motions as supplied by Reto Reudy.
GISS
the
with
consistent
containing
model
more important in low latitudes),
relatively
and
less than the NCAR
transport is
energy
are similar to
In low latitudes the GISS
one another, but much larger than observations.
atmospheric
transports
shows that the model atmospheric
Figure 6.5
ocean
GFDL transports,
are
(which
transports
which are not represented
in the
One might expect that the intensity of the Hadley cell in the
and GFDL models would be stronger than that observed in the real
other models.
NCAR
atmosphere to make up for the omission of ocean transports in low latitudes.
We have not been able to find published references to the strength of the
Hadley cell circulation in the control run for either of these models to verify
this
on
Information
however.
meridional
mean
simulation
circulation
is
The Hadley cell
available for the GISS model (Stone, personal communication).
in the GISS model is approximately correct, being some 10-20% too weak. The
atmospheric energy transport in low latitudes in the NCAR and GFDL models is
15 and 50% greater than in the GISS model,
between
that their
suggesting
Hadley cell circulations are too strong.
In midlatitudes the GISS simulated ferrel cell is too weak, being less than
Starr et al. (1970) have published estimates for
half of its observed strength.
The GISS
observations of Ferrel cell strength using a variety of techniques.
The Ferrel
simulated Ferrel cell is weak relative to the Starr et al. estimates.
cell is a thermodynamically
the
Ferrel
will
cell
indirect reverse
contribute. to
atmospheric energy transport.
cell
is
underestimated
has
impugned
transports
are
transports
and
the
simulated
(Stone,
observations,
correct
instead.
observations
eddy
of
the
momentum
poleward
transports
suggested
The
differences
atmospheric
are
in
Covey (1988)
personal communication).
and
af the
the
model
atmospheric
between
model
atmospheric
that
transport
are
quite
large
and it is unlikely that the observations could
however (between 50 and 80%),
be off by this much.
overestimation
In the GISS model it is believed that the Ferrel
because
error relative to observations
an
cell, and so an underestimate of
In the absence of good reasons why the observations
6
Validation of model fundamentals: energy transport
73
be systematically
should
we
continue
to
strong
relative
to
amount,
a large
such
low by
too
the observations.
believe
transports
energy
atmospheric
the
If
too
are
observations in the GISS model, then they are also too strong in the NCAR and
GFDL models, which have similar values for the peak atmospheric energy
The peak of the atmospheric
transport.
similar
though
even
since,
transport,
where
and
NCAR
the
the
transports
GFDL
to be
contain
ocean
do
models
not
peaks,
transport
energy
total
might be expected
atmospheric
the
Considering now figures 6.4.4 and 6.5 together,
transports in the models are too large relative to
energy transport peaks too.
the
atmospheric
energy
and
observations;
the
observations, are relatively much closer to observations.
approximately
transports
are
meridional
temerature
would
correct
are
gradients
about
suggest
right.
matching
not
while
transports,
energy
total
That the model total
that
If this
simulated
their
is
so,
then
the
atmospheric components of the models are being too efficient (relative to the
We can investigate this
real atmosphere) at transporting energy polewards.
by
considering the meridional temperature
6.4.4
Meridional
temperature
structure
simulated by the models.
structure
In our first attempt to describe the meridional temperature
structure in
the models we fitted the model zonally averaged temperatures to an equation of
the form T(x)=To+T 2 P 2 (x), where x is the sine of latitude, and P2 is the second
We chose this form to match the form of the EBM
Legendre polynomial.
In this representation, To is the planetary
temperature representation.
average temperature, and T 2 gives an indication of the equator to pole
This approach was not very fruitful however, as it
temperature gradient.
turned out that T 2 did not give a very good measure of the forcing of the
energy transport flux. The probable reason for this is that T 2 is influenced by
temperatures
in
high
whereas
latitudes,
the
transport
flux
maximum
peaks
near about 400, and is not much influenced by high latitude temperatures.
well
55*N.
Stone and Miller (1980) have shown that the total energy transport is
correlated with the temperature gradient at 1000mb between 25*N and
Following this approach, we take the temperature gradient AT, as the sea
6
Validation of model fundamentals: energy transport
74
level
temperature
also
calculated
between
difference
AT
as
surface
the
250 and 55* in the models.
latitudes
The results for the sea
similar.
latitudes, and the results were quantitatively
these
between
difference
temperature
We
level AT's are shown in table 6.1.
For the NH, the AT for each of the models is quite close to its value for
The
observations.
observations
temperature
derived from
Agreement
on
A T's
the Sellers
is to
be
represented
are
(1965)
correct, because the total transport
reasonably
the
total
energy
level
transports
AT.
determines
sea
observations.
surface temperature
if the
expected
by
here
are
In the SH, the
GISS model AT is close to observations, while the NCAR AT is too large, and the
GFDL AT is too small.
The GFDL AT is small because the model simulates a
relatively high zonal mean temperature at 55 0 S (60 C too warm).
the
SH is mostly
ocean,
few observing
with
same
an
with
also applies
an atmospheric transport
stronger than observations
transport
atmospheric
to the models
the
atmospheric
observations.
From this we conclude
the
inconsistencies
models.
between
To
models
than
are
much
that the atmospheric
or
resolve
temperature
transports, we need to consider their energy balance
The
observations.
where the AT's are
transports
highlight
the
stronger
in the NH,
yet
in
observations
in the GFDL SH a weak
observations,
efficient
so
Similarly, a AT close to observations as in the GISS SH is not
as in figure 6.5.
consistent
and
Nonetheless,
themselves might be questionable there.
AT is not consistent with
stations,
This latitude in
some
larger
transports
of
structure
close to
the
and
in more detail.
than
are too
apparent
energy
To this
end we will parameterize 1-D EBMs from the output of the 3-D GCMs to analyse
the energy balance implications of the errors in the GCMs.
This approach is
taken up in the next chapter.
6
Validation of model fundamentals: energy transport
75
TABLE 6.1
Sea level
temperatures near 550 and 250 in each hemisphere derived from GCMs
AT is the temperature difference per degree of
and Sellers (1965) observations.
The actual latitudes were chosen on the
latitude over this range of latitudes.
They are:
model grids as close as possible to the above nominated latitudes.
NCAR and GFDL (55.55*
and 24.440), GISS (58.70
and 27.4*), and Sellers (550 and
250).
NCAR
GFDL
GISS
SELL
T at 55 0 S
(K)
T at 250 S
(K)
AT
(K)
T at 25*N
(K)
T at 550 N
(K)
AT
(K)
272.3
280.2
274.3
274.4
295.6
295.6
293.9
292.6
0.75
0.50
0.63
0.61
296.4
296.6
295.6
295.0
275.7
274.9
275.1
274.8
0.67
0.70
0.66
0.67
6
Validation of model fundamentals: energy transport
76
MODEL CLIMATE SENSITIVITY
7
7.1
Introduction
In the previous
between construction
outputs.
simulated
from
and differences
we noted various similarities
chapters
of the three
GCMs considered
For instance,
the model
here,
and it is
one another (though not radically different from observations),
is
concern
different
whether
differences
the
The
sensitivities.
climate
model
in
Of primary
are important or not.
interesting to ask whether the differences
term
energy
transport
could
yield
refers to how
sensitivity'
'climate
different
are
transports
energy
their
and between
much the climate changes (as measured by say its global average temperature)
in response to a perturbation or change in forcing (such as due to changes in
incoming
changes in
due to
or
radiation
solar
of greenhouse
concentration
gases in the atmosphere, etc.).
For
CO 2
doubled
the
three
the
experiments
display
GCMs
similar
Why should this be so when the models have obvious differences
sensitivities.
in the way they are formulated, and in the way they model certain processes?
transports
We know that their energy
show that the
and will
are different,
To test this sensitivity,
climate is sensitive to the energy transport.
we will
This yields
parameterize a 1-D EBM with output from each of the three GCMs.
'equivalent' EBMs, meaning that each EBM is set up to be an equivalent 1-D
We hope that the essential
of the 3-D GCM.
energy balance representation
physics relevant to energy balance contained in each GCM will be represented
1-D equivalent.
in its
then
their equivalent
If the three GCMs really do have similar sensitivities,
1-D
EBM
sensitivities
(assuming that the essential physics has been captured).
sensitivities
turn
out to be
different,
then
the
fact
that
similar
to have
also ought
representations
If the
the models
have
approximately the right control climate, means this must be so for the wrong
reasons.
If the
compensating
differences
sensitivities
factors
between
or
the
turn
processes
models,
out
to be
at
work
some
of
similar, then
(to
account
which,
sensitivity), and we ought to try to identify them.
we
must
there
for
know,
the
be
obvious
affect
the
77
7.2
Climate
sensitivity
to
the
energy
transport
To illustrate how the climate is sensitive to the energy transport, we will
Stone used an EBM to show that the
follow the approach used in Stone (1978).
energy transport flux is insensitive to the climate (except when the dynamical
the
the
radiative
local
is near
system
climate
will
and
result
Stone's
rederive
mode truncated EBM (as
two
parameterization
described
but with albedo
earlier),
from
parameterization
temperature-longwave
same
the
using
and
Wang
The rederivation will also serve as a further introduction to the
Stone (1980).
EBM
We
flux.
transport
'Northtype'
and
The converse of this result is that the climate is very sensitive
equilibrium).
to
low
is
transport efficiency
which
be
will
from
parameterized
model
output
and
observations
in
sections.
following
which
Equation 4.15 in the EBM derivation defines the function xs(q),
describes how the extent of the polar icecap
changes as
the solar constant
The iceline latitude is plotted against the ratio of the solar constant
changes.
Where the
to the current solar constant (taken to be 1365W/m 2 ) in figure 7.1.
solar constant has its current value, the iceline is at its NH value of xs = 0.95
A climate sensitivity for the EBM is readily found by changing the
(72*N).
solar
constant
temperature,
a
by
known
the
calculating
were
performed
not
average
planetary
The climate sensitivity,
To, corresponding to that solar constant.
P, can be defined by P = q aTo/Aq.
experiments
and
amount
In the 3-D GCMs, the climate sensitivity
by
changing
solar
the
constant,
but
by
Greenhouse gases are
doubling the amount of CO 2 in the model atmosphere.
not included explicitly in the 1-D EBM, so we cannot perform the exactly
analagous CO 2 doubling sensitivity test. We can however roughly approximate
this sensitivity test by simply increasing the solar constant by 2% in the EBM,
since the radiative forcings for each experiment are of similar magnitude.
In
experiments with the GISS GCM and a 1-D radiative-convective model, Hansen
solar constant change corresponds to a forcing of
doubling, the global mean radiative forcing is about 4W/m 2
et al. (1984) note that a 2%
4.8W/m
2.
For a CO 2
(Hansen et al. 1981).
similar radiative
similar
global
experiments.
The small heat capacity of the atmosphere means that the
forcings
mean
heat
for
CO 2 change and solar constant change lead to
fluxes
into
The resultant equilibrium
the
surface
planetary
global mean
in
the
GISS
warming of the surface
7
Model climate sensitivity
78
air is about 4 0 C in both the solar constant and CO 2 GISS sensitivity experiments.
In later sensitivity experiments with the equivalent EBMs we will use a 2%
solar constant change as a proxy for a CO 2 doubling sensitivity test.
The meridional energy transport flux in the EBM is modelled by a simple
diffusion
law:
F = -2 7c R 2 (1.x 2 ) D dI/dx
6
where R is the radius of the earth (we use a mean earth radius of 6.367x10 m).
Substituting for I(x) and differentiating
From equation 4.7, I(x) = I + 12 P 2 (x).
yields:
F = 6 D 12 ic R 2 (x 3 -x)
7.1
The transport flux is plotted as a function of x in figure 7.2. The flux maximum
occurs where x=1/43 (~35*), and has a value of about 5.5 terraWatts. This is
very close to the value of the flux maximum in the NH in the Carissimo et al.
observations, but this is not too surprising.
of the flux to internal parameters,
In demonstrating the insensitivity
Stone (1978)
notes that the "ability
of a
climate model to reproduce the observed flux is not a good test of the model's
reliability."
To
calculated
seventh
expanded
reproduce
result
Stone's
with
our
parameterizations,
we
first
x, as a function of D for D values between 0 and 136 by solving the
The fully
order polynomial for xs represented by equation 4.15.
polynomial
{(12D+2)-
1
is:
27/4 8 S2 1 Xs7 +
{(12D+2)- 1 (15/2 8 - 21/4 8 S2 } xs5 +
3
{8 S2 [1/2 + 25/4 (12D+2)- ] - 10 8 (12D+2)-1} xs +
2
{3(12D+2)- 1 [S2 (1-aO-) - 2/7 a2S2 - a2]} xs +
{ 8 [1 + 5/2 (12D+2) 1 ]
{ 1 - 1/5 a2S2 - ao -8
-
-
S 2 (1/2 + 5/4 (12D+2)-] } x, +
(12D+2)-' [ S 2 (1-aXo-8) - 2/7 a2S2 - a2] - 4 Is/q
}
=
0
36
D is a measure of the efficiency of the dynamical fluxes in transporting energy from low to high
When D=O, the system is in local radiative equilibrium, and as D increases the system
latitudes.
deviates more and more from local radiative equilibrium.
7
Model climate sensitivity
79
The resultant
x, curve is shown in figure 7.3.3.
efficiency of the dynamical
polewards with increasing
iceline
transports (D)
until D
At this point the transports are so efficient that the ice has
reaches about 0.39.
been eliminated,
The iceline latitude recedes
and so further increases
in D
are irrelevant
as
far as
the
is concerned.
Figure 7.3.2
from eqn. 4.11.
shows 12 as a function of D, where 12
The flux maximum (Fmax) can now be calculated as a function
of D from eqn. 7.1.
The result, shown in figure 7.3.1 is quantitatively similar to
Stone's result (his figure 4).
Fmax
is insensitive
efficiency for D greater than about 0.17.
not deviate
has been calculated
by more than
to the dynamical
Over the range 0.17 < D < 0.8 Fmax does
from the value 5.5 terraW
15%
transport
current climate in the EBM is about 0.355).
(note: D for the
While Fmax is insensitive to the
transport efficiency, note that conversely, the climate is very sensitive to the
value of Fmax- Moving along the curve there are three values of D which give
an Fmax of about the present NH value.
These D values are 0.29, 0.355 (the
and 0.42, which correspond to radically different climates with
iceline latitudes at 600, 750 (the present) and an icefree earth respectively. The
current value)
value of the flux maximum need change by no more than a few percent over
the range of D values between
could be
anywhere
an
between
about 0.2 and 0.5,
iceage
and
and the resulting climate
an icefree
earth.
For
further
discussion of the Fmax - D curve refer to Stone (1978).
We conclude that the climate is potentially quite sensitive to the energy
transports.
In the following sections we will set up equivalent EBMs to test this
sensitivity for the three GCMs.
7.3
7.3.1
Equivalent
EBM
parameterization
Parameterization
technique
Equivalent EBMs were parameterized with output from each of the three
GCMs and with observations.
Because the hemispheres do not have symmetric
climates, a NH and a SH equivalent EBM were set up for each of the above cases,
resulting in eight cases.
The values of the input parameters in the NH and SH
were also averaged to give a global equivalent EBM for each case (four more
7
Model climate sensitivity
80
twelve
yielding
cases),
parameters
to produce
in
cases
is valid assuming
a global version
of
averages
Taking
all.
hemispheric
and the
linearity,
extent to which the global EBM results are not bracketed by the hemispheric
EBM
results
an indication
give
will
of the
non-linearity
of the
response.
Jumping ahead to the results in table 7.1, it appears that the global equivalent
results are roughly
bracketed by
result.
each hemispheric
To specify the EBM in each case, the required input parameters are the
meridional distribution of solar radiation (SO,S 2 in eqn. 4.3), the albedo (ao,a 2 ,8
in eqn. 4.4), the surface temperature - longwave coupling (A,B in eqn. 4.2), the
sea level temperature at the iceline latitude (Ts), and either the dynamical
The major outputs from
(xs).
or the iceline latitude
transport efficiency (D),
the model are the longwave coefficients from eqns. 4.10-4.12 (10,12), and D or
xs.
Our initial approach was to specify D and calculate xs on the basis of the
D.
specified
The
reasoning
for specifying
D
that we know the flux
was
in each model reasonably accurately from earlier calculations, and
can use eqn. 7.1 and eqn. 4.11 to solve for 12. Knowing I2, we can then solve
maximum
for D and use it as an input to solve for xS.
The problem
with this approach
The
should be apparent from the conclusion drawn in the previous section.
iceline latitude is too sensitive to the value of the flux maximum, and so any
small
errors in the
value of the flux maximum
in the
lead to large errors
derived value of the iceline latitudes.
The approach taken then was to specify the iceline latitude xs, and use
that to derive D for each of the models and observations.
It is important to
have the iceline correct in the equivalent EBMs if they are to represent the
same
climate
states
parameterizations
7.3.2
Mean
solar
constant
as
simulated
are described
annual
by
the
The
GCMs.
results
of
the
in following sections.
meridional
distribution
of
solar
radiation
and
The mean annual meridional distribution of solar radiation at the top of
the atmosphere in the EBM, S(x), is represented by eqn. 4.3. i.e. S(x)=1+S 2 P 2 (x).
7
Model climate sensitivity
81
With S2 =-0.482, S(x) is uniformly approximated within 2% (North, 1975).
distribution of S(x)
is determined from astronomical
The
and is the
calculations,
same in each EBM since all the GCMs incorporate this distribution correctly.
We use the above value of S2 in each case.
To calculate the latitudinal distribution of the albedo in each of the GCMs
and observations it is necessary to know S(x).
For this purpose we interpolated
S(x) values to model latitudes from the values for S(x) tabulated in Chylek and
since this is more accurate than using eqn. 4.3.
Coakley (1975)
Note that, in
the annual mean case, S(x) is symmetric between the hemispheres.
The solar constant, q, used to multiply S(x) to obtain the net incoming
solar flux,
Q,
at a given latitude is not the same in the GCMs.
For NCAR it is
1370W/m 2 , for GISS it is 1367W/m 2 , and for GFDL it is 7% larger at 1467W/m 2 .
We use thes values in the equivalent EBMs.
The observational value of the
solar constant is not known to within better than about 5W/m 2 .
given
choice
current
measurements
is
A reasonable
1365W/m 2 , which is the value we use
for the EBM parameterized to observations.
Note that the EBM climate is quite sensitive to the solar constant as
displayed
in fig. 7.1.
With the EBM as parameterized in this figure,
in solar constant is enough to eliminate the polar ice sheets.
increase
a 2%
All
other things being equal, the 7% larger solar constant used in the GFDL GCM
should eliminate the ice caps in that model.
This is not the case in the GCM
however, since there are ice caps in the GFDL control climate.
Compensating
factors must be at work there.
7.3.3
Iceline
latitudes
and
sea
level
temperatures
The technique for determining the iceline latitude in each of the models
and observations was outlined in section 5.6.
was Alexander and Mobley (1976).
The iceline latitudes (accurate to within a
couple of degrees) are plotted in figure 7.4.
iceline
latitude
the same.
in observations
The data source for observations
in northern
The first thing to note is that the
and southern
hemispheres
is not
The NH iceline is at x,=0.95 (720) as conventionally specified in EBMs
for the current climate, while the SH iceline is at x,=0.89 (63*).
7
The GFDL and
Model climate sensitivity
82
GISS icelines are fairly close to the observed, while the NCAR model simulates
too much ice in both hemispheres
as has been noted previously.
The sea level temperatures at the iceline, Ts, were interpolated from the
profile for each of the models and observations in the
meridional temperature
respective
The
hemispheres.
results
are
shown
in
figure
Since
7.5.
the
meridional temperature profiles are fairly similar in each of the models and
observations, the results for T, mostly reflect the iceline latitude positions
(except for the GFDL SH, which is warm in southern high latitudes).
7.3.4
Outgoing
longwave
radiation
The flux of outgoing longwave radiation at the top of the atmosphere, I,
I is parameterized as given in
is the prime dependent variable in the EBM.
eqn. 4.2 as I = A + BT, where T is the sea level temperature in *C. For each of the
models, output for I was available directly from the AER data.
I is tabulated in Stephens et al. (1981).
For observations,
The latitudinal profiles of I are shown
In
in figures 7.6.3 and 7.6.4 for each hemisphere for models and observations.
each of the curves there is a minimum in outgoing longwave radiation near
Commenting on this in
the equator, and a maximum in subtropical latitudes.
the observations, Stephens et al. note that the near equatorial minimum is the
result of "high topped clouds associated with the ITCZ," and that the subtropical
maxima "are indicative of the subtropical dry zones (both over land and over
ocean)." While all the models simulate this general profile, the GFDL model has
significantly more outgoing IR at virtually all latitudes relative to NCAR, GISS,
Remember that GFDL has more incoming solar radiataion
and observations.
than observations and the other models (its solar constant is 7% larger).
the
GFDL model
is
in
energy
balance
the
higher solar
higher emissions to space, which can be shortwave, longwave,
constant
Since
requires
or both.
The sea level temperature was derived for each of the models from the
model output surface temperatures.
was
derived
from
the
Sellers
For observations the sea level temperature
surface
latitudinal profiles are shown in figures 7.6.5
for models and observations.
profile fairly closely.
data.
temperature
The
resulting
and 7.6.6 for each hemisphere
In the NH all the models match the observation
In the SH the match is not quite so good, with a tendency
7
Model climate sensitivity
83
for all of the models (especially GFDL in mid and high latitudes) to be warmer
than
observations.
The coefficients A and B were determined for each case by least squares
area weighting) the temperature
fitting (with latitude
longwave flux profiles.
7.7.1
profiles to the outgoing
The values obtained for A and B are shown in figures
There is not much that stands out from these
and 7.7.2 respectively.
figures except that in each case (models and observations) B is larger in the SH
than the NH.
To gain an indication of how well I is parameterized by eqn. 4.2 we
calculated I from the A and B obtained above, using the sea level temperature
profiles for T in each case.
Comparing these figures
relative
magnitudes
The results are shown in figures 7.8.1 and 7.8.2.
with figures 7.6.3
of I are preserved
and 7.6.4 it is apparent that the
in each case, but that the equatorial
i.e. figs. 7.8.1 and 7.8.2 show I
minimum profile has been lost in every case.
increasing
steadily toward the equator.
Equation
4.2 does not appear to be
in low latitudes, and we should
sophisticated enough to reproduce I accurately
remember this in interpreting the EBM results.
7.3.5
Albedos
Observations of the planetary albedo in the annual mean as a function
of latitude are tabulated in Stephens et al.,
the observation EBM albedo.
and were used for parameterizing
For each of the GCMs we did not have available
The albedo may be calculated however from the
direct output of the albedos.
net flux of shortwave radiation at the top of the atmosphere, SWnet, which we
The net shortwave flux is related to the albedo
do have for each of the models.
via
7.2
SWnet(x) = q/4 S(x) [1 - a(x)]
where q is the solar constant, a is the planetary albedo, and S is the meridional
distribution of solar radiation.
profiles for each
Equation 7.2 was used to calculate the albedo
of the models.
The results are
shown
together with the
Stephens et al. albedos for each hemisphere in figures 7.6.1 and 7.6.2.
7
In both
Model climate sensitivity
84
there
hemispheres
are
fairly
between
differences
large
the
for the
albedos
The profile of the GISS albedo is in general closest to the
The NCAR and
profile, increasing relatively smoothly polewards.
various cases.
observed
GFDL albedo profiles are flatter through to high latitudes, then rise sharply.
In the EBM, the albedo is parameterized as in eqn. 4.4:
a = ao + a 2 P 2 (x) + Su(x-xs). We pick ao, a2, and 8 by solving a system of three
simultaneous equations.
The first two equations are obtained by equating the
first two insolation components in eqn. 4.9 with their expansions in equations
4.13 and 4.14:
HO = f
90
0
H2 = 5f
0
S(x) [1- a(x)] dx = (1/2 8 S 2 ) x3 + ( 8 -1/2 8 S 2) x - 1/5 aX
2 S2 + 1-CCo-8
S(x)P2 (x) [1- a(x)] dx = (9/4 S S2 ) X5 + 5/2 ( 5 - 8 S2) X3 + 5/48 (S2
-2) x + S2 (1-ao-5) - 2/7 at2S2 - X2
The third equation we chose to be weighted to high latitudes to make sure that
the parameterization represents the albedo reasonably well in latitudes higher
than xs (i.e. that the 8 is well picked).
The chosen integral and its expansion is:
Hh = 1 S(x) [1- a(x)] dx = 9/20 a 2 S 2 x5 - 1/2 (a2S2 + S2 - a2 - aoS2 - 8 S2) x3 .
Xs
[ 1/2(a2 - S2 + aOS2 + 5 S 2 ) + 1- axo - 8 - 1/4 a 2S 2] x + 1 - ao - 8 - 1/5 a2S2
For
of the three
each
equations
above
left hand
the
side
was
integrated
numerically with albedo values taken from figs. 7.6.1 and 7.6.2, and with values
With S2 set to -0.482, the three
of S interpolated at each model's latitudes.
equations were then solved in each case for the unknowns aCo, a2, and 8. The
results are shown in figures 7.9.1, 7.9.2, and 7.9.3. ao is a measure of the global
average
planetary
albedo,
and
does
observations as shown in figure 7.9.1.
shown in figure 7.9.2,
the albedo.
much
not vary
All
between
models
ao values are close to 0.3.
and provides a measure
and
a2 is
of the meridional gradient of
The GISS C2's are close to observations, while the NCAR and GFDL
C2's are only half as large as observations.
This is to be expected from the
relatively flat albedo profiles for NCAR and GFDL in figures 7.6.1
and 7.6.2
(except in high latitudes where there is less area).
7
Model climate sensitivity
85
The
component,
third albedo
8, is shown in figure 7.9.3, and gives an
indication of the strength of the albedo in high latitudes (where there is ice
present).
8 is large relative to observations in the SH for the NCAR and GFDL
models.
S is about the same in both hemispheres in observations, but not in the
The GISS S's are small relative to observations in both
NCAR and GFDL models.
hemispheres.
To test the albedo parameterizations we substituted the values of ao, c 2 ,
and S obtained above into eqn. 4.4 and plotted the result in each case.
Comparing the results obtained thus so in figures 7.10.1
with the
and 7.10.2
albedos directly from models and observations that they were parameterized
(figs. 7.6.1
successful.
of curves
The two sets
in the
features
it is evident that the albedo parameterization
and 7.6.2),
actual
albedo
profiles
are relatively
seem
to
was quite
similar, and most of the
have been
in
captured
We can thus be relatively confident of having
parameterized profiles.
to
the
a fair
representation of the albedos in the EBMs.
7.4
EBM
Equivalent
results
Having parameterized
the equivalent EBMS
calculated outputs for each case.
calculated
from Is
=
A + BTs.
as outlined
above,
we then
First, the value of I at the iceline, Is was
Knowing Is, the transport efficiency, D, can be
calculated from eqn. 4.15.
For the sensitivity study, D was held fixed while the solar constant was
The iceline latitudes, x., corresponding to the
increased by 2% in each case.
increased
constant
solar
order polynomial
were
determined
by
solving
represented by eqn. 4.15.
expression
for xs
in the seventh
For both the current
climate and the climate with increased solar constant, we then solved for the
longwave radiation, temperature, and flux maximum via: I = q/4 Ho(xs) (eqn.
I2 = (Is-Io)/(P 2 (xs))
4.10);
6irR 2 DI
2
(x 3 -x)
(eqn. 4.12); To = (Io-A)/B;
T 2 = 12 /B and Fmax =
where x=1/13 is the latitude where the flux mamimum occurs. It
was then possible to calculate the climate sensitivity,
p
p,
for each case:
7.3
= qt (T0 2 -TO1)/(q 2 -q1 )
7
Model climate sensitivity
86
where the 1 and 2 subscripts refer to cases with the current solar constant and
The results, showing
with the solar constant increased by 2% respectively.
values for all inputs and outputs in the equivalent EBMs are tabulated in table
7.1.
The values of I
A relatively large I
are shown in figure 7.11.
value is
obtained for the GFDL model in the SH, which is related to the very small
negative value for T. in that case. All other T, values are not too different
from
observations.
D
The
case
each
are
shown
7.12,
figure
in
and
vary
For observations D is about the same in both hemispheres with a
The GISS D values are larger than this in both
to 0.26.
considerably.
value
for
values
close
is
which
hemispheres,
consistent
the
with
large
energy
poleward
transports
The NCAR D values are less than observations, as
Note however that the
is the case for the NCAR energy transport in the NH.
magnitude of the energy flux depends on 12 also, and will be discussed below.
simulated in the GISS model.
Figures 7.13 and 7.14 show values for I, 12, and To, and T2 for each case.
For the planetary IR emission, I, the NCAR model shows asymmetry between
the cases for the two hemispheres, whereas most other models and
For 12 (a
observations are relatively symmetric between hemispheres.
measure
of
the
of
gradient
meridional
I),
NCAR
is
large
relative
to
observations in both hemispheres, which will tend to offset the low D values in
The GISS model 12's are small relative to
calculation of the energy transports.
observations, which will tend to offset the relatively high GISS D values.
of To and T 2 for each case we have plotted the EBM
These profiles are
temperature profile via T(x) = To + T2 P 2 (x).
From values
meridional
The solid lines in the figures
represented by the dashed lines in figure 7.15.
show the actual sea level temperature in the relevant model or observations
for each
hemisphere
capture the levelling
particularly
in each
case.
The EBM temperature
off of the actual
for observations
temperature profiles
and the NCAR model
cases.
profiles
do
not
at low latitudes,
For instance,
the
large To value displayed in figure 7.14.1 for the NCAR NH model is not realistic,
7
Model climate sensitivity
87
The
since the EBM temperature profile is much too warm in low latitudes.
unrealistic T values in the EBM reflect shortcomings in the use of such a
The surface
simple model, and not additional errors on the part of the GCMs.
- outgoing IR parameterization
temperature
enough
sophisticated
account
accurately
to
in the EBM
drop
the
for
was not sufficiently
IR
in
near
the
equator, and this could lead to both To and T2 being larger than they should be.
For the global version of the equivalent EBM, the planetary average
0
To, for NCAR and observations is 19.9 0 C and 16.6 C respectively,
0
0
the meridional gradient of the temperature, T2 , is 38.8 C and 34.6 C
temperature,
and
which is too high.
respectively,
The energy transport maximum in the equivalent EBMs for each case is
The corresponding flux maximim from
shown in figure 7.16 as the solid bars.
earlier calculations for the GCMs and observations is shown as the crossfor the
cases
For the global
bars.
hatched
of the flux
latter, the value
For the EBM
maximum plotted is an average of the two hemispheric values.
global values however, the flux maximum is from the
global
EBMs
parameter
values.
The
with
parameterized
averages
hemispheric
the
of
results show reasonable agreement between the equivalent EBM flux maximum
and the flux maximum from GCMs and observations, and this despite
In the EBM results, the GISS
shortcomings in the EBM representation of 12.
transports
are large relative to observations,
and NCAR and GFDL transports
are small relative to observations as for the GCMs and observations in fig. 4.4.
The iceline latitudes input to the equivalent EBMs are shown in figure
7.17 (solid bars) alongside the iceline latitudes obtained from the EBMs when
The GISS
the solar constant was increased by 2% (diagonal pattern bars).
latitudes
iceline
climate
transports
are closest to observations
GISS
the
parameterized
the iceline latitude.
solar
constant
experiment.
37
model
37
has
from
in both
advantage
the
of
SST data,
observed
cases.
For the control
ocean
representing
which helps
in
heat
simulating
The equivalent GFDL EBM loses its polar ice caps for a 2%
increase,
it
which
This suggests
Wetherald (personal communication)
carbon dioxide in the GFDL model.
did
that
not
do
whatever
in
the
GCM
compensating
doubled
mechanisms
CO 2
are
notes that only about half the sea ice melted for doubled
7
Model climate sensitivity
88
being represented
they are still not as strong
in the EBM,
a compensating
influence as in their representation and sum in the GCM.
The ultimate goal of the EBM experiments was to determine the climate
sensitivity for each equivalent EBM to see how it compares with the sensitivity
For the GCMs the climate sensitivity to doubled CO 2 can be
in the GCM.
calculated by assuming that doubling CO 2 is equivalent to a 2% increase in the
Given
solar constant.
in
increase
the temperature
the the CO 2
doubling
experiment in each GCM, and the GCMs solar constant, the GCM sensitivities are
as follows:
NCAR:
p
GFDL:
GISS:
p
=
1370W/m 2 x 3.5K / 27.4W/m2
175K 3 8
=
1467W/m 2 x 4.0K / 29.3W/m2
200K
=
1367W/m 2 x 4.2K/ 27.3W/m 2
210K
The climate sensitivity calculated for each equivalent EBM is shown in
figure 7.18 (and listed in table 7.1).
They are fairly similar in most cases, with
values lying between about 150K and 200K.
and GFDL values in the SH.
poleward
than
in
The major exceptions are the NCAR
In these cases the iceline shifted relatively further
in response
other cases
to
the increased
solar
constant,
which would affect the sensitivity. The NCAR and GFDL EBMs have the highest
values for S (which contributes to the albedo in icecovered latitudes) in the SH.
Considering the climate
sensitivity for the global equivalent EBMs, not
only are they relatively similar for models and observations, but they are also
similar (for models) to the values obtained from the GCM CO 2 doubling
experiments listed above.
The respective equivalent EBM and GCM values are:
NCAR(190,175), GFDL(205,200), and GISS(145,210).
The sensitivities are similar between models and between the EBM and
GCM versions of the models in spite of the many obvious differences between
the models.
For instance, the energy
transports differ, the solar constant is
relatively larger in the GFDL model, the treatment of convection is different
between the GISS
etc.
These
and other models, the icelines differ, albedo profiles differ,
differences
should
influence
sensitivity.
the model
That
the
38
The temperature in the NCAR doubled C02 run was still increasing (relative to the control run), and
so the actual NCAR model climate sensitivity may be a little higher than this.
7
Model climate sensitivity
89
sensitivities are not so different suggests that either compensating factors are
at work, or the GCM climate sensitivity is not sensitive to the differences
between models. We reject the latter explanation on the grounds that the EBM
climate
sensitivity
is
quite
to
sensitive
the
solar
constant
(which
differs
between models), and the EBM climate is sensitive to the energy transport.
7.5
Discussion
Compensation in the EBMs and GCMs could be a coincidence (this is
unlikely, but if so it would not give us much faith in the models, since their
agreement would be meaningless then), or it could be related to the physics in
For the EBMs we have been unable to determine (at the time of
writing) what factors are responsible for compensation leading to a lack of
Potential candidates for
sensitivity to the energy transport differences.
the models.
sensitivity
compensation
include
those
related
to
feedback
ice-albedo
and
It is possible that the relationship between D
water vapour / cloud feedbacks.
D influences the energy transport which influences
and xs may be important.
the iceline position, xs. The position of the iceline changes the strength of the
The iceline position
ice-albedo feedback, which influences climate sensitivity.
The strength of the ice-albedo feedback is also
is also influenced by Ts.
influenced by 8 (replacing no ice by ice). ao and a2 are not important since
they merely specify the albedo in the absence of ice and so do not relate to the
For the cloud and water vapour feedback the only
ice-albedo feedback.
important parameter is B, which provides a direct measure to the feedback.
It is interesting to speculate at this point as to whether compensations
are necessarily present in the GCMS (and their equivalent EBMs) because the
GCMs are tuned to the current climate. Tuning is possible in the GCMs because
The
all the relevant physics are not included with certainty and completeness.
meridional temperature structure is approximately right in each of the GCMs
partly
compensations,
set,
they
because
tuned
to
the
is still much
on physics)
by
tuning
physics
to
current
climate.
If
this
induces
8, xs, Ts, and D may be constrained (not
then the parameters B,
since there
depend
are
the
in the models,
current
and these
climate.
This
parameters
raises
the
possibility that the climate sensitivity is constrained by tuning to the current
climate. A way to test this in the EBM would be to see whether it is possible to
7
Model climate sensitivity
90
change the sensitivity of an EBM (by changing B, 6, xs, Ts, and D) while still
the
maintaining
current
correct
climate
and
with
consistency
observations.
If it is possible, then we can be pretty certain that it is also possible to change
feedbacks related to B, 6, xs, Ts, and D in the more complex GCMs and change the
If it is possible to change these parameters in the
sensitivity in the GCMs too.
EBM and change the sensitivity then we know that the errors in the GCMs are
important, and would have some indication of where to look for them in each
In future work we will try to answer the unresolved questions identified
GCM.
in discussion here.
7.6
GCM
In the
problem
implications
validation
next
chapter
we
consider
aspects
greenhouse
of the
relevant in science and policy frameworks
change
Before moving
together.
to this level however (and by way of introduction) we follow on here with a
brief discussion on validation of GCMs, considering implications
transport
validation
Theories
of the energy
results.
of greenhouse
change
can not be properly
validated
without
validation of the 3-D GCMs, since greenhouse change scenarios rely so heavily
on GCM experiments in the absence of other tools to forecast climate change in
Attacks
detail.
consequently
GCMs.
on
the
concentrated
validity
to
of
greenhouse
a considerable
degree
change
on
have
scenarios
uncertainties
in
the
Newsweek (1989) quotes U.S. White House Chief of Staff John Sununu as
saying "you do not establish policies on the basis of incomplete
arguing against taking steps to mitigate greenhouse warming.
Apart from the
fact that the Government does this all the time where economic
concerned
39
, this begs the validation question.
incomplete is too incomplete,
models" in
models are
We really need to ask how
and how much uncertainty is too much, as there
will always be some incompleteness and therefore uncertainty in any model of
a complex system.
39
Handel (personal communication) notes that he would "place far more faith in forecasts of global
warming from increases in greenhouse gases, than in the belief that we are on the high side of the
Laffer curve so that decreasing the upper tax rates will lead to increases in government revenues."
7
Model climate sensitivity
91
research, the answer to these questions might
not be too important, since, for the accumulation of scientific knowledge we
might be more interested in learning about processes in the models than
For policy purposes, the
concerned with the actual predictions themselves.
For the sake of scientific
greenhouse change scenarios are important, since it is desireable to determine
what the potential impacts on society might be, if policies are to be put in
What then are the implications of model differences (such as in energy
transport), model internal inconsistencies (such as between energy transports
place.
and temperature structure
and iceline), model incompleteness
(such as lack of
ocean energy transports, or lack of potential for ocean transports to change as
climate changes), and other model uncertainties for validation of the GCMs and
greenhouse
their
change
of model
implications
The
projections?
shortcomings
depend
short
answer
is
on just what it is we
that
the
want to
If we want to know whether the regional results
in a particular area for a particular GCM are reliable or not, then the answer is
probably no. 4 0 The effects of ocean circulation are not included in the models,
determine from the models.
and
transports
energy
ocean,
affect
are
not correctly
partitioned
between
atmosphere
and
and this would affect the atmospheric circulation, which in turn would
regional climate (other factors such as poor representation of
hydrology
and clouds
and moist convection
would also be important in this
results are not reliable for the purpose of forecasting
regional climate does not mean that the model results are not reliable period.
There are other questions one can ask of the models for which the above
That is, for some questions, the results may
shortcomings may not be crucial.
regard).
That model
not be very sensitive to the remaining uncertainties, or at least not so much as
In this case
to yield scenarios that would warrant a different policy response.
Alternatively, it
the model results must be acceded some degree of validity.
may not be clear how sensitive the model projections are to the remaining
uncertainties,
though the perceived
risks associated
with the projections
be high enough to warrant policy responses regardless
may
of how much validity
we associate with the models.
40
For most questions at this level the answer is certainly no. For particular regions like midlatitude
continents and for questions related to drought occurrence, GCMs may provide some reliable
information.
7
Model climate sensitivity
92
In either case, the question being asked of the GCMs at present is what
the equilibrium global warming associated with a doubling of CO 2 in the
The results yielded by models are typically in the range 1.5atmosphere is.
0
These results are widely enough quoted in the
5.5 C (Dickinson, 1989).
but
on a sense of validity by reinforcement,
thay they have taken
literature
From whence do they derive their validity?
we must ask: why believe them?
Are they sensitive to the sorts of errors and inconsistencies we have noted in
transport
the energy
following
three parts
what the
radiative
to
response
temperature
is
forcing
the
the
(and
down
broken
with
response
the
into
The second part can
in the
the response
models
simpler
1-D
feedbacks
can
From
present.
feedbacks
of
absence
of
the absence
in
response
1.20 C temperature response
a net positive feedback
considered here) in the range of 4 0 C.
they
yield
It is
the
representation
(particularly
the
cloud
of the
in the
feedbacks)
radiative-convective
be
models,
1989)
i.e. about 1.20 C.
the range quoted
(personal
communication),
response
with
(warming)
to
for
feedbacks,
a
given
(for
that
most
is
the
since
models
three
the
contentious
are
feedbacks
which displays sensitivities between
by Dickinson 4 1).
changing the
positive
feedback
no
feedbacks
strength
1.9K
of clouds (though these results
and 5.2K for various plausible representations
are still not outside
and
determined,
This is demonstrated by the results from the
sensitive to model uncertainties.
UKMO model (Mitchell et al.
feedbacks
the
are incorporated in GCMs
sensitivities
and
and
feedbacks
generally yields results at the lower end of the above range.
When feedbacks to the
involves
part
third
determination of how long it takes the response to occur).
be
surface
the
what
determining
not much
and
small
relatively
involves
part
second
are
they
While there are
CO 2 is.
with doubling
forcing associated
The
about.
argued
first part is determining
The
(1989).
Dickinson
this determination,
in
uncertainties
CO 2 response prediction can be broken down
doubled
equilibrium
The
into
for instance?
than
As noted by Stone
surface
feedback
provide
the
a
cooling
temperature
greater
response
tendency
for
a
4
1They are also not outside the lower end of the range estimated by NRC (1979) for carbon dioxide
doubling of 3*C±1.5 0C.
7
Model climate sensitivity
93
negative
of
feedback
the profile
considering
same
the
strength.
of the feedback
This
be
can
understood
versus
ratio
response
/ no feedback
by
This ratio goes as 1/(1the feedback plotted in figure 2 in Schlesinger (1989).
f), where f is the feedback, and is much more sensitive to positive feedbacks.
This means that to reduce the sensitivity of the GCMs by a significant amount
we must include processes not currently accounted for, that yield substantial
While this does not prove that the GCM sensitivities
negative feedback.
right,
it
makes
the
proving
of
job
them
wrong
more
are
since
substantial,
processes not included in the models that would yield only weak feedbacks will
not suffice to change the results much when they are included.
The doubled CO 2 sensitivity results from the models cannot be tested for
some time however (until greenhouse
gases
to an equivalent
build up
CO 2
doubling, and the surface temperature has time to respond), and so confidence
A variety of validation tests
in the models must be derived from other sources.
are possible, and some of them were outlined in an earlier chapter. These tests
are ultimately important as reasons why those who find GCM results credible
find them credible, and so will be mentioned again here.
Firstly,
GCMs agree
the
qualitatively
in many
respects
in their
CO 2
This test
doubling simulations, and for their simulation of the current climate.
ultimately yields a half glass of water though, in that one's perspective of
whether the glass is half full (models agree) or half empty (models disagree)
depends on what is being compared and what one is trying to show.
Nevertheless,
GCMs
the
do
yield
similar
sensitivities
and
results
despite
The equivalent EBM model
differences in the way they have been formulated.
results suggest (barring coincidences) that the GCMs obtain similar
sensitivities
due
to
compensations
between
processes
related
to
ice-albedo
feedback and cloud and water vapour feedback, so that the outcome is perhaps
not too sensitive to differences in model formulation.
Further evidence supporting GCMs comes from their ability to simulate
the
seasonal
cycle
(which
involves
larger
temperature
changes
than
CO 2
doubling results), past climates (including the last glacial maximum), and the
Confidence in the GCMs is
climates of other planets such as Mars and Venus.
also obtained by
noting consistency with simpler models and physics
7
(which
Model climate sensitivity
94
can be more easily understood), and with observations of temperature and CO 2
in past climate changes (on paleoclimatological time scales and over the last
100 years).
None of this evidence taken alone offers clear confirmation of GCMs and
In validating a GCM there are lots of
their climate sensitivity experiments.
necessary tests that can be performed, but no sufficient ones (short of waiting
It is therefore virtually impossible to validate a
until hindsight is available).
On the other
GCM (or any other model of a complex system) unequivocally.
For
hand, everybody has some confidence in GCM results at some level.
example, the basic physics of the greenhouse effect is reproduced in a GCM
and is not disputed. How far one is willing to go in seeing utility in GCM output
on how much weight one gives to the various
validation tests that are available to support the types of processes that the
results depend on. Those that have some confidence in the GCM doubled CO 2
at a particular level depends
as
sensitivity
results
particular
validation
information
relevant
available
reasonable
test,
to
but
test
generally
through
the
a
so,
do
not
and
the
their
of a
basis
of
all
the
representation
of
the
judgement
subjective
models
on
physics.
The GCM projections are at least plausible at some level, and are gaining
In the next chapter we will outline a
attention in the policy community.
validation
framework
for
greenhouse
change
and
(models
projections)
in
a
science - policy context.
7
Model climate sensitivity
95
TABLE 7.1
Inputs and outputs from the twelve equivalent EBMs.
In each case the first
four letters denote the model or observations that the EBM was parameterized
The last two letters denote the hemisphere (Northern Hemisphere or
to.
Results are
Southern Hemisphere) or average of the hemispheres (Global).
given for both the current solar constant and for the solar constant increased
by 2%. The climate sensitivity calculated for each case is also shown.
CURRENT CASE VALUE SOLAR CONSTANT
NCARNH
NCARSH
GFDLNH
GFDLSH
GISSNH
OISSSH
CBSNNH
O3SNSH
NCARGL
GFDLGL
GISSQL
C3SNGL
WASTWH
0
XS
IS
10
0. 167
0.207
0. 275
0. 353
0. 415
0.365
0.266
0.253
0. 177
0.291
0. 385
0. 258
0. 355
0.920
0.839
0.956
0. 899
0.957
0.906
0.950
0.891
0. 880
0.928
0. 932
0. 920
0. 950
185. 64
188.60
193.44
208.97
188. 55
192.89
181.89
190. 48
186.77
200. 40
190.36
185. 85
191.S5
245.93
235. 30
244.04
244. 68
231. 92
233. 68
236. 78
239. 76
241.45
244. 73
232. 87
238. 31
236. 46
2% INCREASED
0
NCARNH
NCARSH
GFDLNH
CFDLSH
GISSNH
GISSSH
0. 167
0. 207
0. 275
0.353
0. 415
0. 365
OSSNNH 0.266
OBSNSH 0.253
NCARGL
0. 177
GFDLGL
0.291
-GISSGL
0.385
OBSNGL
0. 258
WASTWH 0.355
CASE SOLAR
OBSNNH
OBSNSH
NCARGL
GFDLGL
GISSCL
OSSNGL
WASTWH
TO
T2
FM
A
a
ALO
AL2
DEL
TS
GO
-39. 5
-36.7
-32. 1
-24. 2
-29.4
-28. 6
-35. 3
-33.9
-3. 8
-28. 9
-29. 1
-34. 6
-32. 1
3.85
5. 12
4. 71
5.20
6.06
6.00
5. 04
5.30
4.30
4.80
6.00
5. 46
201.4
196.3
216.9
210. 5
207. 1
203. 6
204. 7
206.8
198.9
213.7
205.4
205. 8
213.0
1. 98
2.29
1.81
2.07
1. 69
1. 95
1.82
2. 10
2. 14
1. 94
1. 82
1. 96
1. 63
0. 280
0. 283
0.341
0. 322
0.340
0.330
0. 317
0.303
0.282
0.331
0.335
0.310
0.:316
0.079
0.076
0.099
0.041
0.208
0. 202
0. 162
0. 168
0.078
0.070
0.205
0. 165
0. 146
0.209
0. 364
0. 148
0.251
0.079
0. 101
0. 164
0. 168
0.286
0.200
0.090
0. 166
0. 186
-7.9
-3. 4
-13.0
-0.7
-11.0
-5. 5
-12. 5
-7. 8
-5. 7
-6.8
-8.2
-10. 1
-13.0
1370.0
1370.0
1467.0
1467.0
1367.0
1367. 0
1365.0
1365.0
1370.0
1467.0
1367.0
1365.0
1365.0
T2
FM
A
ALO
AL2
DEL
TS
GO
3.76
4.32
4.61
4.79
6.10
6.02
4.98
5.28
4.09
4.47
6.03
5. 12
5.34
201.4
196.3
216. 9
210. 5
207. 1
203.6
204.7
206.8
198.9
213.7
205.4
205.8
213.0
0.280
0.283
0.341
0. 322
0.340
0. 330
0.317
0.303
0.282
0.331
0.335
0.310
0.316
0.079
0.076
0.099
0.041
0.208
0.202
0. 162
0. 168
0.078
0.070
0.205
0. 165
0. 146
0.209
0.364
0. 148
0.251
0.079
0. 101
0. 164
0. 168
0.286
0.200
0.090
0. 166
0. 186
-7.9
-3. 4
-13.0
-Q. 7
-11.0
-5. 5
-12. 5
-7.8
-5.7
-6. 8
-8.2
-10. 1
-13.0
1397.4
1397.4
1496.3
1496.3
1394.3
1394.3
1392.3
1392.3
1397.4
1496.3
1394.3
1392. 3
1392.3
5.16
CCNSTANT
XS
IS
10
12
TO
0. 956
0. 952
0.999
1.000
0.990
0.940
0.985
0.923
0.930
1.000
0.965
0.954
1.000
185.64
188.60
193.44
208. 97
188. 55
192.89
181.89
190.48
186.77
200.40
190.36
185.85
191.85
252.44
249. 49
250.23
255. 16
237.06
239. 10
242.64
245.76
249. 56
252.71
238. 15
244. 25
242.99
-76. 70
-70. 85
-56. 96
-46. 19
-50.01
-55. 98
-63. 59
-71.06
-78.75
-52. 30
-53.28
-67. 50
-51. 13
25.8
23.2
18.4
21.6
17.7
18. 2
20.8
18. 5
23. 7
20. 1
18. 0
19. 6
18.4
CLIMATE SENSITIVITY PARAMETER
NCARNH
NCARSH
GFDLNH
GFDLSH
GISSNH
GISSSH
12
-78. 34
22. 5
-84. 02
17. 0
15.0
-58. 10
-50. 13 16. 5
-49. 64
14.7
-55. 78
15. 4
-64. 29
17. 6
-71. 32
15. 7
-82. 82 19.9
-56.09
16.0
-53. 03
15. 1
-68. 04
16.6
14. 4
-52. 25
:
-38.7
-30.9
-31. 5
-22.3
-29.6
-28.7
-34.9
-33.8
-36.9
-27.0
-29.3
-34. 4
-31.4
B
1.98
2.29
1.81
2.07
1.69
1. 95
1.82
2. 10
2.14
1.94
1.82
1.96
1.63
BETA GOT/DG
164.20
309.73
171.02
253. 20
152.20
139. 10
160.74
142.62
189. 77
205. 57
145.07
151. 17
200.03
7
Model climate sensitivity
96
8
GREENHOUSE CHANGE VALIDATION FRAMEWORK
Introduction
8.1
about
greenhouse
there
are
a
validating
cultures
and between
between
also differences
their requirements.
in these
validation process
are
there
theory,
scientific
requirements
and policy
science
between
commonalities
many
While
contexts.
and policy
in science
change validation
for thinking
a framework
of this chapter is to present
The purpose
these
comparing and contrasting
By
in
two
the
arenas, it might be possible to identify ways of
facilitating effort between them to match their respective needs more closely.
We will argue that there are characteristics
which do lead to some degree
of the greenhouse
between
of mismatch in requirements
problem
science
We
and policy validation systems (thus compounding the validation problem).
treat each system as a dynamic system that is capable of responding to changes
We do not
in the other, and adjusting its requirements or effort accordingly.
venture to suggest just how dynamic each system is, but will try to identify
questions that could be presented from one system to a responsive other. Some
of the questions and problems identified in the framework will undoubtedly
turn out (under deeper analysis) not to be problems at all.
not
expect
exhaust
to
the
of
list
Conversely, we do
in
encountered
difficulties
evaluating
science-policy issues in the context of the framework.
greenhouse
The validation framework relies upon notions of how science and policy
systems
operate.
philosophy
validation
The
domain views scientific
theories
outlined
to be
as valid if they are
in
the science
consistent,
largely
and
than alternative
seem to explain more
about the way the world might work
views of the world.
Validation in the policy domain will extend upon the
scientific
version
acceptability.
to
consider
The policy
issues
system
such
as
useability,
is viewed implicitly
interest groups with varying degrees of power and
as
defensibility,
a composition
validated
describing the framework we will describe
and in what context,
and what definitions
of
benevolence.
The validation framework is represented schematically in figure 8.1.
subsequent sections
and
In
what is being
of validation
have been
97
settled upon and what the different components of the validation process are.
The
In this work we will not go much beyond presenting the framework.
following chapter will address only one of the questions derived from the
that of how well validation information is being presented from
framework:
science to policy arenas.
outline
Framework
8.2
framework shown in figure
In broad terms, the validation
the
reinterpretation,
transfer,
interpretation,
development,
involves
8.1
redevelopment
and so on, of information relevant to assessing the credibility of greenhouse
Representation of information flow by inputs, outputs, and
change theory.
feedbacks has been adopted for convenience, and the lack of a connecting line
between any two boxes should not be taken to imply that no relationship exists.
The origin of the problem is represented
somewhere.
in
inputs
top
the
in
anthropogenic
and
gas
corner.
The
primary
atmosphere
and
their
left
the
concentrations
greenhouse
change must come from
such as that posed by greenhouse
A problem
natural
contributions).
we
concentration,
have
inputs
climate
greenhouse
are
evolution
time
As
in the framework
a
result
changes
and
gas
result
of
changes
in
(the
of
as
associated
Building into the knowledge base on input is our understanding of
paleoclimatic history and the historical response of societies to climate change
impacts.
and
other social
perturbations.
The problem
greenhouse
change
at issue as a result of increases in greenhouse gases is
The validation framework is
or greenhouse warming.
geared toward the issue of greenhouse change, not to the greenhouse effect.
The greenhouse effect is
Definitions of these terms were given earlier.
reasonably well understood, and perfectly well validated by the fact that the
is about the predicted amount above the blackbody
Greenhouse change refers here to
equivalent value for the earth of 255K.
changes in the greenhouse effect associated with the increase of greenhouse
The
gases in the atmosphere over roughly the present and coming centuries.
global average temperature
theory
changes
of
greenhouse
(temperature
change
is
increases,
upon to
called
precipitation
8
project
what
changes
and
the
climate
circulation
Greenhouse change validation framework
98
associated with the greenhouse gas buildup in the atmosphere might
The development of the science toward this end has relied upon
changes)
be.
development
sophistication,
increasing
of
models
of
hierarchy
a
of
culminating in the GCMs, and on the reconstruction and interpretation of past
The topic for validation is the projections
and present climate changes.
and nature of the climate
(largely derived from the GCMs) of the magnitude
Whether the projections are
change expected over the coming century or two.
useful and valid for policy or not depends on what questions are asked of them.
A reasonable level of detail in the projections would probably go no further
than that articulated in Jaeger (1988).
range of response
Jaeger outlines in broad terms what the
and precipitation might be, and
sea level,
of temperature,
We do not try to validate whether or not
how they might change latitudinally.
climate changes of the order of a few degrees celcius or so (globally averaged)
We assume that they do.
imply significant environmental and social impacts.
understanding whether or not the expectation of
Validation is directed toward
a climate change of about this order seems reasonable.
communities
at
is
process
validation
The
national
the
primarily
Throughout
level.
and
science
in
here
framed
framework,
the
is mostly
policy, and the science community
policy refers mostly to national
policy
This is a guide only, and it
thought of as those doing science based in the U.S.
is expected that the framework would be useful in thinking about the
validation problem at the international level as well.
Once
potential
it
realized
is
that
policy implications,
and
community.
noting
policy
Once political
as
such
by the scientific
provides
implications
change
to validate
it in some
Publicizing
a
has
greenhouse
community in validating
to the policy arena.
theory must be communicated
science
theory
it is of interest to try
The effort expended
manner.
a
trigger
interest has been garnered,
the
the issue in
the
to
the issue
policy
is on
the
Whether the issue stays on the policy agenda or not depends
in part on perceptions of how well the theory and its projected social and
Many greenhouse commentators have
environmental impacts are validated.
political agenda.
noted the ability of phenomena
change
summer
on the political
(though
agenda.
essentially
such as a hot summer in keeping greenhouse
In the validation
in
meaningless
8
framework
validating
a single hot
greenhouse
change)
Greenhouse change validation framework
99
Whether
does strengthen the public perception of the validity of the theory.
people attribute causal relationship to the event or not, the event serves to
reify
evidence
supporting
plausible
more
the
to
reinforcement
provides
which
theory,
the
theory.
the
Once an issue is on the policy agenda there is at least some incremental
social response/adjustment in the way we live and do business (even if only
The mere fact that an issue is on the policy agenda means that the
slow).
policy process is relevant to the issue, since even if we do nothing about it,
that is a policy response.
inputs to the policy
Validation
than just the
more
come from
level
Once an issue has attention in the policy arena, then
and personal
economic, cultural, environmental,
the political, social,
While
dimensions of the issue will also be communicated in varying degree.
the scientific community might touch on each of these issues, other
scientific
community.
will
communities
provide
much
input.
of this
of the
Each
communities
(science and policy related) provides information back and forth in seeking to
evaluate the relative importance of an issue.
Each community
is
composed
interest
is by no means homogenous.
of researchers
groups,
from
government,
non-government-organizations,
Each sub-community
The science community
universities,
industry,
and
private
organizations.
influence
and unique factors that
is subject to common
public
the type of science they do and the way they view it. Scientists may even see
their responsibilities and duties differently, depending on where they work.
Superimposed on this is their own personal world views and values, which can
influence the way they approach the greenhouse validation problem.
Though
scientists
can
be
as
viewed
interest
groups
beholden
to
particular interests or world views, or to a particular sense of responsibility,
they have instituted a widely agreed upon procedure for regulating the
development of their research.
Imperfect as it might be, that procedure is the
process of peer review
and
publication in refereed journals.
to
the
notion
share
an
involving
adherence
hypothesis,
experiment,
method';
of 'the
scientific
testing,
verification,
8
Scientists also
and
a
process
repeatability
Greenhouse change validation framework
100
The scientific method is ideally guided by a rationality (call it
where possible.
science
rationality)
(usually
within
based
reasonable
upon
framework
a
and
knowledge
Panels
the
of
understanding
issue
convened
to
on
report
issues,
provide
frequently
testimony
channels
government.
Scientists
are
conceptions
and differing
bases
this
industry
For
received.
to
cultures
across
communicating
and
groups
to
of rationality
be
may
an
with
apply
to
based
economic
In each
For the public this may be a socially based rationality.
rationality.
and
for scientists to the media, the public, industry,
communication
information
through more than just the journal
and
reports,
interviews,
working
are
committess
and
documentaries,
the
of
development
is communicated
output
Scientific
differing
of the
change).
(greenhouse
papers.
quantifiable
papers that represent formal descriptions
in science is science journal
of
evaluation
The output of this whole process
characteristics is possible and meaningful).
state
of
measurement
where
and
choices
logical
case some form of reason is still the basis for evaluating information, but the
If
factors considered and the bounds and constraints applied may be different.
for evaluation
the frameworks
to
frameworks
scientists this
they
where
then
isolation,
make
implies
of
understanding
are not the same between two communities
intersect
information
shared
supplementing
the
they
their
enlarge
For
comprehensible.
notions
dimensions
to
need
mutually
scientific
non-scientific
will
with an
of rationality
the
of
in
issue
in
its
communication.
After validation information is presented by scientists it is received and
For the policy process it is important to determine
interpreted by the public.
the potential
projections
or
likely
impacts
from theory.
(social
and ecological)
associated
essentially
about
evaluation
the mid-range
of the
scenario
by the probability of this scenario
by
the political
the
While it is difficult to factor in potential extreme
climatic outcomes and their impact, it is easier to set up
thinking
with
acceptability
impacts.
The
for how climate
might
likely
an expression
for
is
likely
impact
change
multiplied
This is then further influenced
occuring.
of the outcome
and perceptions
of its policy
implications.
8
Greenhouse change validation framework
101
The
modelling
and
theory
change
is in some ways
scenario
of
Development
projections.
as an
determined
from
determined
is
scenario
greenhouse
mid-range
the
greenhouse
mid-range
average of the most influential
results in the literature, and perhaps also by what those scientists comprising
Scientists view of the certainties and
the consensus group will agree to.
will influence how the scenario is constructed.
uncertainties
The probability
of the mid-range scenario is a measure of the credibility of theory and model
Credibility is influenced by
projections as determined at the policy level.
perceptions of the consistency and plausibility and support of the models and
theory, and by perceptions of the degree of consensus among scientists in
supporting
the projections
standing
relative
(the
of dissenting views
would
Views of the uncertainty, such as how critical they are,
and in which direction they are likely to influence the results, would also play
a role in determining how much probability to associate with the scenario.
also be important).
by how effective the
presentation has been in matching
The presentation of the climate
level.
policy
and
of the science
policy requirements.
change projections must be meaningful
at the
information must have clear policy implications
concrete terms that people can understand and
The projection
translateable
be
in the presentation
is influenced
acceptability
Political
into
An example of effective presentation in this regard is shown by
In presenting greenhouse change scenarios from the
Hansen et al. (1988).
0
GISS GCM, rather than just showing results in terms of a 4 C or so temperature
appreciate.
rise
(which
sound
doesn't
like
much
to
most people),
they calculated
the
in such things as the number of days per year with temperatures
exceeding certain amounts in several major cities (which has more local
increase
meaning).
uncertainty
policy
Political
and
how
community
is
acceptability
they
have
interested
is
been
in
also
influenced
characterized
knowing
what
uncertainties is, and how long it might take to do so.
in
the
by
of
views
the
presentation.
The
to
reduce
potential
Uncertainties
frequently
42
a source of inertia in environmental policy , though there are also
characteristics of uncertainty that can be construed as implying action as well
act as
as inaction.
A perception
that uncertianties can be reduced relatively quickly
42
Uncertainties have been framed to act this way by U.S. Chief of Staff John Sununu on greenhouse
policy to date.
8
Greenhouse change validation framework
102
This view of
might bode for delay to resolve them before taking policy action.
in urging delay on the
the uncertainties is taken by Marshall Institute (1989)
basis that uncertainties can be reduced enough to make a difference in 3 to 5
years.
If on the other hand the perception
reduced
quickly,
less
be
might
bode
to
likely
are
there
while
to
costs
then
delaying,
term
of short
delay
for
cannot be
is that uncertainties
uncertainties
the
policy
responses.
Discussion of the role of uncertainties in the validation process will be taken
up again later.
Received views of the potential risks are also important in establishing
and the worst case
seem high,
i.e. if the risks associated with not acting
of a theory.
political acceptability
scenarios
theory
then the
seem plausible,
is
likely to gain higher acceptability in the face of interests that find the costs of
unacceptable.
action
come
projections
such
as: the perceived
perception;
and considerarion
depends on the evaluation
considered acceptable
cost and benefits of alternative policy responses;
justice, equity,
morality; perception
constituencies;
the
of the relative
of greenhouse
relationship
policies and their constituencies;
and
issues
and
to
other
the political and legal structure; the role of
This is not supposed to be
acceptability
of greenhouse
the likely impacts of greenhouse change,
it is relevant
affect
of the
perceptions
scenarios.
In evaluationg
to ask: Are the likely
(to policy)
answer
this
to
importance
question
adaptive
depends
impacts worth
on
who
their
they
social
come
you
The
worrying about yet?
Perspectives
are.
and the problem
of the problem differ widely,
for different people,
society
issues of
It is only intended to give an indication that a myriad of
can potentially
change
risk
strength of coalitions
change
lobbying, sanctions, protest, cooperation, capital etc.
factors
are
of factors
of relative interests and distributive burdens;
perceptions
an exhaustive list.
implications
their
and
projections
change
greenhouse
Whether
change
realms.
environmenal
and
social,
economic,
the
from
greenhouse
of the
acceptability
political
on
influences
Major
on
the
itself is different
depending on what part of the world they live in, how
and
from.
physical
In
a
infrastructure
national
8
is,
context
and
what
greenhouse
segments
change
of
will
Greenhouse change validation framework
103
implicitly or explicitly be given some degree of priority by whether or not it
on the
stays
If it is dropped
policy process.
and engages the
agenda
or
deferred from the policy agenda, then the implicit answer is that the scenarios
are
not worth prioritizing
on the
stays
change
If greenhouse
yet.
policy
agenda with some priority, then either policies (of some kind) can be put in
place or
a place
In either case,
paid.
lip service
the agenda
on
yields
legitimacy to the issue.
do they
and
social
make
we
articulated,
procedurally
just,
substantively
Further,
are
policies
Where
any
to
a result
of greenhouse
warming
or to
other
of
life,
quality
pollution,
are
goals.
implementable
greenhouse
policies
change
greenhouse gas concentratons
nothing'),
they
stratospheric ozone, etc.).
economic well being, resource depletion,
As
whether
ask
have
(energy,
problems
to
and
correct,
difference
environmental
need
(which
in the atmosphere
could
include
will change.
'do
This
in turn will lead to exacerbation or ameloration of greenhouse climate change
Climate will change (both in actuality, and as we measure and perceive
trends.
the changes),
and climate impacts and projections of the impacts will change
This in turn will feed back on
(together with confidence in the projections).
greenhouse
we evaluate
the way
at each of the various
projections
change
levels throughout the coupled system outlined in the framework.
components
Validation
8.3
In discussing validation of GCMS and greenhouse change projections we
have not attempted to define validation according to a single criterion with a
It is virtually meaningless
single yes/no answer.
projections
judgements
values.
are
valid
or
over questions
to say that the models and
that
involves
that come down to matters
of faith,
invalid
period,
since
too
many
belief, and
The answer depends on who you are and what your world view is, and
ultimately ends up revealing more about the person giving the answer than it
provides useful information to the question
simple
whether
manner
present
the
question
uncertainties
science,
transcends
justify
When phrased in this
at hand.
a policy
since
response
it
implicitly
or not.
asks
Schneider
(1989) notes that this is not an issue resolvable by scientific methods.
8
Greenhouse change validation framework
104
We can learn more about the validation question by phrasing it instead
More specifically, is
credible and plausible:
particular greenhouse projections
Are
such as:
in terms of related questions about subsets of the projections
there some level of aggregation of the results at which the results are not only
credible, but also useful from the point of view of policy development?
While
it might be said that framing the validation problem in utilitarian terms avoids
the question, we need to think about the context of the validation question.
Strictly from the point of view of scientific research,
validating
greenhouse
projections
make
doesn't
a focus
immediate
on utility in
sense,
since
we
would like to think that we do not decide scientific questions simply on the
In science though, we probably
basis of whether the results are useful or not.
wouldn't spend so much time worrying about the actual veracity of GCM results
beyond the utility so derived in improving the models and our understanding
Greenhouse science is coupled to policy, since
of important climatic processes.
the projections have social implications.
then the results must be credible,
results as an input to policy considerations,
defensible, and ultimately
useful.
The problem still remains as to how credible
greenhouse projections
have
broken
the
scientific,
validation
process
utilitarian,
For the scientific
projections are credible.
which
the
observations
projections
up
into
meaningful
paradigmatic,
four
component
the
and
components,
and
listed
The four components are
the overriding
question
is
whether
the
To answer this we must ask whether the theory from
derive
credibility
utilitarian
useful
To this end, we
and political.
is
supported
by
and whether the uncertainties
are
consistent,
and other theories or models,
crucial in undermining
For
and how useful particular
should be in a science-policy context.
particular questions or tests under each component.
labelled
are to use the
If policy practitioners
whether
it
is
and in formulating the policy
component
information
main
the
can
be
consideration
derived
from
the
response.
is
whether
projections.
Certain aspects of the projections will not be useful, either because they are
not credible, or because they do not match the needs or ability of the policy
system to comprehend them or translate them into suitable form.
8
By
'suitable
Greenhouse change validation framework
105
form' it is meant that information may be presented in various ways that make
its
more
appear
contents
or
less
tangible,
meaningful,
and
to
palatable
a
system that, we will suppose, processes certain information forms more easily
than others.
For example,
local scale interpretations of global scale changes
and numbers may be more comprehensible to a policy audience.
For the paradigmatic
component, focus returns to the theory and tools
we have a more powerful
theory to explain
theory than greenhouse
aspects of past, present, and potential climate changes?
to project
tools
available
changes
climate
as
Is the theory defensible?
greenhouse gases?
a result
Are GCMs the best
of perturbations
by
intact?
is the political, which
The final validation component
relevant
Does it have consensus support?
is the greenhouse paradigm or research program
In short,
Do
support.
behind the projections to ascertain whether they have widespread
essentially boils
are the projections (and perceptions of what they imply) acceptable
down to:
in their current
form?
Having identified what we see as important questions and tests relevant
to validating greenhouse change theory, we will now discuss more specifically
Categorization
how the questions relate to the policy and science communities.
of the discussion will follow under the terms of uncertainty,
credibility,
and
useability.
8.3.1
Uncertainty
In the policy
arena
theory is seen as credible
since
there
are
perceived policy
political
are critical
uncertainties
opposed
(and
in
implications of the theory who will exploit
This exploitation takes place in an environment
necessarily
The uncertainty
and forceful or not.
constituencies
and strategic
understand
the
context and
in whether
is strategic,
favour)
to
the
the uncertainty.
where many people will not
relevance
of the
uncertainty,
and
so
might be shaken into thinking that the whole theory is damned (or proved).
In the
that
science community
are still important,
except
then there is no obvious
strong
the uncertainties
if there is no real competing
theory,
8
Greenhouse change validation framework
106
In the science
the uncertainty.
to exploit
in position
science constituency
community, to win the support of others it is generally not enough to simply
Scientists will want to know what alternative views
critique the uncertainties.
to
proposed
being
are
or theories
the
supplant
being
ones
old
critiqued.
If well
Otherwise, why should they abandon a reasonably successful theory?
known
greenhouse
support
in
the science
they must
then
community,
Lindzen 4 3 are to acheive
as
such
critics
theory
change
progressive
forward
put
views to explain past observations on which we currently rely on greenhouse
In the science community
theory to some degree.
in
uncertainties
in
perspective
relative
it is easier to place the
to
importance
the
in
certainties
the
theory, since the broad context of the theory and what supports it is generally
course
of
(though
understood
the
in
ignorance
of
degree
the
science
community is also fairly high at times).
For policy purposes it would ideally be nice to know how long it would
take
reduce
to
trying
to
with
up
come
there there
quantitative
answers
be
would
to
less
quantify
to
uncertainties
(which
compounding
the
the
by
leads
uncertainties
their very nature are
to
disagreement
difficult
to
the
The exercise of
uncertaintes are and how long it will take to reduce them.
trying
to
impetus
important
how
just
for
information
important
also
though
programs,
research
scientific
is
This
uncertainties.
the
the
over
quantify),
thereby
uncertainty.
community
In the science
uncertainties can not be reduced.
an appreciation
that some of the
We are stuck with them.
For example, the
there
is
natural variability of climate and the behaviour of volcanoes lead to variations
in
the
record
temperature
greenhouse
signal,
and
evolve in the future.
these
sources
essentially
are
irreducible
which
more
it
make
difficult
more difficult
to predict just
to
slim.
or potentially
categorization
The
reducible)
of
a
out
how temperature
The prospects for reducing uncertainties
quite
extract
will
derived from
uncertainties
and the task of placing
(as
them
in context in validating the theory lends a level of complexity to greenhouse
validation which makes it more difficult for the policy system to incorporate.
43
Richard Lindzen is an MIT meteorologist who has received media coverage in the U.S. on his critique
of greenhouse change theory.
8
Greenhouse change validation framework
107
Credibility
8.3.2
of greenhouse
In assessing the credibility
it is
agree sufficiently
with
projections
whether models and
to ascertain
necessary
theory and projections
This is an exercise for which it is helpful to
observations and simpler models.
have a grasp of the way the climate system potentially works. A knowledge of
the potential role of various forcing factors and of internal variability is
useful
difficult
be
and it may
important,
the
observed
climate
Again,
and projections.
theory
from
expected
whether
assessing
in
the
to communicate
relevance of this to the policy domain.
response
broad
matches
that
is
quite
context
breadth,
the
depth
and
For example, how does one view the
They indicate
Newell et al. (1989) sea surface temperature (SST) observations?
Is that
essentially no trend in measurements of SSTs over the last 130 years.
projections
model
with
consistent
for
doubled
CO 2 which would imply
a
In the
warming to date of between 0.4K and 0.8K (Ramanathan et al., 1987)?
science system there are a variety of Lakatosian type positive heuristics
(Lakatos,
1970) which
program,
and
observations
serve
which
a protective belt around the core of a research
maintain support for the theory in the face of
form
to
apparently
do not
support theory. 4 4
Firstly, in a science
context it is widely recognized that the SST data itself is open to question, since
the measurement techniques used in collecting the data have changed
considerably and are difficult to correct for.
Secondly,
it is also recognized
that it is not even necessary for the oceans to have warmed at all so far, and
The oceans may not have warmed, since
not be inconsistent with the theory.
the ability of the oceans
estimated
range,
and so
to take up heat may be at the high end of the
we may not see the warming at the surface yet.
Additionally, the greenhouse warming expected to be realized to date (0.4-0.8K)
is still small relative to natural variability and influences of other potential
forcing
factors.
These
positive
heuristics
derive
from
knowledge
of the
Lakatos (1970) notes that "few theoretical scientists engaged in a research programme pay undue
They have a long-term research policy which anticipates these refutations.
attention to 'refutations'.
This research policy, or order of research, is set out - in more or less detail - in the positive heuristic
of the research programme. The negative heuristic specifies the 'hard core' of the programme which is
'irrefutable' by the methodological decision of its protagonists; the positive heuristic consists of a
partially articulated set of suggestions or hints on how to change, develop the 'refutable variants' of
The positive
the research programme, how to modify, sophisticate, the 'refutable' protective belt.
heuristic of the programme saves the scientist from becoming confused by the ocean of anomalies."
44
8
Greenhouse change validation framework
108
broader theory and climate context and may not be well incorporated into the
This makes evaluation of the theory more difficult there.
policy realm.
and disagreement between GCMs is also important in
Testing agreement
That models disagree, in the modelling community
establishing credibility.
is
viewed mostly as grounds for improving the models, since a perfect match in
For the modelling community to stay happy, there
all detail is not expected.
and
models,
3-D
the
between
consistency
reasonable
be
must
appropriate
In addition, those disagreements that do exist
agreement with simpler models.
should be traceable to differences in model formulation, and should not lead to
Where the energy transport is
drastic differences in model sensitivity.
Outside the modelling community
concerned at least, this seems to be the case.
however (in science and policy domains) any model disagreement is sometimes
seen more in terms of the inability of science to get its act together.
This has
the affect of making the theory less defensible.
Defensibility of the theory and projections is crucial, since if society is
going
to
to
efforts
policy
commit
and
regulating
making
then
changes,
policymakers want to be able to defend policy decisions against critics arguing
In the policy arena, to defend policy decisions and the
the uncertainties.
credibility
theory,
of
demonstrate
that
there
policymakers
is a reasonable consensus
of consensus
theory
in the
public
Kerr (1989)
be quite
At the present time,
mind.
degree
the
characterizing
differently.
and dissent can thus
of
dissent
over
fall
among scientists
and
back
supporting
Views of the relative
the theory, and that dissent is relegated to the fringes.
degree
to
want
least
at
will
important
evaluating
can be
found
projections
quite
examples
greenhouse
in
in Science magazine characterizes Lindzen's position
as part of the fringe under the heading "Greenhouse skeptic out in the cold. A
prominent meteorologist says the greenhouse warming will probably be a
bust; experts in and out of the climate community staunchly disagree with this
latest iconoclast."
are
challenging
By contrast, Stevens (1989) in the New York Times ("Skeptics
dire
'greenhouse'
views")
characterizes
both
dissenters
and
advocates of the theory as on the fringes of a climate community that has for
the most part not made up its mind on the issue.
defend
their
decision
to
support
greenhouse
A policymaker wishing to
response
policies
may
point
happily to the Science article, but not quite so happily to the Times article.
8
Greenhouse change validation framework
109
Whereas a modeller
realms, though probably more of an issue in the latter.
might respect scientists such as Lindzen, but dismiss
incomplete
(by
general
community
must
deal
Nevertheless,
it is important
theory, tolerating majority
Clark
and Majone
to ensure a "fair play
does
community
then the policy
essentially
a 'head count'
conceptual
1985).
is
contexts has
approaching
for
community
can hardly be blamed
approach.
The
weighing up
inquiry
If the science
level."
framework
conceptual
a
have
scientific
when working in policy
deep
at any
explored
problems,
in evaluating
of ideas"
"implications of how
the
note that
not
scientist.
emminent
an
policy
the
and minority viewpoints (Clark and Majone,
prepared to handle minority viewpoints
not been usefully
is
he
that
fact
the
with
yet
as
critique
his
unpublished,
and
standards)
science
and policy
in science
is important
accorded to dissenters
weight
The
such
for resorting to
of consensus
support
and relative standing of dissent among scientists by the policy community is a
It is an attempt to quantify
proxy measure of the credibility base in science.
researchers
via informal poll among influential
theory
represents
(following Kuhn,
the dominant paradigm
science provide necessary
the
science
community
of whether the
1962),
or whether
replaced by a progressive shift in
the research program has been successfully
view (following Lakatos).
the questions
Perceptions of widespread support for the theory in
If
grounding for defending the theory in policy.
for evaluating
can develop frameworks
science-policy
problems, then perhaps it will be possible for the policy community to respond
change the necessary
progressively 4 5
and
sophistication
more
with
conditions
for defensibility.
The net result of relative differences
in assessing credibility
of theory
in science and policy domains is relative differences in the degree of stability
In science, protective heuristics and a validation philosophy
afforded theory.
that
takes
into
account
a
gestalt
knowledge of the climate system
for
theory
in
the
face
of
theory,
of
models,
observations,
is more resilient in maintaining
by
threats
uncertainties
and
and
credibility
perceived
As outlined in Clark and Majone (1985), a Lakatosian progressive shift in the policy arena "may be
said to be progressing as long as it succeeds in disposing of issues, i.e., in moving them from the stage
of contention to a class of issues which the actors in the policy process judge to be in a state of
satisfactory, if temporary, resolution."
45
8
Greenhouse change validation framework
110
appears
validation philosophy
the
community
In the policy
inconsistencies.
to be less pervasive, and so relative stability of the theory against apparently
philosophy may be
The science validation
is diminished.
evidence
contrary
It is less
less pervasive in the policy community for a number of reasons.
Continuity and
widely known and not institutionalized into the policy process.
which must deal with many
focus is also lacking in the policy community,
issues, of which greenhouse change is only one.
buffers
that entail
related
theory
skepticism
any
science-policy
agenda.
The
reasons for this can
found in the point made by Lindzen (1989)
have
- tied
paralyzed
quickly
misplaced
excruciating
into
problems
rigorously
be
cannot
and
be
its
knots by
which
proven
would be
response, our society
forth a statutory
If each called
impossible.
of environmental
unlikely
exceedingly
objectively
although
of
that "given a week or so, I would
long list
a
generating
no difficulty
issue
accepting
toward
that enters the policy
a system
requires
community
policy
the
theory,
stabilizes
and
protects
in place that
of buffers
a system
community has
the science
While
well meaning
but
enthusiasm".
Useability
8.3.3
If greenhouse change scenarios are to be useful as a basis for policy,
believable
they
must
be
meaningful
(have
policy
then
policy implications).
they
so sensitive
not
would
sensitive
and not too
implications),
(or at least not
uncertainties
(otherwise
as to yield
be
defensible),
to the residual
results with
different
Trimming GCM output down to a level of specificity at
which the scenarios are justified by this type of reasoning is a difficult task.
is an example
(1988)
Jaeger
changes
in temperature,
this,
of an attempt to do
sea level,
specifying
expected
terms,
together
in broad
and precipitation
with outer limits at the 1/10th probability on what might occur.
as
over what
a community
level
of detail
Scientists will
to
accept
not easily
agree
projections,
some believing that one shouldn't engage in the exercise at all.
Where
scenarios or guidelines
are derived from GCMs,
their validity is obtained on the basis of physical
Hansen
et
al.
(1989)
note
that
the
frequency
8
plausibility.
and
severity
confidence
the
in
For instance,
of
droughts
Greenhouse change validation framework
111
in greenhouse
increases
simulations
change
with the GISS
GCM.
Though
drought is a regional phenomenon and the GCMs are generally not regarded as
being reliable
for
forecasts,
regional
in
warmer temperatures
note that
(1987)
drought
In a study with the GFDL GCM, Manabe and
intensification seems fairly robust.
Wetherald
behind
physics
the underlying
midlatitudes
continental
lead to an earlier occurence of the snow melt season followed by a period of
intense evaporation. This leads to a CO 2 induced reduction of soil moisture in
Enhanced summer dryness leads to a reduction in low cloud amount
summer.
and
which
precipitation,
surface
continental
precipitation
the
early
influence the location
processes leading to
and
distribution.
intensification
location
policy
the
change
communities
to
at
it
maintain
a
of
the soil
level
low
in
longwave
such as atmospheric
These factors
will change.
of droughts
are not necessarily
theory
the physical
Nevertheless,
in a general
continental
midlatitude
areas)
sense (averaged
are
manifestly
associated with the prediction.
plausible, and confidence is thereby
Greenhouse
decrease
and introduce uncertainty into regional forecasts
drought behaviour
of exactly where
the
Both
the
reaching
energy
evaporation further reduce
and timing of droughts
in the GCMs,
well simulated
solar
evaporation.
help
and
the
Hansen et al. note that there are other factors that
patterns and ocean temperature
time
potential
summer
throughout the summer.
over
both
and the increase of potential
during
moisture
and
increases
and projections
In science
slightly differently.
science
and
the actual projections
and
are
used
in
What is
are not quite so important.
level at which they are accepted
important is whether we can learn more about the system with the theory (as
conceptual and numerical models) than with alternative models.
in the sense that it provides
useful
a superior tool for
The theory is
about
learning
In policy, the projections are useful at
climate system and climate processes.
some specific level if they have broad support in the science community,
clear and meaningful
the
climate change
implications whereby
and
may be related to
change.
social
Though
of
many
the
questions
asked
by
the
science
and
policy
communities are common, the emphasis and extension eventually diverges.
science,
validation
in
terms
of the
most
successful
paradigm
or
In
research
program affords theory a degree of stability, while still leaving it quite open to
8
Greenhouse change validation framework
112
attack
and lively
dissent.
terms
of whether
they
In policy,
defensible
are
attacks
against
more
in
arbitrarily
of being
In the face of political opposition, it must be clearly demonstrated that
based.
A way of doing this seems to
the projections are reasonable and not arbitrary.
be
are validated
the projections
via
showing
the
that
part, then the issue presumably remains
satisfied in substantial
projections
is
If this is
implications.
impacts have policy
and that the likely
dominant,
the
supporting
view
consensus
on the policy
agenda.
the
Assessing
characteristics
change
Greenhouse
8.4
to
according
projections
of greenhouse
validity
criteria
such as credibility and usefulness is made difficult by the fact that we as a
about
determining
in a manner that takes into account the characteristics
of complex
do
society
credibility
seem
not
to
how
know
make
to
decisions
The development of a critical capacity for the
issues like greenhouse change.
appraisal of scientific enquires with policy implications
and Majone,
(Clark
surrounding
of the uncertainties
The nature
1985)
is still in its infancy
the
greenhouse issue alone seems to defy our ability to assess the relative merit of
the projections
using
and credibility.
We
problem
will
and notions
frameworks
conventional
outline
various characteristics
uncertainty
of the
greenhouse
for
assessment
difficulties
might contribute
and consider how they
about
and resolution of the problem at the policy level.
The
problem
greenhouse
is
global
in
cause
and
effects,
but
the
contributions to the causes derive from many different social sectors (there is
no one culprit) and countries.
and between
countries
Cooperation between
is required
(unilateral
action
industry and government
is insufficient),
but this
is made difficult when the effects and costs and benefits of action are different
for different regions.
The perception of 'winners'
and 'losers' compounds this
difficulty.
Greenhouse
change
involves
large
systems,
both
the
socio-economic
and the ecological, so the inertia of the systems becomes important.
Lags in
the climate response make it difficult to validate the projections and to prove
that the issue is serious.
Meanwhile, delay in industrial response means that
8
Greenhouse change validation framework
113
If the climate sensitivity is large, then
greenhouse gases continue to build up.
the stored response will grow bigger with time, and it is largely irreversible.
In a methodological sense, Science (to avoid refuting all theories at all
times) is forced to build resiliency and stability into the evaluation of theories.
Policy (to avoid accepting all issues at all times for its agenda) is forced to build
instability into the evaluation of theories so that only the most well supported
issues survive by virtue of the overwhelming evidence supporting them, or by
way of political acceptance of the implications of the issue.
The
public,
forecast
changes
temperature
global
and impacts occuring
small and distant.
can appear both
impacts of greenhouse change
The
next century
sound
small
like
numbers
and beyond are difficult to factor
People in
the activities causing the problem are not viewed as reprehensible.
to
seem
countries
as
consumption
the
The results of the problem are potentially severe, but
into current decisions.
industrialized
to
quite
normal,
no
sufficient
view
large
scale
of
the consequences
and
fuel
fossil
energy
behaviour
this
are
unintended.
There
are
of greenhouse
tests
change,
only
necessary
The
ones, which can be exasperating to those seeking certainty to act upon.
uncertainties are severe and will be slow to be reduced.
uncertainty
of the issue
allow respectable
The complexity and
scientists to take
opposing views.
This is confusing to the public.
The
greenhouse
projections
rely
in part
on complex
climate
which are difficult to interpret and validate, and expensive to run.
and nations have GCM modelling capability.
small isolated groups.
Policymakers
community of modellers, which
projections
models
Few groups
This expertise is confined to a few
are forced to rely on a relatively small
makes
it harder to make
the case that the
are well tested.
The net result of these difficulties is an apparent mismatch between the
problems
faced
by
scientists
in
evaluating
and
presenting
information
greenhouse change and its credibility, and the needs of the policymakers.
on
The
policymakers would like to defend their policies with some surety and support,
8
Greenhouse change validation framework
114
the
describe
concrete
and give
comprehensively,
problem
of the
indications
Meanwhile, they are operating in an environment
implications of not acting.
in which the risk perception is low or distant, and the perception of the costs
of action is high and action difficult to negotiate, while perceptions of benefit
from effective
divided
both
scientists
incorporate
Much
critically.
and
apt
are
to
of the uncertainty
policymakers
inertia
With much political
responses are low.
uncertainties
community,
incorporated
so
policy
are
stuck
exploited
be
and a
than
rather
cannot be reduced however,
with
and
it,
must
learn
to
it into issue evaluation.
Reappraisal
of the evaluation
issues
of science-policy
can
draw
upon
For the part of the
potential inherent in both science and policy communities.
scientists, they can change the way they research and present material so that
it more effectively matches the needs of policymakers, and is more effective in
providing validation information
community can
work on setting critical
uncertainties,
minority
evaluation
and
viewpoints,
following Clark and Majone.
The science-policy
in policy relevant terms.
other
standards that
incorporate
evaluation
dimensions
The policy community must be geared up
to
playing a role in setting evaluation standards so that they will be useful once
applied in the policy
Whether it will be necessary for the political
arena.
framework to undergo some form of transformation or paradigm shift if it is to
be able to come to terms with the challenges presented by greenhouse change
is left as an open question.
In further discussion here, we are going to confine the focus to the role
We return to one of the
played by scientists in validating greenhouse change.
problems
identified
in
the
framework
of
information from science to policy communities.
try
to
show
presenting
that
the
validation
scientists
are
information
not
a
in
doing
communicating
In the next chapter we will
a
good
comprehensible
further suggest that this leads to unnecessary confusion
over validation of greenhouse change.
validation
job
uniformly
manner. 4 6
in
We
in the public debate
What we would really like to know is
46
1t should be stressed that the next chapter is not an attempt to illustrate the entire framework, but
to address one of the communication questions identified in the framework (Is validation information
being presented well?) in so far as it relates to one of the validation components in the framework (the
scientific component viewing validation in terms of credibility).
8
Greenhouse change validation framework
115
improving the
validation
way science
in the policy
shift
forcing a paradigm
is presented
information
benefit of
the marginal
is
What
of uncertainties
a clarification
Would
arena?
i.e.
difference.
any
makes
whether this
to the policy
and validation evidence help
hypothetical
an extreme
In
arena?
in
case: if the evidence validating theory was persuasive for all scientists (dissent
being non-existent among scientists with standing) and the uncertainties
essentially zero (but political inertia and opposition were still present), would
this make a difference to the policy evaluation and response?
we
now
For
of the
presentation
framework
we described
evaluation
of likely
probability
of the
presenting
might make
science
non-conventional,
improved
the
validation
well
as an
times
scenario
acceptability.
but possible outcomes as
an
process
validation
mid-range
the political
times
scenario
the
by
given
In
a difference.
the outcome of the policy
impacts
which
via
avenues
suggest
only
can
By
the
science
as the mid-range
and comprehensible stories of how they come about, the fragility of
By presenting
the conventional scenarios might be appreciated in context.
validation evidence clearly and in context, and outlining sources and
scenario,
implications of real disputes in the scientific interpretation, the probability of
Less room would be
the scenarios could be gleened more realistically.
available
to
needlessly
or
diminish
carrying out debates over non-issues.
then people
only,
are likely
enhance
the
probability
estimates
by
If debate can be minimized to real issues
to be less confused,
and the probabilities
they
If the presentation of
associate with the scenarios are likely to fluctuate less.
information to
the science is more effective in supplying useable
policymakers so that the impacts and policy implications are clearer, then the
scenarios may be more acceptable.
If validation
and faithfully,
can
be
information
then the risks
appreciated
in
their
can
presented
be
and uncertainties
comprehensively,
associated with the scenarios
context.
appropriate
clearly,
This
will
be
important
information for the policy system if it is, or can be, sensitive to the relative
degree
of risk and uncertainty.
8
Greenhouse change validation framework
116
9
9.1
GREENHOUSE CHANGE VALIDATION PRESENTATIONS
Introduction
In
relating
this
chapter
we
take
up
the
question
to the validation of greenhouse
scientific
community
greenhouse validation
outlined
in
the
the
previous
mostly in terms
policy
presentations
political), though we
greenhouse
to
chapter
community.
We
with regard
to
(scientific,
utilitarian,
should in time.
theory
how
will
from the
not
all validation
consider
components
paradigmatic,
which
involves
and
testing whether
are credible.
in general
relates
to
its broader
context is complex, and has been outlined by Ravetz (1971).
manageable
information
For now we will consider validation
and scenarios
The problem of how science
well
change is being presented
of the scientific component,
change
of
in answering the validation
social
To keep this study
question we are going to concentrate
on subsets of science and policy which we see as integral to the translation of
validation
information
science journal
media
(policy
literature
input).
scientific validation
fora
for
from
science to policy.
(science
The
original
information
communication
output)
The
and
subsets chosen
interpretation
and most
sophisticated
is the science journal
beyond
the
scientific
of this
the
in
the
of sources
literature.
journals
are
of
Successive
(popular
articles,
interviews, etc.) tend to become increasingly simplistic relative to the sources
of information in the journals.
If the validation information is not presented
well in its most sophisticated form, then the validation problem will be made
successively
topic
more difficult in other fora.
of current
concern,
the
journal
literature
directly in the press upon publication.
press
change
is
influential
theory has
Policymakers
influenced
by
Much
are
in
part
the
of
the
perception
or eroded
media
is
Interpretation
public
been bolstered
public
of
shaping
With greenhouse
frequently
and
reported
a
on
of the literature in the
of
whether
by the latest
audience,
change being
the
greenhouse
research
policy
results.
process
is
perceptions.
recent
scientific
greenhouse change has concentrated
and
public
debate
on
validating
on whether or not it is possible to detect
greenhouse change signatures in the climatic record to date.
The role of the
117
the
Interpreting
is not conceptually
validation
regard.
of
greenhouse
change
In response to long term sustained
difficult.
gases, events
greenhouse
by
of the climate
forcing
this
in
debate
public
context
the
in
drought
U.S.
1988
in the
attention
received
drought
U.S.
1988
as the
such
U.S.
1988
This does not say that the 1988 U.S.
drought are likely to happen more often.
drought is conclusive evidence for greenhouse change, nor does it say that the
Yet, misinterpretation
1988 U.S. drought is unrelated to greenhouse change. 4 7
scientific
of
as
such
information
distressingly
this is
Schneider
common.
(1989) notes that the trial by media of greenhouse change and the drought was
By analysing some of the
a non-scientific issue from the very beginning.
journal articles
that were cited by the media in the 1988
debate, we hope to show here that misinterpretation
in part
is
related
of validation information
literature
the scientific
way
context for that
Scientists
results
wield
is intended
obligation
incumbent
consistent
knowledge.
the
influence
or desired.
upon
with
issues,
information
to
is
is
not being
regardless
of
inform
the public
relevant
of
presentation
by
a position,
the
of
nature
domain
such
in
Being
scientists
the
public
the
in
public
contentious
on
influence
manner
validation
change
well.
presented
relevant
greenhouse
In short,
that
implications
are being picked up in the press without providing adequate
audience.
The
is presented.
with policy
papers
is writing journal
community
scientific
the
to
drought non-issue
body
whether
such
the minimum
debate
of
in
a
scientific
We argue here that part of this obligation can be met by placing
implications
of
scientific
research
in
context
appropriate
to
a broader
audience than that of the immediate field at hand.
47
For instance, in a particular region of the U.S., droughts of a particular severity (exceeding some
If greenhouse
arbitrary measured threshold) may occur say 10 times in a hundred years on average.
warming increases the likelihood of drought occurence in the region, then severe droughts may now
occur say 15 times in a hundred years on average. We cannot however separate out the 5 additional
droughts from the 10 that would have occured anyway (in the absence of a change in climate forcing).
It is thus not possible to conclude that a particular drought is caused by greenhouse warming, or that a
particular drought is unrelated to greenhouse warming.
9
Greenhouse change validation presentations
118
9.1.1
Approach
from the
We will frame the problem of translating scientific knowledge
within
on
influences
the
scientists,
to
contributing
in
degrees
politicians,
media,
of scientific
translation
poor
and the
external
to
varying
Further,
results.
all
public
general
or
and
involved
all
are
channel
communication
their
journalists,
science
Scientists,
arenas.
different
the
barriers across
occurs due to communication
construction, poor translation
With this
of communication.
terms
in
largely
arena
public
to
scientific
interact
in
a
In this work, we do not
complex manner in the formulation of public policy.
attempt to articulate what the role of each group is, nor the extent of their
though
interaction,
interesting
are certainly
these
questions
to
influence
or
take up.
We will however, begin to address the role of scientists through the
of
particularly
towards how these results are interpreted
and
literature,
journal
the
in
results
research
their
expression
look
Most of
in the media.
the examples used here will be of journal papers that elicited direct coverage
The papers have been chosen for their relevance in clarifying
in the press.
We have not attempted to do a survey of the greenhouse
the themes presented.
change
literature.
Media
in
presentation
has
papers
considered
been
We leave open the question at this stage as to whether
mostly in the U.S.
science
of the
interpretation
and media
coverage
is significantly
of science
different
other countries.
To simplify analysis initially, we will make the assumption that science
journalists and the media are reasonably competent and faithful in their
We will not consider in any depth
attempts to interpret scientific information.
here problems
at the media end of the communication
distort translation of information.
The structure and organisation of the media
Herman and Chomsky (1988)
can lead to significant distortion of information.
outline this with respect to communication of U.S.
sources
of constraints
communication
on
of scientific
the
funnel that hinder or
and
media,
information
on scientists,
we
foreign policy.
role in
the media's
will
be
intend to
left
for later
make
focusing
here largely
problem
of communicating scientific results effectively.
a first
Potential
shaping
the
work.
By
cut on the
This is not intended
to be wholly conclusive, but to provide a basis for further consideration.
9
Greenhouse change validation presentations
119
Background
9.1.2
validation
and
Detection
of any
induced
greenhouse
climate
change
Wigley and Jones (1981) articulate this as follows: "The
presents a conundrum.
By
effects of CO 2 may not be detectable until around the turn of the century.
will probably have become
this time, atmospheric CO 2 concentration
sufficiently
high
we
(and
will
significantly
climatic change
will have
increases)
larger than any which has occurred
such a change
To avert
century could be unavoidable.
decisions
further
to
committed
be
to be made
(for example
to
reduce
that
a
in the past
it is possible
that
anthropogenic
CO 2
emissions) some time before unequivocal observational 'proof of the effects of
In light of this, establishing high confidence in
C 02 on climate is available."
detection of greenhouse change may well be irrelevant to the formulation of
That is, a point may be reached at which policy decisions do not depend
For now at least, the policy
on resolution of residual scientific uncertainty.
debate on greenhouse change has not reached this level (whether it should
policy.
have
or not),
and communication
of increments
of scientific
knowledge
on
change is relevant to the policy response.
greenhouse
The
within
of evidence
presentation
a framework
in which
response
may be required prior to conclusive resolution of the science is bound to be
Furthermore, until modelling studies of the transient response to
contentious.
greenhouse gas build up over the last one hundred years are carried out with
three-dimensional climate models including realistic representations of ocean
circulation
and mixing,
good
indicators
quantitative
of greenhouse change to date will not be available.
numbers, there is plenty of scope for disagreement
expected
from theory
has occurred or not.
of the expected
strength
Without good quantitative
over whether the change
Nevertheless,
some guidance
for
presentation of theory is still possible.
In
recent
years,
greenhouse
change
not only in the scientific community,
has
received
but by the general
increased
attention,
public, and in the
This widening of attention from within the scientific
community to without has meant that greater import is now attached to the
Two of the recent papers
presentation of scientific results in the area.
bearing on the detection or lack thereof of greenhouse effects in climatic data
political sphere as well.
9
Greenhouse change validation presentations
120
have received high profile
(Hanson et al., 1989, and Trenberth et al., 1988)48
in the U.S.
coverage
York
(New
coverage
newspaper
prominent
received
also
(1981)
Gavin
and Kukla
Earlier papers, such as those by Hansen et al. (1981)
Times, 1989c).
and
1989a, and New York
national press (New York Times,
Times, 1981a, and New York Times, 1981b) in bringing the greenhouse issue to
the
public.
That the general public
papers
on
of the papers,
interpretations
their
influence
do
presentation
of
manner
social
serious
scenarios have
that the public
and environmental
context is known to the community
known (as yet)
in
confusion
unnecessary
representative
of climatologists,
outside this community,
of
many
the
public
papers
in
This requires
The
literature
confusion for the lack of appropriate context.
While much of the
it is generally not well
papers
that
challenge
the
or essential
research
have
the
research
careful in the field.
in these
such
was created,
Note that this is not an attempt to
conclusions
papers
created
are
As such, it may be instructive to
in
these
appears little reason to conclude that it isn't satisfactory.
presenting
above
cited
go through them in turn and consider where or how confusion
with a view to avoiding this in future.
an
and the failure to provide it creates
sphere.
the
and it is vital
implications,
of the research.
of the relevant context
understanding
policy
Greenhouse climate change
issue be an informed one.
debate on the
and
perception
public
makers opinion in the area of climate change. 4 9
and
Scientific results
context for the work therein.
provide the appropriate
correctly,
on the papers to
burden
an additional
places
more
or
above,
cited
those
like
change
greenhouse
will be digesting
and policymaking community
are among
the
papers,
for there
In fact, the groups
most thorough
and
Rather, it is the presentation of the results that will be
48
0ur selection of examples is skewed toward papers interpreted to be in conflict with greenhouse
change theory, or at least interpreted as not supporting present expectations. We can speculate on why
Clark (personal
these examples may be easier to find than examples supporting theory.
communication) notes that the greenhouse change journal literature has been subject to fairly good
peer review over quite some decades now, resulting in tight and careful presentation. When the peer
review process has had sufficient time to critique theory, then presentations advancing the theory may
be fairly refined, and less subject to incautiousness.
49
For example, it is unlikely that the U.S. ban on chlorofluorocarbons as aerosol propellants would
have occured as early as 1978 without the presentations of Rowland and Molina and their peers in the
atmospheric chemistry field.
9
Greenhouse change validation presentations
121
considered; that which was presented and how, that which was not, and how it
was interpreted in the press.
greenhouse
of
detection
on
papers
journal
scientific
of
Discussion
9.2
in
implicated
or
change
With some waxing and waning of attention, scientific papers bearing on
detection of greenhouse change have received coverage in the press
of global warming,
consequences
the
carbon
increasing
of
impact
probable
consequences
in a paper entitled 'Climate
(1981)
greenhouse change theory is Hansen et al.
describing
In
dioxide'.
outlining
decade
earlier papers this
of
Illustrative
1980's.
the
throughout
potential
the
this and other papers quite rightly outlined
warming,
of global
disastrous
consequences
greenhouse
warming,
lest
be
into
factored
the hubris
we have
the
less
but
probable,
It is crucial that less probable, but
sometimes potentially catastrophic impacts.
potentially
and
to
the
defy
policy
debate
on
Law,
and
Murphy's
50
However, in doing
assume that only the most probable outcome will occur.
so, it is also important to be careful about how such possibilities are presented.
in presentation can result in an overemphasis on
'gloom and doom' scenarios to the detriment of the overall debate.
Lack of attention to detail
For instance, Hansen et al. (1981)
with warming in the vicinity of the West Antarctic ice sheet as
"Danger of rapid sea level rise is posed by the West Antarctic ice
associated
follows.
sheet,
present the possibility of sea level rise
which
disintegration
is
grounded
and
melting
below
in
level,
sea
of
case
making
general
it vulnerable
warming.
The
to
rapid
summer
If this temperature rises ~5* C,
in its vicinity is about -5*C.
deglaciation could be rapid, requiring a century or less and causing a sea level
rise of 5 to 6m. If the West Antarctic ice sheet melts on such a time scale, it will
temperature
temporarily overwhelm any sea level change due to growth or decay of landbased ice sheets. A sea level rise of 5m would flood 25 percent of Louisiana and
and many other lowlands throughout
the
Florida,
10 percent of New Jersey,
world."
If the West Antarctic ice sheet were to contribute to sea level rise, the
From Clark (1985), "it is generally accepted that among the greatest blunders of military and
political analysis is focusing on what one's adversary will probably do, to the exclusion of what he
might."
50
9
Greenhouse change validation presentations
122
Both pieces of information must
important points are how much, and how fast.
This information is in the
go together to be meaningful in impact evaluation.
The paper tells us that the sea level rise would be 5 or 6m over a time
paper.
The West Antarctic ice sheet would take of order
period of a century or less.
one hundred years to melt if conditions for melting were attained. 5 1
It is thus
quite important to distinguish the above statement from one that implies that
the sea level could rise 5 or 6m at any time within the next century.
use
occasionally
scientists
however,
Unfortunately
terms
certain
with
particular meaning in a scientific context that can lead to confusion in other
contexts
where
particular
the
meaning
not
is
example,
For
understood.
Hansen et al. (1981) use the word 'rapid' three times in the above quote to refer
On geological time scales, a change in sea
to deglaciation and sea level rise.
level
rapid.
or glaciation
The
over a period of one hundred
general
public
not generally
do
however,
incredibly
years is indeed
think on
such
time
scales, and rapid in the public context implies time scales probably a couple of
In the press, this inevitably
orders of magnitude less than one hundred years.
led to initial confusion,
and eventual decoupling of the potential change with
New York Times (1981a) interpreted
the associated rate of change in sea level.
the paper of Hansen et al. in a way that already made the association between
the amount of sea level rise and its rate of change ambiguous, as follows.
seven
scientists
atmospheric
global
a
predict
unprecedented magnitude' in the next century.
of
warming
"The
'almost
It might even be sufficient to
melt and dislodge the ice cover of West Antarctica, they say, eventually leading
to a worldwide rise of 15 to 20 feet in the sea level. In that case, they say, it
would 'flood 25 percent of Louisiana and Florida, 10 percent of New Jersey and
many other lowlands
this imply
that the
throughout the world'
flooding would
occur over the course
simply sometime within the next century?
Times
article
covering
another
change (Kukla and Gavin,
or less."
within a century
Does
of a century, or
Two months later, in a New York
paper in the journal
Science
on greenhouse
1981), the association between 'how much rise' and
'how fast' was already lost in the public arena.
New York Times (1981b) closes
51More recent estimates since Hansen et al. were writing have relaxed the time required to melt the
West Antarctic ice sheet (if it were to melt) to closer to two hundred years. Semi-apocalyptic
references to the West Antarctic ice sheet have now virtually receded from discussion, though this
took quite some time to come about.
9
Greenhouse
change validation
presentations
123
with
of
consequences
"The
paragraph
the
dioxide
carbon
rising
Some contend that the
have been a matter of intense debate.
warming would melt and dislodge the West Antarctic ice sheet,
concentrations
resulting
raising sea levels 15 to 20 feet."
Up until recently,
by
characterized
the
change has
on greenhouse
public debate
level
to possible rapid sea
similar references
rise,
been
without
Of course,
qualifying statements on the period over which it would take place.
a 5 or 6m rise over 100 years would still be disastrous, but it does not quite
engender the same gloom and doom as perceptions of an instantaneous 'rapid'
rise. We do not imply that the paper of Hansen et al. (1981) is responsible for
this shaping of the debate, but that it is illustrative of the type of presentation
we turn now to more recent papers,
illustrative not of the potential for creation of unwarranted gloom and doom,
but of the potential for the unwarranted toning down of greenhouse change
to it.
that could contribute
Similarly,
theory.
The paper by Hanson et al.
(1989)
titled "Are atmospheric
'greenhouse'
in the climatic record of the contiguous U.S. (1895-1987)?",
The paper makes an
pertains directly to the detection of greenhouse change.
implicit leap into the policy realm by writing on this issue, and a more explicit
one by choice of title, and by setting up NASA climatologist James Hansen as a
effects apparent
Hansen's testimony to U.S. Congress is quoted
'straw-man' in the introduction.
in the introduction to the paper as follows: "the global warming is now
sufficiently large that we can ascribe with a high degree of confidence a
cause and effect relationship to the greenhouse effect..."
with this or not, Hanson et al.'s paper alone (or
references)
is not of sufficient
climatologists
well
know
(but
breadth
not
to
refute
necessarily
the
Whether one agrees
without
supporting
statement.
Hansen's
and
press),
as
As
Hansen
pointed out in response to the paper in the press, the contiguous United States
cover only 1.5% of the surface area of the earth, and we would not expect such
Only one extra
a small area to be even close to representative of the globe.
By quoting Hansen and
sentence would have been required to point this out.
then
failing
unnecessary
place
the
paper's
confusion
was
created
to
research
in
findings
the public
in
the
sphere
Reuters coverage of the paper [Boston Globe, 25 Jan 1989]).
9
proper
context,
for
example
(see
Reuters were so
Greenhouse change validation presentations
124
unaware of the relevance of the paper to the detection of greenhouse change
that they printed the following: "Their findings contradict suggestions by
other scientists that the global
climate is gradually warming
rainfall is
and
New York Times (1989a)
declining because of the so-called greenhouse effect."
did not confuse the meaning of the paper in the same way, but it was clear that
they had spoken in some detail with one of the authors about its implications.
As Reuters showed on this occassion, not all reporting agencies will trouble
to do that,
themselves
more pervasive
al.
and Reuters
than that of Geophysical
is probably
public opinion
influence on
Hanson et
Research Letters (where
was published).
A further source of distress in Hanson et al. is that the paper makes no
reference back to the areal distribution of temperature changes in those
works whose interpretation it casts implicit aspersions on, namely Hansen and
While both these global data sets
Lebedeff (1987) and Jones et al. (1986,1987).52
show
a
global
contiguous
warming
trend
United States, Hansen
trend, and the Jones et al. (1987)
the eastern U.S.
of Hanson
the
over
last
and Lebedeff
not representative
contradict the available
years;
for
the
show an insignificant
(1987)
data for the past 40 years shows cooling for
and warming for the western U.S.
et al.
hundred
one
of
53
Thus, not only is the data
global trends,
but it
doesn't even
trends in the global data
indications of regional U.S.
This is important contextual information
sets that do indicate global warming.
that should be included in the paper.
similar
Along
lines,
the
GFDL
model
predictions
for
precipitation
changes over the United States are referenced in Hanson et al., and then the
conclusion of no significant precipitation trend in the observations is
presented.
This represents
a failure
to place the theory
resulting in the creation of a straw-man
will).
in proper context,
(or 'straw-theory'
of theory
if you
The reader is not told, but deserves to know, that regional precipitation
forecasts from
current
models are highly provisional
and uncertain,
and that
52
1n this case, even from a purely scientific point of view, one has an obligation to note recently
published data covering the same region.
53
1n addition, but less representative, Angell (1989) has analysed global temperatures over the last
thirty years. At the surface, Angell's data shows a marginally significant warming trend for the globe,
but no trend for the north temperate region (within which the U.S. lies).
9
Greenhouse
change validation presentations
125
there
are
substantial
and
(Schlesinger
differences
Mitchell,
for the
precipitation trends
between
predictions
Quoting
1987).54
U.S.
from
of
predictions
models
regional
It is important to
is fine.
for comparison
different
recognize however, that the degree of precision offered in the paper goes well
the
beyond
instance,
present
a
offered
The
consensus.
scenario
consensus
rainfall in the mid-latitudes" (Jaeger,
doubling
Bellagio
and
"perhaps,
for
workshops
a decrease
in
summer
Note further that that was for a
1988).
a situation we
at equilibrium;
and which is not expected until about the middle of next
Without some form of caveat on the comparison of one model group's
for doubled carbon dioxide levels with pre
detailed predictions
data,
of
of carbon dioxide in the atmosphere
are yet to attain,
century.
Villach
a much
stronger
theory
of greenhouse
refutation
1988 climatic
is presented
to
the
policy level discussion than is justified.
data should treat the expected
is systematically
deficient
in the
greenhouse change from theory with the same
The paper by Hanson et
sophistication that is afforded the analysis of the data.
al.
is apparent
with whether greenhouse change
A paper concerned
Both the
in this respect.
implicit comparison
the explicit comparison with the GFDL precipitation
with Hansen's data, and
predictions presented the theory in an overly simplistic manner, and failed to
provide context for the work.
paper
the
all have
affect
Moreover, the systematic
greenhouse
of impugning
deficiencies
change
in the
It
theory.
is
healthy to be critical of any theory, but critical debate is not strengthened by
In short, the approach to the policy implications of the
setting up straw-men.
research in the Hanson et al. paper is misleading.
related events
The
al.
et
Trenberth
that
development
(1988)
were
of the
discussed in relation
presented
associated
Midwest
to tropical
a plausible
with the
United
1988
States
mechanism
North American
drought
sea surface temperature
displacement of the jet stream over North America.
of causally
in
(SST)
drought.
particular
patterns,
is
and
In setting this work in
context, the authors point out that "the greenhouse effect may tilt the balance
54
Though speculative, there may be simple physical mechanisms that support the GFDL model
If such mechanisms are relied upon here, then
predictions for a decrease in precipitation in the U.S.
they should be mentioned, and any reliance upon them made explicit. Note also, that even for different
runs with the same model, there may be significant differences in precipitation distribution.
9
Greenhouse change validation presentations
126
such that conditions for droughts and heat waves are more likely, but it cannot
Though seemingly adequate in setting
be blamed for an individual drought."
On the one hand, an individual drought does not
the picture as we shall see.
fact
the
other hand,
that
evidence
conclusive
represent
we
that
explain
can
is
change
greenhouse
any
with
us.
on
the
drought
On the
basis
of
atmosphere and ocean conditions means that, on the basis of
antecedent land,
this evidence, we cannot dismiss the 1988 U.S.
as
only half
this provided
to greenhouse theory,
for the study relative
context
drought and greenhouse change
unrelated.
reference
greenhouse
The
in Trenberth
served
al.
et
attract
to
only
New York Times (1989c)
people's attention, and then confusion reigned again.
gave the paper an emphasis that belied an inability to interpret
Their 'Science Times' section ran
the paper's relevance to greenhouse theory.
with the heading "Scientists link '88 drought to natural cycle in tropical
for instance,
"Greenhouse
subheading
Pacific,"
researchers
agree,"
and
was
effect
introductory
paragraph
the
not
"Last year's
time,
this
culprit
killing
drought
in the United States was caused by massive, naturally occurring climatic forces
in the tropical Pacific Ocean and had little to do with global warming caused by
the greenhouse effect,
according
The Scientific American
to new evidence."
1989) presented Trenberth et al. similarly to the Times,
(Burnham,
role in greenhouse change detection to the paper.
an unwarranted
Though
the
paper
et
of Trenberth
al.
actually
doesn't
wrong, it was set up to be misinterpreted in the above manner.
Trenberth et al.'s explanation
drought
is
Trenberth
unrelated
et al.
to
of the 1988 U.S.
change
greenhouse
offered an
phenomena, but didn't say
that
occurs
in
This leap
occurs
actually make it themselves.
two
First,
steps.
in terms of natural
and greenhouse change
on this basis.
step follows whereby
drought and greer-house change
on interpretation
The leap from
and that one
failing to provide the latter information, the second
the leap is made that the U.S.
anything
always hold
such explanations
drought
say
drought to the notion that the
explanation for the drought
cannot separate the 1988 U.S.
By
attributing
by others,
while Trenberth
In effect, Trenberth
are unrelated.
et
al.
never
et al.'s presentation leads
science journalists and other interpreters to the edge of a cliff, and then the
interpreters do the work of throwing themselves and their readers off.
9
Greenhouse change validation presentations
127
Palmer and Brankovic
et al.'s paper in Science,
Following Trenberth
(1989) published a similar paper in Nature titled 'The 1988 US drought linked to
The style of Palmer and Brankovic is
anomalous sea surface temperature.'
The 'greenhouse effect' is mentioned in the first
identical to Trenberth et al.
line of the paper, but no subsequent attempt is made to place the research in
Quite predictably, the
the paper in full context relative to greenhouse change.
on Palmer and Brankovic
1989) commented
1988
greenhouse
on
drought
concluded with the following.
U.S.
Strategic
probably
arose
the
Trenberth
and Palmer
(1989),
Brankovic
and
whatever
however,
thing,
normal
of
consequence
"Don't blame
piece
short
AMS
The
says."
(AMS,
society
under the heading
a simpler and more
Namias
effect."
greenhouse
"One
from
Meteorological
writing
absolutely
is
with the greenhouse
in
explanation
Nature
on
the
the nonsense
Namias closes with the
the
clear:
variability,
atmospheric
logical
articulated
papers
leap made from these papers more clearly than most.
sentence
a further boost
"Their study supports the conclusions of many
causes,
natural
from
et al.
study
effect,
U.S.
expressed in such forums as the National Climate Program
Planning Seminar held last November, that the drought
climatologists,
Office's
than
The American
and Brankovic's paper.
Palmer
received
are unrelated
change
and greenhouse
drought
1988
the
that
et al.
of Trenberth
interpretation
on
propagated
notion
and
has
drought
no
was
a
connection
effect."
To stem the tide of confusion, it was left to James Hansen (In These
Times, 1989; New York Times, 1989b) to place the paper of Trenberth et al. more
directly in context thus: "Every drought can be related to concurrent and
antecedent climatic factors such as the position of the jet stream, soil moisture,
That will be true even if the
snow cover, and ocean temperature patterns.
greenhouse
effect
greatly
In a strictly scientific
Brankovic
since
are
the frequency
severity
and
of drought."
sense, the papers of Trenberth et al. and Palmer and
virtually
greenhouse
increases
change
manner of Trenberth et al.
irrelevant
or not,
the
to
one could
detection
always
of
greenhouse
explain
droughts
change,
in the
It is possible that the reference in Trenberth et al.
to greenhouse change was an attempt to lay to rest the public belief that the
1988 U.S. drought represented unequivocal evidence for the arrival of
greenhouse change.
While that belief is indeed an unfortunate
9
distortion of
Greenhouse change validation presentations
128
the public debate over greenhouse change, so too is the misinterpretation of
Trenberth et al. that the 1988 U.S. drought and global climate change are
If global climate is changing in response to long term changes in
greenhouse gas forcing, then this will be manifest in shifting statistical
outcomes by changes in the frequency or intensity of some events in some
unrelated.
places.
One
Simply put, drought will be a more likely occurrence in some places.
cannot rule out the plausibility of relationships between greenhouse
change
climate
1988
the
(even
occurrences
drought
particular
and
U.S.
drought) on the basis of Trenberth et al. and Palmer and Brankovic.
on the research of Trenberth
A more direct statement
implications
American,
Scientific
science
for
confusion
could
deserve
to
choose
ignore
propagated.
the
York
New
the
that
and Nature interpretations
AMS,
journalist
information, but they
the
and its
if contained in the
such as those above,
theory
prevented
have
might
paper,
a
for greenhouse
et al.
55
Times,
Of course,
contextual
relevant
at least the information to make that decision
themselves.
Knowing
what
level
of context
reach a broad audience is difficult.
Hanson
et
al.
papers,
the
in
is required
scientific
papers that
In the aforementioned Trenberth et al. and
confusion
the
papers
created
was
probably
foreseeable with some thought, so the simple answer is to provide more context
In
to avoid the common or likely sources of confusion or misinterpretation.
many cases however, it will be unclear as to whether the press will view the
paper
as
information.
relevant,
such
that
needs
one
to
include
much
contextual
Even then, it might still be unclear as to where confusion might
stem, and what would be required to address it.
1n placing the research more directly in context in papers, there is no reason why one shouldn't use
in addition alternative methaphors (Schneider, 1988) that are more easily understood by a broader
audience. In Trenberth et al. for instance, one might liken the relationship between the drought, the
greenhouse problem, and jet streams to that between nuclear reactor accidents, lack of adequate safety
The lack of adequate safety standards in nuclear
standards, and cracks in the containment vessel.
of accidents, however, any given accident
occurence
to
the
plants leads to conditions more conducive
Similarly,
can always be explained in terms of cracks in the containment vessel and such like.
greenhouse change might lead to conditions more conducive to the occurence of droughts in some areas,
however any given drought can always be explained in terms of jet streams, SSTs, and such like.
55
9
Greenhouse
change validation presentations
129
This
Angell (1989) is probably an example of a paper in this grey area.
"Variations
paper,
temperatures,
and
trends
in
1958-87," has relevance
the concluding
tropospheric
and
stratospheric
global
to detection of greenhouse change.
In
section of the paper, the evidence for and against greenhouse
change as analysed in the paper is laid out.
The author then concludes that
"In view of the pros and cons listed above, I believe it is premature to state
The
categorically that a greenhouse effect is already being observed."
question presents
further context
itself as to whether this
outlining the limitations
thirty year data set.
statement and the paper required
of what one can
conclude
from
a
Predicted warming due to greenhouse gas increases will
appear as a climatic variation on approximately this time scale and longer, and
one probably doesn't have sufficient information on natural fluctuations or
out greenhouse
effects in a thoroughly
convincing
climatic
noise to separate
manner.
Kelly et al. (1982) note that "variables for which only 20 or 30 years
of data exist are of little value on their own as candidates for detection of CO 2
Thus, on the basis of a thirty year data set, it is unlikely that one
could ever make the categorical statement that the author was not prepared to
Rather, the paper could provide supporting evidence, albeit fairly
make.
effects."
strong, and should be looked at in conjunction with other longer period data
The big picture might suggest conclusions not implied by the smaller
sets.
one.
These sorts of considerations are self-evident to a climatologist, but those
outside the field need more information to begin thinking in this manner.
This attention to Angell (1989)
isolated statement in the paper.
may seem to give unfair emphasis to an
However, it would be naive to assume that
such statements directly addressing the greenhouse issue would not be picked
up in the press and perhaps subject to overly simplistic interpretations other
At the level of the scientific literature it is far better to
err on the side of sophistication and too much context, than to oversimplify too
If this trend continues, it will
early, as with the present trend in papers.
become increasingly difficult to conduct an informed policy debate.
than those intended.
9
Greenhouse change validation presentations
130
of
Characterization
9.3
coverage
media
the
for
about the
we will say a few things
articles,
the writing of science journal
above examples
of the
proceeding to consider implications
Before
media coverage in general, since it also bears on this.
in
of
aware
common
several
with
characteristic
That characteristic
scrutiny.
issues
on
papers
journal
writing
should
scientists
that
manifestations
be
public
current
undergoing
of a paper
to present coverage
is the tendency
salient
particular
a
(and
here
reviewed
papers
display
media
the
topics),
other
on
elsewhere
journal
science
the
of
coverage
In
attaching greater import to the story contained in the paper than is probably
The media seem to want to validate theory by applying what Lakatos
justified.
(1970)
calls 'instant
that can
experiments'
as 'crucial
The media frequently interpret
rationality'.
refute (or confirm)
a theory
results
new
instantly.
In
Lakatos' view of science "there are no such things as crucial experiments, at
least not if these are meant to be experiments which can instantly overthrow a
research programme."
For those papers that might be classified
as in some
way at odds with conventional theory, the story of the conflict with the theory
Thus Hanson et al. and Trenberth et al. are
tends to be overemphasized.
as
presented
the
also
media
to
tend
with the theory.
consistent
In doing so,
that
are
creates
the
observations
individual
overemphasize
Similarly,
theory.
change
for greenhouse
setbacks
significant
the media sometimes
This occurs as a
impression that the single observation proves the theory.
media response to observations such as a hot summer locally, or record global
warmth in one year.
of
channels
communication
literature. 5 6
That is,
other
through
than
directly
weather service
from
conditions.
the
summaries,
or even just plain every day observation
scientists,
science
In these cases, the misinterpretation usually stems from
science
journal
interviews
and experience
with
of local
We have been unable to find good examples of direct coverage of
journal
literature
whose
presentation
resulted
in
misinterpretation
such that a single observation was interpreted as proving the theory.
56
For instance, Schneider (1989) notes that no atmospheric scientist he is aware of has ever made a
statement attributing the 1988 U.S. drought to greenhouse change.
9
Greenhouse change validation presentations
131
For those papers that might be classified as supportive of theory, the
of the potential impacts or consequences of theory tends to be over
story
Thus, greenhouse
emphasized.
drugs
new
miracle
with
change is associated with rapid sea level rise,
superconductivity
cures,
the
with
promise
revolution, cold fusion with an end to greenhouse problems,
transport
of
a
and so
on.
Since our concern here is on the science journal literature, we will be
content to simply note the above for now without further exploration.
of the media and
Furthermore, we will not consider more subtle characteristics
beyond the few salient features covered in
their effects on science coverage
this section, since that is beyond the scope.
General
9.4.1
In
on
issues
guidelines
determining
relevant
questions for
standards
presentation
Science
9.4
to
scientific
whether the issue
guidelines
the
for
formulation
authors
to
being presented
presentation
the
of
consider.
public
The
has significant
of scientific
we
policy,
pose
author must first
public
policy
research
several
decide
implications
or not, and if so, whether the policy debate is still sensitive to the presentation
If both these criteria are satisfied, then
of relevant scientific material.
additional care in presentation is warranted.
We have argued that the issue of
satisfies both these criteria, but this is also true of many
other issues. On some issues, the second criterion is not met, and the standards
advocated here for communication with the public sector are not as important.
greenhouse change
One might argue for example that studies of the health effects of leaded petrol
have public policy implications, but that the public policy with respect to
leaded petrol is essentially
in place
now (in the U.S.),
and unlikely to be
changed by anything but radical new results.
The assessment of whether the above criteria are met or not may not
In such cases it probably makes sense to err on the side of
always be clear.
supplying additional appropriate context for communication with a broader
9
Greenhouse change validation presentations
132
audience,
since the cost of doing so is minimal,
and the insurance
gained
against possible creation of confusion in the public sphere is worth having.
In considering
a particular
whether
piece
of
research
is relevant
to
public policy issues or not, it is important to keep in mind whether it might be
In such cases, one should be clear in
construed as relevant, even if it is not.
presentation that the material is not relevant to the issue, particularly
if the
issue is mentioned at all in the paper.
Once the decision to provide context appropriate to a broader audience is
made, one must decide just how much additional context is necessary.
A loose guideline is that the competent reader
earlier.
point was broached
This
from outside the field at hand should understand the intended implications of
the research (no more, and no less) from a single careful reading of the paper.
This guideline
is set in this way to attempt to describe the position of the
majority of journalists
or science journalists
who
might interpret
the paper
This does not mean that such readers should understand all the
for the public.
details of the research; only that they be clear about its implications.
The decision as to how much context to give is also moderated by the
choice of journal
best,
of publication.
are probably
The simplest guidelines
and these would be that the more widely read
context for a broader audience should be supplied.
the journal,
should
given
be
to
context
in
the more
In particular, the journals
Science, and Nature are closely scrutinized by the science press,
attention
the
presentation
and so more
of articles
appearing
therein.
In reference to the characteristic of media coverage to generate a story,
scientists
should
consider
their
whether
supportive of theory or at odds with theory.
paper
will
be
considered
as
This is a somewhat artificial
construction, since most papers usually contain results that work both ways to
some degree.
What is meant here then, is to consider that information in the
paper which might be spotlighted by the media, and how it would be construed
in relation to theory.
9
Greenhouse change validation presentations
133
If the paper is supportive of theory, then the potential for the media to
theory
describing
given
be
should
attention
or
the impacts
overemphasize
should
of the theory
consequences
context
supplying
to
be
also
in
this
of the tendency
aware
exists,
and more
area.
Scientists
of the
media
in
To
subsequent coverage to present individual observations as proof of theory.
it might be useful to devote further
help the media overcome this tendency,
discussion in the paper to clearly outlining the role that observations play in
theory.
confirming
If the paper might be viewed as at odds with conventional theory, then
for the media
the potential
and more
information relevant to the real or apparent conflict is
attention to contextual
Papers should be quite clear as to whether a conflict exists or not
warranted.
between
exists,
conflict
the
to overemphasize
conventional
If a
and the observations or theory presented.
theory
conflict does exist, then it would be instructive to outline the source of the
While it
conflict and the implications of the conflict for conventional theory.
might be viewed that this is standard practice anyway, it has not been done in
the papers of this nature discussed here.
for presentation of greenhouse
More specific guidelines
theory
change
will be offered in the following section.
Scientific
9.4.2
standards
argument to
The
pre
presentation
post
and
is that the failure to
this section
be developed in
provide adequate context in presenting science validation information leads to
evaluation
inappropriate
of
evaluation
theories
requires
an understanding
program
in
expected
to
which
the
understand
and
theory
following
observations
of the
is
the
in
information
of the
public
arena.
Critical
the
scientific
method
broader context of the
embedded.
of
intricacies
While
research
research
the
public
programs
or
topic and
cannot
the
be
full
implications of every new result, they are not being helped at all by scientists
We suggest that scientists think about
who don't try to provide basic context.
applying
critical
evaluation
standards
to
validation
material
information they present has been interpreted by a competent
think about implications
of the scientific
9
method
after
lay person.
both before presenting
the
i.e.
the
Greenhouse change validation presentations
134
material and also after it has been interpreted.
resolve
method
alone isn't enough.
of information
interpretation
that
points out
The
much
easier
adherence
to
evaluation
policy
make
communication)
(personal
Clark
and
uncertainties
scientific method
validating
to
relevant
guidelines.
critical
scientific
is not enough
since the
greenhouse
can
incidences
reduce
(Schneider,
U.S.
1989)
Nevertheless, the
over non-scientific
issues
information
greenhouse
of the
media
the
by
"trial
validation
effect"
involving the
such as that
1988
drought.
Trenberth et al. (1989)
or
such as those of Hanson et al.
of recent papers
Authors
direct
of
and
change
It may be that thinking about
in presenting
context
a broader
in
evaluation
clearer.
or
the
placing it in context is not an objective value-free exercise.
scientific method provides useful
analysis will
and policy
method in science
of the scientific
following
good
We do not wish to imply that a
implicit
and
have jumped into the fray of the greenhouse issue with
comparisons
between
such
however,
observations.
Unfortunately
theory directly
and systematically.
greenhouse
papers
They have
change
have
addressed
not
confronted
and
the
attention to the
At this point, people are taking notice,
paper in the context of that subject.
are incurred.
theory
greenhouse change
theory in the title or text of the paper, deliberately calling
and obligations
(1989)
Appropriate context
is then required, which
is
Taking side shots at theory is a fair practice only when context is
Using Hanson et al. (1989) as an example, suppose that Hansen and
not given.
given.
Lebedeff (1987)
did
show a significant temperature
Hanson et al. (1989) did not.
trend for the U.S.,
and
It would then be quite appropriate to give context,
and then impugn Hansen et al.'s global data set on the basis of disagreement
Though one might disagree with the
with Hanson et al. over the U.S.
conclusion, it would be a fair attack, and one would not be calling Hanson et al.
to task for their handling or lack thereof of contextual issues.
The issue of confronting theory appropriately
it is worthwhile to outline it in more detail.
papers reviewed here will be presented
confronting
is important enough that
A summary of examples from the
with regard to scientific standards
for
theory.
9
Greenhouse change validation presentations
135
The scientific method is based in part on the presentation of theories,
and critical evaluation of the theories, particularly in light of observations or
Greenhouse change theory provides certain predictions
alternative theories.
might be altered over time.
about the manner in which the climatic system
Critical
evaluation
response
predicted
this
theory
theory
can
of
by
consistent with the theory.
and explained
detected
be
of
determination
requires
whether
in
the
a manner
For the process of critical examination of theory
via observations to be valid and successful, it is necessary to compare theory
and observations in a manner that does justice to the sophistication and nature
Though none of the papers discussed here have strictly fallen short
in this regard by the standards of the scientific method, there are additional
That is, these papers are being
relevant standards that should be applied.
of each.
interpreted
in the policy realm, but in varying degree, have failed to provide
adequate context for that realm.
As a result, when one applies the techniques
(without adequate context),
of critical evaluation of theory in that realm
the
critical comparison process fails to do justice to the theory, and is therefore
In the examples from the more recent papers presented earlier,
invalidated.
from the perspective of the policy realm, one sees observations that appear to
This can be
be inconsistent with theory, and so doubt is cast on the theory.
misleading if the conclusions are different from those in the scientific realm
(where context is known, but not necessarily given) where doubt (or support)
is not necessarily
implied.
The lack of context leaves policy makers
in the Indian proverb
blind men examining the elephant
position of the six
Each examines a different part of the elephant's
(Saxe, 1955).
anatomy, and
None has sufficient
different creatures.
believes that they are encountering
in the
information to identify the elephant.
In the process
failure to
provide
context
for
adequate
change theory, the
of greenhouse
evaluation
of critical
transference
the
from
policy realm can be related to several shortcomings or pitfalls.
observations
comparing
representative
representative
of
of
something
with
greenhouse
relevant climatic
theory;
comparing theory
theory
to
theory
accurately;
greenhouse
theory;
change
behaviour or
or observations
change
comparing
to
9
observations
or
not
greenhouse
change
to
greenhouse
change
describe
real or potential
failing to reference
in
scale to
not relevant
failing
theory;
to
These include:
established
well
not
scientific
consequences
contradictions
of
of the
Greenhouse change validation presentations
136
comparison that is
Very often the
and combinations of the above.
theory;
made in the policy realm in one of the above categories will not be the one
that should have been made, nor might it be the one that was intended or
The successful science journal
implied by the authors of the scientific paper.
be the one that provides
context appropriate to
prevent this occurence by well intentioned interpreters.
which the papers reviewed here contain shortcomings.
This is the sense in
paper in this regard would
reviewed
implicit
or
Explicit
presentation
generated comparisons
here
at
of
the
in
material
recent
papers
of the
the policy
level
illustrative
not
well
established
above pitfall categories as follows:
--
Comparing
-
of greenhouse
representative
with
observations
change
something
theory:
or
precipitation
of U.S.
comparison
in
data with particular GFDL model predictions (Hanson et al., 1989).
behaviour
scale
or
with
data
temperature
to
greenhouse
change
data
climatic
relevant
(Hanson
al.,
et
U.S.
of
comparison
theory:
temperature
global
of
representative
not
observations
Comparing
---
1989);
comparison of U.S. precipitation data for the past one hundred years with
model predictions of precipitation changes for a doubled CO 2 climate (Hanson
et al., 1989); comparison of a short period temperature data set (thirty years)
with greenhouse change temperature predictions (Angell, 1989).
--
Comparing
-
theory
theory
to greenhouse
or
change
observations
not
relevant
comparison
theory:
of
to
greenhouse
a physical
drought to greenhouse theory (Trenberth et al.
for the 1988 U.S.
change
mechanism
1988, Palmer
and Brankovic 1989).
Other examples of pitfalls that fall into these categories exist elsewhere
Note that the ratio of those in the popular press to those in
in the literature.
It takes only one journal article lacking in
the scientific journals is high.
appropriate
context
popular press.
is
repeated
information
to
generate
a
plethora
of
misleading
articles
in
the
Early misinterpretation of a journal article by one news source
over and over in coverage by later sources who base their
on
Frequently,
the
misinterpretation.
the pitfalls
in the literature
relate to
implicit
comparison
between theory and observations using inappropriate time or space scales.
9
i.e.
Greenhouse change validation presentations
137
too
small an area,
appropriate,
but
or too
the
short a time series.
variable
selected
for
Sometimes
comparison
on
the scales
those
are
scales
compared with the greenhouse signal in a different variable from theory.
comparing
lagged)
land
SST
with
and
data
isolation
greenhouse
sea
greenhouse
in
theory
combined.
theory
is
(where
the
estimates
of global
Comparison
another
common
greenhouse
with
trap.
temperature
theory
This
e.g.
time
is
change
for
unrepresentative
of
relates
failure to distinguish between that which is reasonably
theory, and that which is not.
response
is
principally
to
a
well grounded in the
The global warming of surface air temperature
is well grounded as a piece of theory; the exact time scale over which it might
happen is not.
That the warming at the surface should be greater (in the long
run) in high latitudes
than low latitudes
has near consensus
precipitation changes over the central U.S.
Examples
of failure to reference
in theory; that
will increase or decrease does not.
real or potential contradictions of the
theory have not been described here, but it is easy enough to imagine their
effect
on the public debate.
They would have the effect
theory in a more confident light than was justified.
that paleoclimatic
climates
data present
and predictions
a reasonable
for a greenhouse
between them should be checked.
inconsistencies
between
reconstructions
For example, to the extent
analogue
change
Hansen et al.
of
of presenting the
between
warmed earth,
(1981)
regional
that the climate
wetter than greenhouse
inconsistency,
over some
climate
Much of the theory
predictions
of
doubled
another,
or with
what
patterns
the
scenarios.
It
land areas in past warmer periods
was
physical
descriptions,
Until
would
further exploration.
change
stems from climate
model
Where model results agree with one
expect
of
This may not be an
that
from
simpler
part
of
models
theory
is
and
simple
strengthened.
as theory is also increased when consistent with
particular
grounding or consensus,
and
on greenhouse
grounding
Confidence in the predictions
observations.
referencing
CO 2 scenarios.
one
consistency
in
change scenarios would indicate.
but deserves
warmer
have noted potential
altithermal period with model predictions of greenhouse change
appears
past
climate
model
predictions
support or refutation of theory
attain
reasonable
based on comparison
of observations with these predictions can not readily be assumed, and context
pertinent to this should be included in presentations of results.
9
Greenhouse change validation presentations
138
The real climatic system is likely to contain
is not included in the climate models,
Surprises
will
important
to
contradict
well
occur. 5 7
significant behaviour that
nor represented
by their predictions.
In testing theory with observation, it will be vitally
and note whether the surprises or inconsistencies
consider
more
those
or
theory,
grounded
speculative
aspects
of the
Failure to do this will
predictions that relate to shortcomings in the models.
Establishing confidence in detection or
result in a debate over 'straw-theory'.
refutation of greenhouse
temperatures,
(surface
measurements,
upper
and
precipitation
on a broad
relies
change
coverage
temperatures,
air
hydrological
of observations
budget
radiation
circulation
measurements,
changes, sea level, ocean mixing, ice and snow cover, cloud distribution, etc.),
Establishing
and their comparison with relevant parts of the theory.
also requires better understanding
confidence
of the theory.
of
changes
of the more speculative
aspects
In particular, the role of cloud feedbacks and the modelling of
cloud distribution and type, the role of oceans and ocean
and the internal behaviour of the climatic system and the way it
The synthesis of a mix of
with external environmental changes.
circulation,
interacts
observations
are
studies
with
likely
an evolving
to
pertain
is
theory
to
only
an enormous
a small
task,
of this.
part
and
individual
To
prevent
misinterpretation of a subset of the required information in the public sector,
it is important that studies be placed in context relative to the overall detection
The implications of research that is likely to be construed as relevant
to greenhouse change should be stated directly and clearly.
schema.
9.5
The
science-media
interface
The short set of examples presented here is illustrative of the type and
degree
to
which
scientific
misinterpreted in the press.
journal
papers
on
greenhouse
How is this to be explained?
change
are
One could blame the
From that point of view then, there was no confusion created by
cannot resist
It could be that newspapers
the scientific papers.
sensationalizing issues and then downplaying them, and frame everything as
media solely.
Surprises in the climatic response to greenhouse forcing will include amplifying and damping
actions, and there is little reason to presume that either action is more likely to predominate than the
other.
57
9
Greenhouse change validation presentations
139
a story, which gets played out as we have seen (overemphasis on conflict or
Perhaps they are constrained from acheiving faithful
consequences).
On the other hand, one
structural factors or other means.
interpretation by
Perhaps scientists are not able to communicate
could blame scientists solely.
or
disciplinary
bounds.
to
write
of which the
truth
occasionally
are
scientists
Perhaps
traditional
transcends
audience
their
that
aware
not
are
effectively,
content
papers that could easily be misinterpreted for whatever reason.
probably
as
doing
not
are
extremes,
represent
If this is so, then
in between most of the time.
lies somewhere
scientists
blame
allocating
cases
These
well
as
in
could
they
We have suggested
information, and have some role to play in addressing this.
that
reduce
help
might
explanation
measures
simple
relatively
like
additional
supplying
incidences
of
validation
presenting
and
context
dispute
and
misinterpretation
This is all very well except that we have not considered how
over non-issues.
For
scientists might be motivated to change or supplement current practices.
instance, what is motivating researchers to present results as they do now?
Where could change be motivated - through scientists, or the editors of science
Would structural change in
journals, or perhaps through science journalists?
These questions
science and policy arenas be necessary to motivate change?
are beyond the scope of this work, though the inability to answer them means
that
question
Analysis
of whether
of
incorporating
the
1988
validation
U.S.
information
or
validate
to
attempts
used as examples.
of
validation
of simply
goal
beyond our original
here go
we cannot
is
being
presented
illuminate
drought and U.S.
addressing
well
or
greenhouse
temperature
trends
the
not.
change
have
been
They show that at least in terms of the scientific component
(establishing
credibility),
information
validation
is
not
being
presented as well as it could be.
9.6
Conclusions
We began with an assumption that scientific journal articles relevant to
validating greenhouse
change have an effect in shaping the public debate on
Scientific journal papers represent the initial point of
greenhouse validation.
communication between scientists and the press.
of scientific
papers
in
such cases,
As a result of the influence
scientists have
9
an obligation
to present
Greenhouse change validation presentations
140
their results in such a way that the results have a reasonable chance of being
interpreted
context
faithfully in the public
describing
the research
domain.
This requires
and its implications
the supplying of
that is appropriate
broader public audience as well as to specialists in the field.
the
scientific journal
for
presentation
sophistication
articles represent probably the
of
in
scientific
this
results,
forum
and
leaves
to a
It was noted that
most sophisticated
that
loss
greenhouse
of
forum
context
science-policy
and
related
discussions without complete sources of information.
Recent
published
scientific
papers
relevant
to
greenhouse
have not supplied context appropriate to a broader audience.
have
reported
ramifications,
on
scientific
considerable
confusion
aim
yielding
communication
of the
of an
informed
papers
and
with
When the press
greenhouse
misinterpretation
has
policy
taken
place
This is detrimental to the public discourse, since a
in the course of translation.
major
journal
validation
public
of
debate.
should
information
scientific
If we
assume
that the
be
the
press
is
reasonably competent and faithful in its attempt to translate scientific
papers,
then some blame must be placed upon writers of the scientific literature
to
failing
supply
appropriate
context.
failure
The
to
supply
for
appropriate
context can lead to the unwarranted support or erosion of theory relevant to
the scientific
and public greenhouse
The translation
of scientific
debates.
results
from
the scientific
to the
public
domain depends upon the context offered.
When we
apply the standards of
comparison of theory
science
the
which
itself
has
inappropriate
been
so
observations
translated
comparisons
misrepresented
information)
and
badly
occur
in
in
into
when
the
(by
is presented
that straw theory
public
context
translation
to
scientific research
domain,
is lacking.
failing
we
find
Theory
to
in the public
that
may
be
include
pertinent
arena.
Knocking
down straw theory then has the effect in public of damning the whole canon
of the theory, when this may not be implied in the scientific domain (where
context
yielding
based
at
is known)
unwarranted
on
consequences
theory
in
of theory
all.
The
bolstering
the
converse
of
public
are allowed
also
theory.
arena
applies
Unwarranted
is
created
to misinterpretation
support
indirectly
to take on unnecessary
9
shades
for
action
when
the
of gloom
Greenhouse change validation presentations
141
In either case, the result is an unfortunate skewing of the public
and doom.
debate on which the formulation of public policy depends.
avoiding
for
guideline
simplest
The
unnecessary
of
misinterpretation
scientific journal papers is to write them so that the competent reader outside
the specialty field has a reasonable chance of interpreting clearly that which
Scientific authors should think
is implied by the paper, and that which is not.
critically
of how
evaluated
after
further
as
manifest
on
authors
a
reader
has
are
reached
in
some information
have
a comparison,
such
theory
of
is
or
frequently
conflict
with
direct
their
to
where
on
If
place.
taken
misinterpretation
consequences
be
will
and observations
theory
Since
required.
is
overemphasis
an
scientific
theory,
warranted
than those
context
such
by
interpretation
other
conclusions
then
their comparison between
attention.
An
improvement
validation
of
credibility
establishing
theory
of the
less
(such
debate over non-issues
could
theory
greenhouse
confusing
by
relevant
information
of validation
in presentation
reducing
make
to
policy
level
of
public
incidences
as the debate over the role of the
1988 U.S.
It is likely that there is also significant potential for improvement
greenhouse science presentation with respect to the utilitarian and
drought).
in
paradigmatic
validation
presented,
science
for
and
community
incorporating
well
as
as
the
scientific
component
For the utilitarian component more useable information can
considered here.
be
components
and
the
views
presenting
paradigmatic
and
how
component
they
are
viewpoints
minority
scientists
formed,
critically
can
work
and
and
examine
in
on
context.
For their part, the scientists appear to have scope for adapting their work and
Such adaption can not be expected
presentation to facilitate policy validation.
however without attention to the factors motivating scientists.
9
Greenhouse change validation presentations
142
SUMMARY AND CONCLUSIONS
10
greenhouse
In the thesis we set out to identify problems in validating
particular,
for
GCMs
3-D
reliance upon
the three major
climate studies of NCAR, GFDL and GISS.
in
used
GCMS
In
scenarios.
change
greenhouse
developing
for study
we chose
We focused on GCMs, noting the heavy
context.
change in a science-policy
greenhouse
We acknowledged that it is not possible
to test everything in a GCM, and noted that there are no sufficient tests of
a validation
for
energy transports
the
chose
We
change.
greenhouse
case
study of the GCMs since the energy transports are fundamental to climate.
To calculate energy transports in the GCM annual mean control climates
it was necessary to follow an indirect technique of deriving the total energy
transport from the net flux of radiation at the top of the model atmosphere.
For the GISS model only, it was possible to compare the indirectly calculated
energy
directly from the model motions.
transport with that calculated
agreement
We
good.
quite
was
that
conclude
method
indirect
the
The
is
a
satisfactory technique for deriving the total energy transport in a GCM.
In
(1985).
the
significant,
observations,
GISS
the
and
showed
GCM
and
NCAR
more
GFDL
Carissimo
et al.
were
more
transport
than
less
showed
The GCM atmospheric
transport than observations.
energy
total
models
of the total
differences
the
where
Hemisphere
Northern
the
from
and observations
energy transport for the three GCMs
profile
the meridional
between
differences
We identified
energy
energy
total
transports (which
for NCAR and GFDL models are the same as the total energy transport since
there is no ocean energy transport in these models) were similar to each
other,
were
but
much
than
larger
observations
of the
energy
atmospheric
On the basis that it is unlikely for the observations to be in error by
transport.
the large amount (50-80%) required to bring them up to the model values, we
argued
that
the
atmospheric
transports
temperature
structure
are
with
the
models
having
more
large.
too
an
We
observed
believeable
and
that
compared the models
meridional
temperature
the
GCM
meridional
structure
They were similar, which would be inconsistent
derived from Sellers (1965).
with
are
observations
a
larger
atmospheric
energy
transport
than
143
observations.
We
conclude
that
the
atmospheric
energy
transport
is
too
efficient in the GCMs.
We
then set
out to
consider implications
energy transport between the three GCMs.
that the climate
asked
the
is potentially
question:
quite
given
of the
differences
We used a 1-D EBM to demonstrate
sensitive to the energy
that
the
GCMs
have
transport.
significantly
(~4 0 C)
transports, why do they display similar climate sensitivities
C0
We
different
to doubled
We investigated by using a simplified 'North-type' 1-D EBM of the climate
2?
system,
where
transport.
it
is
easier to
trace
implications
of differences
in
with output from the GCMs.
data.
energy
To set up 1-D EBMs that would be in some sense equivalent to the 3-D
GCMs (hopefully capturing the essential physics) we parameterized
Because
parameterized
EBMs
in total
the
each
1-D EBMs
We also parameterized the EBM with observational
climate
is
hemisphere
not
symmetric
separately,
where the parameter values
were
and
across
also
averages
the
set up
hemispheres
global
we
equivalent
of the hemispheric
values.
The major limitation found in fitting the EBMs to the GCMs and observations
was related to
the temperature
- outgoing IR parameterization.
The simple
linear fit between temperatures and outgoing IR was too simplistic and did not
capture the local minimum in outgoing IR in the tropics noted in observations
and the models.
However, this did not appear to be a major limitation of the
ability of the EBMs
parameterization
was
to perform climate sensitivity
quite
good,
and
showed
experiments.
the
The albedo
significant
differences
between albedo profiles in the GCMs, with GISS matching observations more
closely.
To test the sensitivity of the equivalent EBMs we used a 2% increase in
solar constant as a proxy for doubling CO 2 .
The results yielded essentially
similar
sensitivities
observations).
similar
to
The
the
atmosphere.
the
3-D
the
equivalent global
GCM
equivalent
GCMs
EBM sensitivities
sensitivities
obtained
for
and
in
energy
parameter
transports
values
were
between
them,
significantly
models
and
for the models
were
doubling
to
the
energy
transport.
The
fact
because
different.
sensitivity is sensitive to the value of solar constant,
sensitive
(for
CO 2 in the
This agreement cannot be because the EBMs are not sensitive to
differences
constants
between
their
EBM
solar
climate
and the EBM climate is
that
10
the
equivalent
EBM
Summary and conclusions
144
coincidence,
unlikely
compensating
Whatever
work.
at
are
mechanisms
compensating
an
either
implies
similar
are
sensitivities
that
or
factors
are
represented in the EBM, they are not as efficient as in the GCMs however. In
the GFDL equivalent EBM the polar icecaps were eliminated for a 2% solar
constant increase, which was not the case in the GCM CO 2 doubling experiment.
The energy transports obtained from the equivalent EBMs were similar
in magnitude to the values in the GCMs and observations (though a little
It is not clear why the equivalent EBM climate sensitivities were not
smaller).
candidates
for
are
the
EBM
the
in
differences
transport
energy
the
of
compensation
The main
transport.
energy
in the
differences
sensitive to
parameters B, D, xs, Ts and 8, related to water vapour and cloud feedbacks and
If compensation is occurring, and if these parameters
ice-albedo feedback.
are constrained by tuning the GCMs to the current climate, then this raises the
possibility that the climate sensitivity is constrained by tuning to the current
climate.
concluded
We
may
be
not
necessary
work.
We
do
transport
at
and
errors
that
invalidate
noted
the
energy
the GCMs
since compensating
factors
that confidence
greenhouse
change
in
to
related
inconsistencies
GCM
projections is derived from a consideration of all relevant theory, models, and
The GCMs do not stand alone, but are part of a hierarchy of
observations.
models used to test observations in developing greenhouse change projections.
step
a next
As
context we described
The
communities.
evaluating
in
greenhouse
a validation framework
purpose
of setting
incorporating
down the
and contrasting the way- science
comparing
in
change
science and policy
framework
and policy
science-policy
a
was to
systems
aid in
approach the
In science, greenhouse change theory was said to be valid
in the sense that it is embedded within the most progressive research program
(that capable of explaining the most about present, past and potential future
In policy, greenhouse change theory was said to be evaluated by
climates).
validation problem.
whether
it is
defensible
requires showing
We
noted
the
against
that the theory
lack
science policy issues.
attacks
of being
is consistent, has
of development
of adequate
arbitrarily
based.
This
credibility, and is useful.
evaluation
procedures
for
We suggested that this is perhaps one reason why the
10
Summary and conclusions
145
policy community tend to measure credibility in terms of whether there is a
consensus
supporting
linked
the
to
implications.
theory
ability
to
in
science.
demonstrate
Defensibility
that
the
in
likely
policy
impacts
was
have
also
policy
We noted that the priority assigned to greenhouse change on the
policy agenda also depends on the political acceptability of the theory and its
implications.
We argued that some of the characteristics of the greenhouse issue lead
to
a mismatch
such
as
between
science
scales,
source
time
coordination
requirements).
uncertainties
into
differences
issue
between
requirements.
In
and policy
diffusivity,
This
compounds
evaluation.
the
two
science
In
is
physical
that
necessary
(related
there
lead
to
to
factors
manifestation,
the difficulty
addition
communities
it
requirements
of incorporating
are
to
methodological
diverging
develop
and
a set
evaluation
of protective
heuristics that lend a research program a degree of resiliency in the face of
difficulties.
In policy it is necessary to raise the costs of uncertainty to allow
only well supported or acceptable issues on the agenda.
We
remained
open
to
the
possibility
that
both
science
and
policy
communities have potential to change and adjust to requirements of the other
community
in
questions
question
policy
facilitating
identified
of
how
arenas.
in
well
critical
the
evaluation
validation
validation
improve
then there may be some marginal benefit.
to
make
its decisions
uncertainties
which
the
system
of greenhouse
improved
perceptions
of risks
requires
acceptability
with
enhanced
theory.
presentation
and
high
and
we
the way
chose
they
the
to
from
present
various
study
the
science
to
information
The policy community will be able
of the inherent
We outlined
of validation
low
Of
is presented
knowledge
uncertainties.
risk
issues.
framework
information
If scientists can
of
We
potential
information
noted
uncertainty
that
to
the
act
risks
mechanisms
and
by
might
change
current
policy
when
political
is low.
To answer the question of how well validation information is presented
now
we
considered
interpretation.
commmunication
just
the
science
journal
literature
and
its
media
These two components are integral to the wider science-policy
network.
We
demonstrated
by
example
10
that
there
is
Summary and conclusions
146
considerable misinterpretation
direct
result
occasional
of
media
over
debate
As a
of science journal literature in the media.
of
coverage
issues
greenhouse change in the media.
that
science
are
papers
there
irrelevant
to
journal
essentially
is
even
validating
For the case involving the 1988 U.S.
drought
we argued that this is partly due to a failure by scientists to provide context
We
appropriate to a broader audience in presenting validation information.
note
finally
might prove
that improvements
to
be
useful
in
communication
if greenhouse
change
of validation
unfolds
within
information
the
ranges
expected, and if the policy system can become more sensitive to the relative
degree of risk and uncertainty.
10
Summary and conclusions
147
FIGURES
5.1
Zonal mean 850mb - 950mb lapse rate from Oort (1983) data
850MB - 950MB LRPSE RRTE FROM OORT
.0055
.0050
.0040
.0035
.0030
.0025
.0020
.0015
.0010
.0005
0
-. 0005
-..
coso
-50
-40
20
0
-20
LATITUDE (DEGREES)
d0
60
80
148
5.2
GISS topography on the GISS grid
GISS TOPOGRAPHY (M)
LONGITUDE
Figures
149
5.3
NCAR CCM1 R15 surface geopotential height
NCAR CCM1 R15 SURF GEOPOTENTIAL HEIGHT (M)
LONGITUDE
Figures
150
5.4
NCAR (N) and GISS (G) zonal mean topography
NCRR RND GISS ZONRL MERN TOPOGRRPHY
2800
2600
2400
2200
20001800
1600
1 0oo
1200
1000
800
q00
200
-80
-60
-40
0
20
-20
LRTITUOE (DEGREES)
40
60
80
Figures
151
5.5
NCAR (N), GFDL (P), and GISS (G) control run surface
temperatures (solid lines) and sea level temperatures
(dashed lines)
NCRR GFDL GISS SURF
SERLVL TEMPS
305
300
295
290
285
280
275
S270
U-j 265
-280O
255
250
2q5
240
235 i
230
e
-80
-60
-40
-20
0
20
LATITUOE (EGREES)
40
60
80
Figures
152
6.1
Annual mean zonal mean total meridional (northward) energy
transport (T), atmospheric energy transport (A), and ocean
energy transport (0) for Carissimo et al. (1985) observations
(dashed lines), and calculated directly from the GISS control
run model motions (solid lines)
ENERGY TRANSPORT - 0BSVNS PND GISS DIRECT
Ln
E
2
CL
0
>-
-1
LJ
LU -2
-3
-4
-5
-6
-80
-60
-40
-20
0
20
LRTITUOE [DEGREES)
40
60
80
Figures
153
6.2
Annual mean zonal mean total northward energy transport
calculated directly from the GISS model motions (D), and
indirectly from the net flux of radiation at the top of the
atmosphere
the
in
GISS
model
(I)
(control
run)
GISS TTL NWRO
EN TRAN RNL- OIR INOIR (1Co::)
I
5
/-K -
-n3
C
n0~
L-I
-
-
~2
w -
-80
-60
-40
-20
0
20
40
LATITUDE (DEGREES)
60
60
Figures
154
6.3.16.3.6
Annual mean zonal mean shortwave flux at the top of the
atmosphere (SWTP), longwave flux at the top of the atmosphere
(LWTP), shortwave flux at the surface (SWSF), longwave flux at
the surface (LWSF), sensible heat flux at the surface (SHSF),
and latent heat flux at the surface (LHSF) for the NCAR (N1,N2),
GFDL (P1,P2), and GISS (G1,G2) models. The numbers '1' and '2'
refer to equilibrium results for 1 x CO 2 and 2 x CO 2 simulations
6.3.2
6.3.1
LWTPRt&.
NC PR GI 1CZ2 2C02 FLUX
SWTPRNLNC PR GI 1C02 2C22 FLUX OATR
DATR
280
-40
-60
260
-80
-100
240
-120
-140
220
-160
Z
-180
-200
-220
200
-240
180
-260
160
-280
-300
140
-320
-340
120
-360
-80
-60
40
20
0
-40 -20
LRTITUDE (DEGREES)
60
80.
40
2
-20 0LATITUDE (DEGREES)
-80
80
50
LWSFRNM.NC PR GI 1C02 2CO2 FLUX ORTR
SWSFRN.. NC PR GI IC0Z 2CO2 FLUX OATR
es
-20
80
-40
7S
-60
70
-80
65
-100
60
Q-120
55
-
-140
-U-
x 50
160
s
U
-180
45
40
-200
35
-220
30
-240
25
-250
20
.
-80
6.3.3
-60
-40
-20
0
20
40
LATITUDE (DEGREES)
60
80
.
-80
-60
-40
20
40
0
-20
LATITUDE (DEGREES)
60
80
6.3.4
Figures
155
Annual mean zonal mean shortwave flux at the top of the
atmosphere (SWTP), longwave flux at the top of the atmosphere
(LWTP), shortwave flux at the surface (SWSF), longwave flux at
the surface (LWSF), sensible heat flux at the surface (SHSF),
and latent heat flux at the surface (LHSF) for the NCAR (N1,N2),
GFDL (P1,P2), and GISS (G1,G2) models. The numbers '1' and '2'
refer to equilibrium results for 1 x CO 2 and 2 x CO 2 simulations
6.3.16.3.6
6N3.6
63 .5
LHSF REl..NC PR GI 1C02 2C02 FLUX 0RTA
SHSF ANL NCPR GI 1C02 2C02 FLUX OATS
40
30
20
10
0
X-10
U.
-20
-30
-40
-50
-60
-80
-60
-40
40
20
0
-20
LRTITUDE (DEGREES)
60
80
-80
-60
-40
0
20 40
-20
LRTITUDE (DEGREES)
60
80
Figures
156
6.3.7
Annual mean zonal mean net radiative flux at the top of the
atmosphere from NCAR (N), GFDL (P), and GISS (G) model
control runs, and from Stephens et al. (1981) observations (0)
NET FLUX TOP ATM - NC PR GI 06
-80
-80
20 - 40
0
-40 -20
LRTITUOE (DEGREES)
60
80
Figures
157
Annual mean zonal mean total northward energy transport for
NCAR, GFDL, and GISS models (control run). The curves
labelled 'N' are for integrations starting at the north pole, the
curves labelled 'S' are for integrations starting at the south
The
pole, and the curves labelled 'C' are corrected curves.
dashed curve is from Carissimo et al. (1985) observations (0)
6.4.16.4.3
6.42
6.4.1
NWRRO
ENERGY
TRANSPORT
- RNNURL - GFOL
- NCAR
TRANSPORT - ANNURL
NARD ENERGY
5
3
0
-1
-2
N
-6
-10
-
Z -3
-
-60
-80
-40
20
40
-20
0
LATITUDE (DEGREES)
60
-6
80
-80
-60
-40 -20
0
20
40
LATITUDE (DEGREES)
60
80
NWAR EN TRANSPT-RNNL-CORRTD-NC PR GI 08
- GISS
TRANSPORT - ANNUAL
ENERGY
NWRRD
7
7
6
5
5
5
-4
u3
c
-
-
3
2
2
-1
M0
CE -1
-2
~-1
-2
S-2
-3
LU
-3
-5
-6
-7
-6
-80
6.43
6.4.4
-60
40
0
20
-40Q -20
LRTITUDE (DEGREES)
60
80
-80
-60
-40
-20
0
20
40
LATITUDE (DEGREES)
60
80
6.44
Corrected annual mean zonal mean total northward energy
transport for NCAR (N), GFDL (P), and GISS (G) models (control
run), and for Carissimo et al. (1985) observations (0)
Figures
158
6.5
Annual mean zonal mean northward atmospheric energy
transport for NCAR (N), GFDL (P), and GISS (G) models (control
run), and for Carissimo et al. (1985) observations (0)
NWARO RTM EN TRRNSPT-RNNL-NC PR GI 08
S0
L -3
LU
-5
-6
-80
-60
-40
4O
0
20
-20
LATITUOE (DEGREES)
60
80
Figures
159
Iceline latitude (xs) versus ratio of the solar constant to the
current solar constant (q/qo) for the 1-D EBM
7.1
Q/QY
1.0
RL WS QO=1365.0 IS=191.85 D=0.355
-
.9
.8
.7
.6
<
.5
.3
.2
.1-
0.86
.88
.90
.92
.94
Q/Q0
.96
.98
1.00
1.02
Figures
160
7.2
Meridional energy transport flux as a function of the sine of
latitude (x) for the 1-D EBM
RL WS 00=1365.0 IS=191.85 D=0.355
FLUX
5.5
5.0
4.5
.0
3.5
U,
X
3.0
X
X
2.5
~]
LL
2.0
1.5
1.0
I
I
I
.1
.2
.3
.4
I
.6
.5
LRTITUCE X
I
.7
.8
.9
1.0
Figures
161
Maximum energy transport flux, second component of
outgoing longwave radiation (I2), and iceline latitude (x,) as
functions of the dynamical transport efficiency (D) for the
1-D EBM
7.3.17.3.3
MRXFLUX
RL WS 00=1355.0 15=191.85
12
RL W5 00=1365.0 IS=191.85
-20
7.0
6.5
-40
6.0
5.5
-60
E 5.0
-4.05
-100
>3.5
-120
G- 3.0
2.5
1
2.0
1.5
-150--
1.0
-120
.5
0
0
.1
XS
.2
.3
.4
.5
0
.6
.7
.8
.9
1.0
.7
.8
.9 ,
1.0
-200
0
.1
.2
.3
.4
.5
0
.6
.7
.8
RL.14S00=1365.0 I~S=191. 85
1.00
.98
.96
.94
.92
.90
.88
.86
.84
.82
.80
.78
.76
.74
.72
.70
.68
.56
.1
.2
.3
.4
.5
0
.6
Figures
.9
1.0
162
7.4
Iceline latitude (xs) for NCAR (ncarnh, ncarsh, ncargl), GFDL
(gfdlnh, gfdlsh, gfdlgl), and GISS (gissnh, gisssh, gissgl)
models (control run), and from Alexander and Mobley (1976)
In each case, 'nh' is
observations (obsnnh, obsnsh, obsngl).
for the Northern Hemisphere, and 'sh' is for the Southern
The global (gl) values are averages of the
Hemisphere.
hemispheric values in each case
Xs
1.00
0.95
0.90
0.85
0.80
C
U)
CU
0
7.5
)
CCY)
C
CM
U)
W)
U)
U
U)
~CCC
C
U)
.0
0
0)
,
0)
.-
0
C
0
0
As in fig. 7.4, but for the sea level temperature at the iceline
Temperature observations are from Sellers (1965)
latitude.
C)
CO
(d
C
C
.0
C
:6
0~ 0
.0 .0 .0
.0
C
U)
:a
U)
-
C
U)
.0
.0
W)
C
UC
U)
.
)0
V)
U)
C
-
0
D
L-
U)
CO
~
~
0)C
Figures
163
7.6.17.6.6
Annual mean zonal mean planetary albedo (ALBEDO), net
flux of longwave radiation at the top of the atmosphere (I),
and sea level temperature (T) for NCAR (N), GFDL (P), and GISS
(G) models (control run), and from Stephens et al. (1981)
observations (0 for Albedo and I) and Sellers (1965)
observations (0 for T). The curves are plotted separately for
the Southern Hemisphere (SH) and Northern Hemisphere (NH)
as a function of the sine of latitude (x)
75
RL8EO0 !H
NC PR GI 08
.70
.70
.65
.65
.60
.55
.55
.50
-.
.50
.5i
S.40
.60
.35
.30
.20
-1.0
-. 9
-. 8
-. 7 -. 5 -. 5 -. 4 -. 3
LATITUDE X
SH
270
260
250
240
230
. 220
210 Z 200
-. 2
-. 1
0
NC PR GI 08
.20
0
.1
.2
.3
.4
.5
.6
LATITUDE X
.7
.8
.9
1.0
.8
.9
1.0
.8
.9
1.0
NCPR GI 08
NH
280
270
260
250
~240 ~
S230
190
220
.- ' 180
170
160
150'
140
130
120
-1.0
210
200N
190
180
-. 9
-. 8
-. 7
-. 6 -. 5 -. 4
LATITUDE X
SH
305
300
295
290
285
280
S275
270 265
........-........
.25
.25
-. 3
-. 2
-. 1
0
0
.1
.2
.3
.4
.5
.5
.7
LATITUOE X
T
NC PR GI 08
NH
NC PR GI 08
-.
300
295
290
285
280
275
--
-260 2S5 250 '
245
240
235
230
-1.0
170
270
-. 9
-. 8
-. 7
-. 6 -. 5 -. 4
LATITUDE X
-. 3
-. 2
-. 1
0
265 260
255
250
245'
0
.1
.2
.3
.4
.5
.6
LATITUDE X
.7
Figures
164
7.7.17.7.2
As in fig. 7.4, but for parameterized values for A and B from the
temperature - outgoing longwave radiation parameterization
I =A + BT in the equivalent EBMs.
771
300
200-
100
0
.C
V)
-~
C
C
-
o
-
-
U)
C
()
C
.C
V)
~
-C
~
U)
0a
0
0
a
0
772B
3
2-
0
a
C
U)
C
co
Cd
0
C
Y
C
U)
-U)
a)
A)
c
U)
U)
U
a)
C
c
C
)
00
U)U
U
cis
&
7&
a)&a
U
c
Figures
165
7.8.17.8.2
Parameterized outgoing longwave radiation (I) as a function of
the sine of latitude in Southern and Northern Hemispheres (SH
and NH) for NCAR (N), GFDL (P), and GISS (G) models, and for
observations (0). I is calculated via I=A+BT with the
parameterized A and B values and with the actual zonal mean
temperatures for each case (not with the two mode Legendre
representation of the temperature)
PRMOI NH
270
280
NC PR G1 8
260
260
250 240
S230
220
~220
200
210
180
200
190
160
180
140
120
-1.0
-
170
-. 9
-. 8
-. 7
-. 5 -. 5 -. 4
LATITUDE X
-. 3
-. 2
-. 1
0
0
.1
.2
.3
.3
.5
.4
LATITUDE X
.7
.3
.9
1.0
Figures
166
7.9.17.9.3
As in fig. 7.4, but for parameterized values for the albedo
components ao, a2, and 5 in the equivalent EBMs
alphaO
173
=
-2C =
r
cj_
=
-
0
0I
C)
CM
-
C
C
.0
=C
C
C0U
,
0
C'
9
0
7
-0CB
alpha2
0.3
0.2
79.2
0.1
0.0
0)
C
M
00
0
C
:2
M
:
U)
.
C
0
)
0-
-I
C
.0
0)B
delta
793
=~
=~
C
=
C
C
=~
C0
=
=
U
=
C
=
C
BB
B
B
0
Figures
167
Parameterized albedo as a function of the sine of latitude (x) in
7.10.17.10.2 Southern and Northern Hemispheres (SH and NH) for NCAR (N),
GFDL (P), and GISS (G) models, and for observations (0)
.75
.70
.60
.660
.55
.60
.55-
.5
.50
.40cc-
.35
.25
.20
-1.0 -. 9
.20
-. 8
-. 7
-. 6 -. 5 -. 4
LATITUOE X
-. 3
-. 2
-. 1
0
0
.1
.2
.3
.5
.5
.4
LATITUOE X
.7
.8
.9
Figures
1.0
168
7.11
Outgoing longwave radiation at the iceline latitude (Is) from
each of the twelve equivalent EBMs
Is
300
-
200
100
0
C
C
-
0
7.12
CD
()
Cl5
n
Cn
a
0
U)
M
=C)
C
)
-n
0)
-
)
C
=~=)
C
Ul)
(I
C
C)B
Dynamical transport efficiency
equivalent EBMs
Cl)
UC
U)
,
M
(l)
U
)
a.0
(D) from each of the twelve
~C
C
C
U)
=
U)
C
- U)
0
-
c
CC)
~
C)
U
-0
0
Figures
169
7.13.1Outgoing longwave radiation components
7.13.2 each of the twelve equivalent EBMs
(Io) and (I2)
from
10
250
2452402
235230
225
220
(/~
-0
-
C
~cl-
U
0
0
12
20
08
-20
cJ-40
-60-80-100
C V)
C
C
C
:5==
U)
:5
C
a
U)
C
C
(n
C
0
0C
-
Figures
h~CD
TO
T2
0
C.
\
-
.
.
ncarnh
ncarnh
ncarsh
nc ars h
gfdlnh0
gfdlnh
gfdlslh
gissnh
gisssh
gissnh
gisssh
obsnnhobsC)
obsnsh
obsnnh
t-)
ncargi
gfdlgl
gissgl
cgfdlgl
g
0
gissgl
obbnggl
obsnglw
171
7.15.1Sea level temperature from the original NCAR, GFDL, and GISS
7.15.8 GCMs and observations (solid lines), and calculated via the
equation T(x)=To+T 2 P 2 (x), where the To and T2 have been taken
from the equivalent EBMs (dashed lines). The results are
plotted as a function of the sine of latitude (x) for each
hemisphere
310
305
300
295
290
-285
315
305
295
-
--
5280
-275
270
265
260
285
275
265
255/
250
-1.0
-. 9
-. 8
305
300
295
290
285
2 280
,275
270
265
260
255
250
-1 .0 -. 9
-. 7
-. 5 -. 5 -. 4
LATITUDE X
-. 3
-. 2
-. 1
,
-. 8
-. 7
-. 6 -. 5 -. 4
LATITUDE X
rT 5SH T oTn
R
ANn
-. 3
-. 2
-. 1
PRMln
.1
e
280
270
260
250
240
-1. 0 -. 9
-. 8
.
.
-. 7
"
-. 6 -. 5 -. 4
LATITUDE X
T
-. 3
'
'.
-. 2
-. 1
'
.
-
.2
.3
.4
.5
.6
LATITUDE X
C!NH T
- -G
74RT
.7
.8
.9
1.
.8
.9
1.
.8
.9
1.0
.8
.9
1.0
PNn PR~n
275
270
265
260
255
250
305
300
295
290
285
2250
275
270
265
260
255
250
290
-
0
.1
.2
.3
305
290
295
.4
.5
.6
LATITUDE X
nJ FNH ,T
-
0
o{ T
.7
ANn PR~n
-
.1
.2
.3
y
300
280,-'
270
-
0
305
300
295
290
285
2280
300
310
255
0
.4
.5
.6
LATITUDE X
,NN
T
fi
PNf.
.7
Ppg
285
-'
275
260
265
250
255
240
230
-1.0
'
-. 9
-. 0 -. 7
-. 6 -. 5 -. 4
LATITUDE X
-. 3
'
-. 2
'
-. 1
0
245L0
.1
.2
.3
.4
.5
.5
LRTITUDE X
.7
Figures
172
7.16
Energy transport flux maximum from the equivalent EBMs
(solid bars) and from the original GCMs and Carissimo et al.
For the latter the global
observations (cross hatched bars).
values
hemispheric
'gl' values are averages of the
Flux Maximum EBM and GCM/OBS
N
E
C
0O
C
V)
CU
C
0
0
C
C,
U'
U'
'
C,0
C
0'
.0
.0
V)
0
0
a'
U'-
Flux max
GCM Fmax
M'
.
-
C
Figures
173
7.17
Iceline latitudes (x,) as input to the equivalent EBMs in each
case (solid bars), and as calculated in the equivalent EBMs
when the solar constant was increased by 2% in each case
(cross hatched bars)
Iceline latitude Xs
1.00
0.95
m Xs
E
Xs 2
0.90
0.85
0.80C
AA
C
)*
0
0
C
Figures
174
7.18
Climate sensitivity calculated from the equivalent EBMs
Climate sensitivity Beta
400
300
200
CM
4
C C
0
C
0
0
C
Figures
'h*1
Cbd.a
thoy
deeo 6
oiy
coner~s w6pbQm
eatnng h,~f
"
K.~
P.A
f.22asdeneteht
klhu.
conanuame
. es.
clenl; 0
nkefic
go-h d- reionity
lproonlwho#4~~~~~~~~1k
IIIWl?
pf.
w
arrmc
ulcrcpiuIrcied
Lw.
sc.,c: prsonatiha
Scetii PMq 0-
llu
I cho
.- plr
e."?Is
ae do barte ?
,
.4
e~an onteucoi
va,
ak
~
aebw
k
nrvn
h cec nw ginfreap~imAAi h
ht steoignlbnfti
conij
oliyl rcfnkibeig cplukV Hw I thrikprmrea en
ritcalcpscryw~ld~clpedDowthe"m~iso
conmnut
Doesam?l
poic reepon
Af. kk-6
*w,"? IA a~poro byob
and .1kd?
It . der
Aftteuah1kii rg~
0b.h1
00
176
8.1
Greenhouse
change
validation
framework
Figures
177
Greenhouse
change
validation
framework
Is it a decent theory? Is it supported by observations and simpler models? Is
it consistent? What are the uncertainties and how can they be reduced? Are
the results sensitive to the uncertainties? Do we have a better theory or
explanation? Can it be improved? What is believable in the projections?
What are possible impacts? Is the research program progressing? Does it
have more explanatory power than other programs
scientific papers on state of
kntowledge and development
of understanding of
greenhouse issue
Relevance to science faithfully interpreted? Is the communication effective
and usable? How can we do better? Are the uncertainties properly
communicated? Does the policy response depend on the uncertainties?
Critical capacity well developed? Does the information match the needs of
policy? Are links being exploited? How is the risk presented and
received? How are we learning? How can the science change to match
policy needs better? What is the marginal benefit in improving
communication of the science in trying to force a paradigm shift in the
policy arena? Is the validation philosophy being presented?
science presentation
concemns
evaluate perceived
taken into account in the policy process
beingindividualgroupsoanaional/global/resentfutre
EAre
costs/benefits of policies
evaluate perceived risks
evaluate interests and
dis tributive burdens
co nsider
usie
sue
andr
equity, morality
ide ntify coalitionscies
pol itical constituencies
tsidr toexplanations
reatioship
co nider relationships
erised ndplceto
otLh
er issues and policies
an<
Is
it a decent theory? Do observations and models agree? Is it a
defensible theory? Consensus? Do respectable scientists differ? Are the
differences due to uncertainties in the science? Can they be reduced?
How long to reduce them? What is the evidence against? Do alternative
or dissent have much support? Are the projections
meaningful? At what level do the mainstream scientists agree on them?
What do they imply?
defnsbl
theory
Consensus
differencesthscenri
du to urainaotie
hanecnlimte
impact
slate
ssthery
peplacaaetsurditprio
mdr n
potcal a
likel
ge gsenaioduxo
ofageclthimario x
ccmaued
(social change
orrert
ieptabihtylo
t
unpact
Dvauat
ic
su stat ikely impt.
worthro
ngnao
rcdrlycorc
rcs
inernatsonatheiraeans
Fmp
havene
Does ithmeoliy
/
to greenhouse warming or
global
differ?
Arac thet
agee? or
eer polc
lc
whosabeperlspev?dfrP
intenatinal
globN
scintst
teerdcd
and
dissentae goalsp
=aeo
proabiit
Do repetal
nthesinea
ano -epo
aed
other issues?
Figures
178
REFERENCES
AER Climate Group, 1987: Data inventory for P182. Investigation of GCM
current and double CO 2 climates. AER Climate Group, November 1987,
26pp.
AMS,
1989: Don't blame 1988 drought on greenhouse effect, study says. A MS
Newsletter, American Meteorological Society, 10 (4), p4. April 1989.
Alexander, Richard C., and Robert L. Mobley, 1976: Monthly average seasurface temperatures and ice-pack limits on a 10 global grid. Mon. Wea.
Rev., 104, 143-148.
Angell,
Variations and trends in tropospheric and stratospheric
J.K., 1989:
J. Climate, 1, 1296-1313.
global temperatures, 1958-1987.
Arrhenius, Svante, 1986: On the influence of carbonic acid in the air upon the
temperature of the ground. Phil. Mag. (Ser. 5), 41 (251), 237-276.
Arrhenius, Svante, 1908: Worlds in the making.
Harper & Bros., New York.
Bohren, Craig F., 1987: How do Greenhouses work?
in a glass of beer, Wiley, New York, 83-85.
In Craig F. Bohren, Clouds
U.S. study: greenhouse effect not seen in data.
Boston Globe, 1989:
globe, Wednesday Jan 25 1989.
Broecker, Wallace S.,
328, 123-126.
Bryan,
1987: Unpleasant surprises in the greenhouse?
Kirk, 1982: Poleward heat transport by the ocean:
models. Ann. Rev. Earth Planet. Sci., 10, 15-38.
Boston
Nature,
Observations
and
Bryan, Kirk, and Micheal J. Spelman, 1985: The ocean's response to a C0 2 induced warming. J. Geophys. Res., 90 (11), 679-688.
A closer look at last year's drought.
The summer of '88.
Burnam, L., 1989:
Scientific American, March 1989, 21.
Byrne, Gregory, 1989: Save the 'data'!
Science, 244, 290.
Carissimo, B.C., A.H. Oort, and T.H. Vonder Haar, 1985: Estimating the meridional
J. Phy. Oceanogr, 15,
energy transports in the atmosphere and ocean.
82-91.
Chamberlin, T.C., 1987: A group of hypotheses bearing on climatic changes.
Geol, 5, 653-683.
J.
Chylek, Peter, and J.A. Coakley, Jr., 1975: Analytical analysis of a Budyko-type
climate model. J. Atmos. Sci., 32, 675-679.
179
Clark, William C., 1982: Carbon Dioxide Review.
Oxford Univ. Press, 469pp.
The critical appraisal of
Clark, William C., and Giandomenico Majone, 1985:
Science, Tech., and Human
scientific inquires with policy implications.
Values, 10 (3), 6-19.
Covey, Curt, 1988: Atmospheric and oceanic heat transport: simulations versus
observations. Climatic Change, 13, 149-159.
Dickinson, Robert E., 1989: Uncertainties
review. Climatic Change, 15, 5-13.
of estimates
of climate change:
A
Fourier, Baron Jean Baptiste, 1827: Memoire sur les temperatures du globe
Mem. de l'Ac. R. d. Sci. de l'Inst. de
terrestre et des espaces planetaires.
France, 7, 572-604.
Grotch,
Stanley L., 1988: Regional intercomparisons of general circulation
DOE/NBB-0084, TR041,
model predictions and historical climate data.
291pp.
United States Department of Energy,
Gutowski, W.J., D.S.
balance of
response to
United States
Gutzler, D. Portman, and W.-C. Wang, 1988: Surface energy
three general circulation models: Current climate and
DOE/ER/60422-H1, TR042,
increasing atmospheric CO 2 .
Department of Energy, 119pp.
Hansen, James E., 1988: The greenhouse effect: Impacts on current global
Statement to United States House
temperature and regional heat waves.
of Representatives, Committee on Energy and Commerce, Subcommittee
on Energy and Natural Resources, July 7, 1988.
Statement
Hansen, James E., 1989: Modeling Greenhouse climate effects.
presented to United States Senate, Committee on Commerce, Science and
Transportation, Subcommittee on Science, technology, and Space, April
20, 1989.
Hansen, J., D. Johnson, A. Lacis, S. Lebedeff, P. Lee, D. Rind, and G. Russell, 1981:
Science, 213,
Climate impact of increasing atmospheric carbon dioxide.
957-966.
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Reudy, and L.
Travis, 1983: Efficient three-dimensional global models for climate
studies: Models I and II. Mon.. Wea. Rev., 111 (4), 609-662.
Hansen, J., A.
Lerner,
Climate
Maurice
Lacis, D. Rind, G. Russell, P. Stone, I. Fung, R. Reudy, and J.
1984: Climate sensitivity: Analysis of feedback mechanisms.
Processes and Climate Sensitivity, Geophysical Monograph 29,
Ewing Volume 5, 130-163.
Hansen, J., G. Russell, A. Lacis, I. Fung, D. Rind, and P. Stone, 1985: Climate
response times: Dependence on climate sensitivity and ocean mixing.
Science, 229, 857-859.
References
180
Hansen, James, and Sergej Lebedeff, 1987: Global trends of measured surface
air temperature. J. Geophy. Res., 92 (Dl), 13,345-13,372.
Hansen J., I. Fung, A. Lacis, D. Rind, S. Lebedeff, R. Ruedy, G. Russell, and P.
Stone, 1988: Global climate changes as forecast by Goddard Institute for
Space Studies three-dimensional model. J. Geophy. Res., 93, 9341-??.
Hansen, J., D. Rind, A. DelGenio, A. Lacis, S. Lebedeff, M. Prather, R. Ruedy, and
T. Karl, 1989: Regional greenhouse climate effects. In Preparing for
climate change, Proceedings of the second North American conference
on preparing for climate change, December 6-8, 1988, Climate Institute,
Washington, D.C. 1989.
Are atmospheric "greenhouse"
Hanson, K., G.A. Maul, and T.R. Karl, 1989:
effects apparent in the climatic record of the contiguous U.S. (18951987)? Geophys. Res. Letters, 16 (1), 49-52.
Hastenrath, Stefan, 1980: Heat budget of tropical ocean and atmosphere.
Phys. Oceanogr., 10, 159-170.
J.
Hastenrath, S., 1982: On meridional heat transport in the world ocean. J. Phys.
Oceanogr., 12, 922-927.
Held, I.M., and M.J. Suarez,
Tellus, 26, 613-629.
1974: Simple albedo feedback models of ice caps.
Manufacturing consent.
Herman, Edward S., and Noam Chomsky, 1988:
political economy of the mass media. Pantheon, New York, 412pp.
Idso,
The
S.B., 1980: The climatological significance of a doubling of earth's
Science, 207, 1462-1463.
atmospheric carbon dioxide concentration.
Imbrie, J., and J.Z. Imbrie, 1980: Modeling the
variations. Science, 207, 943-953.
climatic response to orbital
In These
In These Times, 1989: Greenhouse neglect: missed signs of the Times?
Times, Jan 25-31 1989.
Jaeger, Jill, 1988:
Developing policies for responding to
climatic
change.
WMO/ID - No. 225, 53pp.
Jones, P.D., T.M.L. Wigley, and P.B. Wright, 1986: Global temperature variations
between 1861 and 1984. Nature, 322, 430-434.
Jones, P.D., T.M.L. Wigley, C.K. Folland, and D.E. Parker, 1987: Spatial patterns
Climate Monitor, 16, 175-185.
in recent worldwide temperature trends.
Commentary on 'Detecting
Kelly, P.M., T.M.L. Wigley, and P.D. Jones, 1982:
effects on climate'. In W.C. Clark ed., The Carbon Dioxide Review, Oxford
Univ. Press, 469pp.
Kerr,
Richard A., 1989a: Hansen vs.
Science, 244, 1041-1043.
the
world
on the
greenhouse
threat.
References
181
Kerr, Richard A.,
1118-1119.
1989b:
Greenhouse skeptic out in the cold.
Keepin, Bill, 1986: Review of global
Ann. Rev. Energy, 11, 357-392.
energy and
Science, 246,
carbon dioxide projections.
Kellogg, W.W., 1977: Effects of human activities on global climate.
WM O
Technical Note 156, WMO No. 486, World Meteorological Organization,
Geneva.
Kellogg, William W., and Zong-Ci Zhao, 1988: Sensitivity of soil moisture to
doubling of carbon dioxide in climate model experiments. Part I: North
America. J. Climate, 1, 348-366.
Kuhn, Thomas, 1970: The structure of scientific revolutions.
Press, Chicago, 210pp.
Kukla, G., and J. Gavin, 1981:
497-503.
Univ. of Chicago
Summer ice and carbon dioxide.
Science, 214,
Lakatos, Imre, 1970: Falsification and the methodology of scientific research
programmes. In Imre Lakatos and Alan Musgrave eds. Criticism and the
growth of knowledge.
Cambridge University Press, 91-196.
Lacis, A.A., and J.E. Hansen, 1974: Parameterization for the absorption of solar
radiation in the earth's atmosphere. J. Atmos. Sci., 31, 118-133.
Lindzen, Richard S., 1989:
Greenhouse
Unpublished manuscript.
warming:
science
v.
consensus.
Luther, Frederick M., Robert G. Ellingson, Yves Fouquart, Stephen Fels, Noelle
A. Scott, and Warren J. Wiscombe, 1988: Intercomparison of radiation
codes in climate models (ICRCCM): Longwave clear-sky results - A
workshop summary. Bull. Amer. Met. Soc., 69 (1), 40-48.
Manabe, Syukuro, 1983: Carbon dioxide and climatic change.
39-82.
Adv.
Geophy., 25,
Manabe, Syukuro, and Richard T. Wetherald, 1975: The effect of doubling the
J.
C 02-concentration of the climate of a general circulation model.
Atmos. Sci., 32, 3-15.
Manabe, Syukuro, and Kirk Bryan, Jr., 1985: C0 2 -induced change in a coupled
ocean-atmsophere model and its paleoclimatic implications.
J. Geophy.
Res., 90 (C6), 11,689-11,707.
Manabe, S., and R. T. Wetherald, 1987: Large-scale changes of soil wetness
induced by an increase in atmospheric carbon dioxide.
J. Atmos. Sci., 4 4
(8), 1211-1235.
Manabe, S., and R.J. Stouffer, 1988: Two stable equilibria of a coupled oceanatmosphere model. J. Climate, 1, 841-866.
References
182
Marshall Institute, 1989:
Scientific perspectives on the greenhouse problem.
The George C. Marshall Institute, Washington D.C., 37pp.
Miller, James R., Gary L. Russell, and Lie-Ching Tsang, 1983: Annual oceanic
Dyn. Atm.
heat transports computed from an atmospheric model.
95-109.
Oceans, 7,
Mitchell, J.F.B., 1988: Simulation of climate change due to increased
atmospheric CO 2 . In M.E. Schlesinger (ed.), Physically-based Modelling
and simulation of climate and climatic change - Part II, Kluwer
Academic Publishers, 1009-1051.
Mitchell, J.F.B., C.A. Wilson, and W.M. Cunnington, 1987: On CO 2 climate
sensitivity and model dependence of results. Q. J. R. Meteorol. Soc., 113,
293-322.
Mitchell J.F.B., C.A. Senior, and W.J. Ingram 1989:
feedback? Nature, 341, 132-134.
Namias, Jerome, 1989:
Cold waters and hot summers.
CO 2 and climate: a missing
Nature, 338, 15-16.
Carbon dioxide and climate: A scientific
National Research Council, 1979:
assessment. National Academy Press, Washington, D.C., 22pp.
Carbon dioxide and climate: A
National Research Council, 1982:
assessment. National Academy Press, Washington, D.C., 72pp.
second
Newell, Nicholas E., Reginald E. Newell, Jane Hsuing, and Wu Zhongxiang, 1989:
Global marine temperature variation and the solar magnetic cycle.
Geophys. Res. Letters, 16 (4), 311-314.
Newell, Reginald E., and Thomas G. Dopplick, 1979: Questions concerning the
possible influence of anthropogenic CO 2 on atmospheric temperature. J.
Appl. Meteror., 18 (6), 822-825.
Newsweek, 1989: Is it all just hot air?
Newsweek, 20 Nov. 1989, 64-66.
Study finds warming trend that could raise sea levels.
New York Times, 1981a:
New York Times, Aug 22 1981.
New York Times, 1981b:
Oct 19 1981.
Evidence is found of warming trend.
New York Times,
New York Times, 1989a: U.S. data since 1895 fail to show warming trend.
York Times, Jan 26 1989.
New York Times, 1989b:
Hot summers in 1990's.
New
New York Times, Jan 11 1989.
Scientists link '88 drought to natural cycle in tropical
New York Times, 1989c:
Pacific. New York Times, Science Times, Jan 3 1989.
New York Times, 1989d: Scientist says Budget Office altered his testimony.
York Times, 8 May 1989, 1.
New
References
183
North, Gerald R., 1975: Theory of energy-balance climate models.
32 (11), 2033-2043.
J. Atmos. Sci.,
North, Gerald R., Robert F. Cahalan, and James A. Coakley, Jr., 1981: Energy
balance climate models. Rev. Geophy. Space Phy., 19 (1), 91-121.
Oort, A.H., 1983:
Global atmospheric circulation statistics, 1958-1973.
NOAA
Prof. Pap. No. 14, Rockville, Maryland, April 1983, 180pp.
Oort, Abraham H., and Thomas H. Vonder Haar, 1976: On the observed annual
cycle in the ocean-atmosphere heat balance over the Northern
Hemisphere. J. Phy. Oceanogr., 6 (6), 781-800.
Palmer, T.N., and C. Brankovic, 1989: The 1988 U.S. drought linked to anomalous
Nature, 338, 54-57.
sea surface temperature.
scenarios for a C0 2 -
Pittock, A.B., and M.J. Salinger, 1982: Towards regional
warmed earth. Climatic Change, 4, 23-40.
Ramanathan, V., 1981: The role of ocean-atmosphere interactions
climate problem. J. Atmos. Sci., 38, 918-930.
in the CO 2
Ramanathan, V., 1988: The radiative and climatic consequences of the
changing atmospheric composition of trace gases. In The changing
atmosphere, eds. F.S. Rowland and I.S.A. Isaksen, John Wiley & Sons Ltd.,
159-186.
Ramanathan, V., L. Callis, R. Cess, J. Hansen, I. Isaksen, W. Kuhn, A. Lacis, F.
Luther, J. Mahlman, R. Reck, and M. Schlesinger, 1987: Climate-chemical
R e v.
interactions and effects of changing atmospheric trace gases.
Geophy., 25 (7), 1441-1482.
Ramanathan, V., Bruce R. Barkstrom, and Edwin F. Harrison, 1989: Climate and
the earth's radiation budget. Physics Today, May 1989, 22-32.
Ravetz,
Scientific
Jerome, R., 1971:
Clarendon Press, Oxford, 449pp.
knowledge
its social problems.
and
Rind, David, Arthur Rosenzweig, and Cynthia Rosenzweig,
future: a joint venture. Nature, 334, 483-486.
1988: Modelling the
Rind, D., R. Goldberg, and R. Ruedy, 1989: Change in climate variability in the
21st Century. Climatic Change, 14, 5-37.
Saxe, John G., 1955: The blindman and the elephant.
treasury of the familiar, MacMillan.
In Ralph Woods ed., A
Quantitative analysis of feedbacks in climate
Schlesinger, Michael E., 1989:
In Understanding Climate Change, A. Berger, R.E.
model simulations.
Dickinson, and J.W. Kidson eds., Geophysical Monograph 52, IUGG
Volume 7, American Geophysical Union, Washington D.C., 177-187.
References
184
Schlesinger, Micheal E., and John F.B. Mitchell, 1985: Model projections of the
In
equilibrium climatic response to increased carbon dioxide.
Projecting the climatic effects of increasing carbon dioxide, M.C.
MacCracken, and F.M. Luther (eds.), DOE/ER-0237, 81-147.
Schlesinger, Micheal E., and John F.B. Mitchell, 1987: Climate model simulations
Rev.
of the equilibrium climatic response to increased carbon dioxide.
Geophy., 25 (4), 760-798.
Schneider, Stephen H., 1984: On the empirical verification of model-predicted
CO 2 -induced climatic effects. Climate Processes and Climate Sensitivity,
Geophysical Monograph 29, Maurice Ewing Volume 5, 187-201.
Schneider, Stephen H., 1987: Climate Modeling.
72-80.
Scientific American, May 1987,
Schneider, Stephen H., 1989: Global warming: Is it real and should it be part of
In Global change and our common future.
a global change program?
National Research Council, National Academy
Papers from a forum.
Press, Washington D.C. 1989, 209-219.
Schneider, Stephen H., and Robert E. Dickinson, 1974: Climate modeling.
Geophy. Space Phy., 12 (3), 447-493.
Rev.
Schneider, S.H., and S.L. Thompson, 1981: Atmospheric CO 2 and climate:
Importance of the transient response. J. Geophy. Res., 86, 3135-3147.
Sellers, William D., 1965: Physical
Press, Chicago, 272pp.
climatology.
The University of Chicago
Somerville, R.C.J., P.H. Stone, M. Halem, J.E. Hansen, J.S. Hogan, L.M. Druyan, G.
Russell, A.A. Lacis, W.J. Quirk, and J. Tenenbaum, 1974: The GISS model of
the global atmosphere. J. Atmos. Sci., 31, 84-117.
Somerville, Richard C.J., and Lorraine A. Remer, 1984: Cloud optical thickness
feedbacks in the CO 2 climate problem. J. Geophy. Res., 89 (D6), 96689672.
Spelman, M.J., and S. Manabe, 1984: Influence of oceanic heat transport upon
the sensitivity of a model climate. J. Geophy. Res., 89 (Cl), 571-586.
Momentum and zonal kinetic
Starr, V.P., J.P. Peixoto, and N.E. Gaut, 1970:
years of hemispheric data.
five
energy balance of the atmosphere from
Tellus, 22, 251-274.
Stephens, G.L., G.G. Campbell, and T.H. Vonder Haar, 1981:
budgets. J. Geophys. Res., 86 (C1O), 9739-9760.
Stevens, William K., 1989: Skeptics are challenging
New York Times, 13 Dec. 1989, pl.
dire
Earth radiation
'greenhouse'
Stone, Peter H., 1978a: Constraints on dynamical transports of energy
spherical planet. Dyn. Atm. Oceans, 2, 123-139.
views.
on a
References
185
Stone, Peter H., 1978b: Baroclinic adjustment.
Stone,
J. Atmos. Sci., 35 (4), 561-571.
Peter H., 1984: Feedbacks between dynamical heat fluxes and
Climate Processes and Climate
temperature structure in the atmosphere.
Sensitivity, Geophysical Monograph 29, Maurice Ewing Volume 5, 6-17.
Stone, P.H., S. Chow, and W.J. Quirr, 1977: The July climate and a comparison of
the January and July climates simulated by the GISS general circulation
model. Mon. Wea. Rev., 105 (2), 170-194.
Stone,
Empirical relations between
Peter H., and Dennis A. Miller, 1980:
seasonal changes in meridional temperature gradients and meridional
fluxes of heat. J. Atmos. Sci., 37 (8), 1708-1721.
Interhemispheric asymmetry in
Stouffer, R.J., S. Manabe, and K. Bryan, 1989:
climate response to a gradual increase of atmospheric CO 2 . Nature, 342,
660-662.
Trenberth, K.E., G.W. Branstator, and P.A. Arkin, 1988:
North American drought. Science, 242, 1640-1645.
Origins of the 1988
Twomey, S.A., M. Piepgrass, and T.L. Wolfe, 1984: An assessment of the impact
of pollution on global cloud albedo. Tellus, 36B, 356-366.
Wang, Wei-Chyung, and Peter H. Stone, 1980: Effect of ice-albedo feedback on
global sensitivity in a one-dimensional radiative-convective climate
model. J. Atmos. Sci., 37 (3), 545-552.
Washington, Warren M., and Gerald A. Meehl, 1984: Seasonal cycle experiment
on the climate sensitivity due to a doubling of CO 2 with an atmospheric
general circulation model coupled to a simple mixed-layer ocean model.
J. Geophy. Res., 89 (D6), 9475-9503.
Wigley, T.M.L., and P.D. Jones, 1981:
Nature, 292, 205-208.
Detecting CO 2 induced climatic change.
Wigley, T.M.L., and B.D. Santer, 1988: Validation of general circulation climate
In M.E. Schlesinger (ed.), Physically-based Modelling and
models.
simulation of climate and climatic change - Part II, Kluwer Academic
Publishers, 841-879.
Wilson, C.A., and J.F.B. Mitchell, 1987: A doubled CO 2 climate sensitivity
J.
experiment with a global climate model including a simple ocean.
13,315-13,343.
Geophy. Res., 92 (DI1),
Wunsch, C., 1980: Meridional heat flux of the North Atlantic Ocean. Proc. Nat.
Acad. Sci. U.S.A., 77, 5043-5047.
References