Climate Models An Assessment of Strengths and Limitations

Climate Models
An Assessment of
Strengths
and Limitations
U.S. Climate Change Science Program

Synthesis and Assessment Product 3.1
July 2008
FED ERAL EX ECUTIVE TEAM
Director, U.S. Climate Change Science Program...........................................................William J. Brennan
Director, Climate Change Science Program Office .......................................................Peter A. Schultz
Lead Agency Principal Representative to CCSP;

Associate Director, Department of Energy, Office of Biological
and Environmental Research .........................................................................................Anna Palmisano
Product Lead; Department of Energy,

Office of Biological and Environmental Research ........................................................Anjuli S. Bamzai
Synthesis and Assessment Product Advisory Group Chair;

Associate Director, EPA National Center for Environmental
Assessment.....................................................................................................................Michael W. Slimak
Synthesis and Assessment Product Coordinator,

Climate Change Science Program Office ......................................................................Fabien J.G. Laurier
OTH ER AGEN CY REPRESEN TATIVES
National Aeronautics and Space Administration ...........................................................Donald Anderson

National Oceanic and Atmospheric Administration ......................................................Brian D. Gross
National Science Foundation .........................................................................................Jay S. Fein
This document does not express any regulatory policies of the United States or any of its agencies, or provide recommendations for
regulatory action. Further information on the process for preparing Synthesis and Assessment products and the CCSP itself can be found
at www.climatescience.gov.
Climate Models
An Assessment of
Strengths
and Limitations
Synthesis and Assessment Product 3.1

Report by the U.S. Climate Change Science Program
and the Subcommittee on Global Change Research
AUTHO RS:
D avid C . Bader, Lawrence Livermore N ational Laboratory
Curt Covey, Lawrence Livermore N ational Laboratory
W illiam J. Gutowski Jr., Iowa State University
Isaac M. Held, N O AA Geophysical Fluid D ynamics Laboratory
Kenneth E. Kunkel, Illinois State W ater Survey
Ronald L. Miller, N ASA Goddard Institute for Space Studies
Robin T. Tokmakian, N aval Postgraduate School
Minghua H. Z hang, State University of N ew York Stony Brook
ACKN O W LED GEMEN T
This report has been peer reviewed in draft form by individuals chosen for their diverse perspectives and technical ex-
pertise. The expert review and selection of reviewers followed the OMB’s Information Quality Bulletin for Peer Review.
The purpose of this independent review is to provide candid and critical comments that will assist the Climate Change
Science Program in making this published report as sound as possible and to ensure that the report meets institutional
standards. The peer-review comments, draft manuscript, and response to the peer-review comments are publicly avail-
able at: www.climatescience.gov/Library/sap/sap3-1/default.php.
We wish to thank the following individuals for their peer review of this report:
Kerry H. Cook, University of Texas Austin
Carlos R. Mechoso, University of California Los Angeles
Gerald A. Meehl, National Center for Atmospheric Research
Phil Mote, University of Washington Seattle
Brad Udall, Western Water Assessment, Boulder, Colorado
John E. Walsh, International Arctic Research Center
We would also like to thank the following individuals who provided comments during the public comment period:
California Department of Water Resources: Michael Anderson
NOAA Research Council: Derek Parks, Tim Eichler, Michael Winton, Ron Stouffer, and Jiayu Zhou
NOAA Office of Federal Coordination of Meteorology: Samuel P. Williamson
NSF: Marta Cehelsky
The public review comments, draft manuscript, and response to the public comments are publicly available at:
www.climatescience.gov/Library/sap/sap3-1/default.php
Intellectual contributions from the following individuals are also acknowledged: John J. Cassano, Elizabeth N. Cassano,
Peter Gent, Bala Govindasamy, Xin-Zhong Liang, William Lipscomb, and Thomas J. Phillips.
ED ITO RIAL TEAM

Technical Editors ................................................................................Judy Wyrick, Anne Adamson,
Oak Ridge National Laboratory
Report Coordinators ..........................................................................Judy Wyrick, Anne Adamson, Shirley Andrews,
Oak Ridge National Laboratory
Technical Advisor ..............................................................................David Dokken, CCSPO
Graphic Production ............................................................................DesignConcept
Recommended Citation for the entire report

CCSP, 2008: Climate Models: An Assessment of Strengths and Limitations. A Report by the U.S. Climate Change Sci-
ence Program and the Subcommittee on Global Change Research [Bader D.C., C. Covey, W.J. Gutowski Jr., I.M. Held,
K.E. Kunkel, R.L. Miller, R.T. Tokmakian and M.H. Zhang (Authors)]. Department of Energy, Office of Biological and
Environmental Research, Washington, D.C., USA, 124 pp.
iv
Climate Models: An Assessm ent of Strengths and Lim itations
TABLE O F CO N TEN TS Executive Summary............................................................................................1
CH APTERS
1 ........................................................................................................................ 7
I ntroduction
2 ...................................................................................................................... 13
D escription of Global Climate System Models
3 ...................................................................................................................... 31
A dded Value of Regional Climate Model Simulations
4 ...................................................................................................................... 39
M odel Climate Sensitivity
5 ...................................................................................................................... 51
M odel Simulation of Major Climate Features
6 ...................................................................................................................... 85
Future Model D evelopment
7 ...................................................................................................................... 91
Example Applications of Climate Model Results
References ........................................................................................................ 97
v
The U.S. Climate Change Science Program
vi
EX ECUTIVE SUMMARY
Scientists extensively use mathematical models of Earth’s climate, executed

on the most powerful computers available, to examine hypotheses about
past and present-day climates. D evelopment of climate models is fully con-
sistent with approaches being taken in many other fields of science deal-
ing with ver y complex systems. These climate simulations provide a
framework within which enhanced understanding of climate-relevant
processes, along with improved observations, are merged into coherent
projections of future climate change.This report describes the models and
their ability to simulate current climate.
The science of climate modeling has matured through finer spatial resolution, the inclusion of a greater number of physical
processes, and comparison to a rapidly expanding array of observations.These models have important strengths and limita-
tions. They successfully simulate a growing set of processes and phenomena; this set intersects with, but does not fully cover,
the set of processes and phenomena of central importance for attribution of past climate changes and the projection of fu-
ture changes. Following is a concise summary of the information in this report, organized around questions from the “Prospec-
tus,” which motivated its preparation, and focusing on these strengths and weaknesses.
still needed for full scientific evaluation of the

W hat are the major components and state of the science.
processes of the climate system that are
included in present state-of-the-science The set of most recent climate simulations, re-
climate models, and how do climate mod- ferred to as CMIP3 models and utilized heavily
els represent these aspects of the climate in Working Group 1 and 2 reports of the Fourth
system? IPCC Assessment, have received unprecedented
scrutiny by hundreds of investigators in various
areas of expertise. Although a number of sys-
Chapter 2 describes the four major components tematic biases are present across the set of mod-
of modern coupled climate models: atmosphere, els, more generally the simulation strengths and
ocean, land surface, and sea ice. The develop- weaknesses, when compared against the current
ment of each of these individual components climate, vary substantially from model to
raises important questions as to how key phys- model. From many perspectives, an average
ical processes are represented in models, and over the set of models clearly provides climate
some of these questions are discussed in this re- simulation superior to any individual model,
port. Furthermore, strategies used to couple the thus justifying the multimodel approach in
components into a climate system model are de- many recent attribution and climate projection
tailed. Development paths for the three U.S. studies.
modeling groups that contributed to the 2007
Intergovernmental Panel on Climate Change Climate modeling has been steadily improving
(IPCC) Scientific Assessment of Climate over the past several decades, but the pace has
Change (IPCC 2007) serve as examples. Expe- been uneven because several important aspects
rience and expert judgment are essential in con- of the climate system present especially severe
structing and evaluating a climate modeling challenges to the goal of simulation.
system, so multiple modeling approaches are
1
The U.S. Climate Change Science Program Executive Summary
cycle in solar irradiance; paleoclimatic infor-

H ow are changes in the Earth’s energy
mation, particularly from the peak of the last Ice
balance incorporated into climate mod-
Age some 20,000 years ago; aspects of the sea-
els? H ow sensitive is the Earth’s (mod-
sonal cycle; and the magnitude of observed
eled) climate to changes in the factors
warming over the past century. Because each
that affect the energy balance?
test is subject to limitations in data and compli-
cations from feedbacks in the system, they do
not provide definitive tests of models’ climate
The Earth’s radiant energy balance at the top of
sensitivity in isolation. Studies in which multi-
the atmosphere helps to determine its climate.
ple tests of model climate responses are con-
Chapter 2 contains a brief description of energy-
sidered simultaneously are essential when
transfer simulation within models, particularly
analyzing these constraints on sensitivity.
within the atmospheric component. More im-
portant, Chapter 4 includes an extensive dis-
Improvements in our confidence in estimates of
cussion about radiative forcing of climate
climate sensitivity are most likely to arise from
change and climate sensitivity. The response of
new data streams such as the satellite platforms
global mean temperature to a doubling of car-
now providing a first look at the three-dimen-
bon dioxide remains a useful measure of climate
sional global distributions of clouds. New and
sensitivity. The equilibrium response—the re-
very computationally intensive climate model-
sponse expected after waiting long enough
ing strategies that explicitly resolve some of the
(many hundreds of years) for the system to
smaller scales of motion influencing cloud
reequilibrate—is the most commonly quoted
cover and cloud radiative properties also prom-
measure. Remaining consistent for three
ise to improve cloud simulations.
decades, the range of equilibrium climate sen-
sitivity obtained from models is roughly con-
sistent with estimates from observations of
H ow uncertain are climate model results?
recent and past climates. The canonical three-
In what ways has uncertainty in model-
fold range of uncertainty, 1.5 to 4.5°C, has
based simulation and prediction changed
evolved very slowly. The lower limit has been
with increased knowledge about the cli-
nearly unchanged over time, with very few re-
mate system?
cent models below 2°. Difficulties in simulat-
ing Earth’s clouds and their response to climate
change are the fundamental reasons preventing
Chapter 1 provides an overview of improvement
a reduction in this range in model-generated cli-
in models in both completeness and in the abil-
mate sensitivity.
ity to simulate observed climate. Climate mod-
els are compared to observations of the mean
Other common measures of climate sensitivity
climate in a multitude of ways, and their ability
measure the climate response on time scales
to simulate observed climate changes, particu-
shorter than 100 years. By these measures there
larly those of the past century, have been exam-
is considerably less spread among the models—
ined extensively. A discussion of metrics that
roughly a factor of two rather than three. The
may be used to evaluate model improvement
range still is considerable and is not decreasing
over time is included at the end of Chapter 2,
rapidly, due in part to difficulties in cloud sim-
which cautions that no current model is supe-
ulation but also to uncertainty in the rate of heat
rior to others in all respects, but rather that dif-
uptake by the oceans. This uncertainty rises in
ferent models have differing strengths and
importance when considering the responses on
weaknesses.
these shorter time scales.
As discussed in Chapter 5, climate models de-
Climate sensitivity in models is subjected to
veloped in the United States and around the
tests using observational constraints. Tests in-
world show many consistent features in their
clude climate response to volcanic eruptions;
simulations and projections for the future. Ac-
aspects of internal climate variability that pro-
curate simulation of present-day climatology for
vide information on the strength of climatic
near-surface temperature and precipitation is
“restoring forces”; the response to the 11-year
necessary for most practical applications of cli-
2
mate modeling. The seasonal cycle and large- atmosphere, is sensitive to deficiencies in sim-
scale geographical variations of near-surface ulated winds and salinities, but a subset of mod-
temperature are indeed well simulated in recent els is producing realistic circulation in the
models, with typical correlations between mod- Southern Ocean as well.
els and observations of 95% or better.
Models forced by the observed well-mixed
Climate model simulation of precipitation has greenhouse gas concentrations, volcanic
improved over time but is still problematic. Cor- aerosols, estimates of variations in solar energy
relation between models and observations is 50 incidence, and anthropogenic aerosol concen-
to 60% for seasonal means on scales of a few trations are able to simulate the recorded 20th
hundred kilometers. Comparing simulated and Century global mean temperature in a plausible
observed latitude-longitude precipitation maps way. Solar variations, observed through direct
reveals similarity of magnitudes and patterns in satellite measurements for the last few decades,
most regions of the globe, with the most strik- do not contribute significantly to warming dur-
ing disagreements occurring in the tropics. In ing that period. Solar variations early in the 20th
most models, the appearance of the Inter-Trop- Century are much less certain but are thought
ical Convergence Zone of cloudiness and rain- to be a potential contributor to warming in that
fall in the equatorial Pacific is distorted, and period.
rainfall in the Amazon Basin is substantially un-
derestimated. These errors may prove conse- Uncertainties in the climatic effects of man-
quential for a number of model predictions, made aerosols (liquid and solid particles sus-
such as forest uptake of atmospheric CO2. pended in the atmosphere) constitute a major
stumbling block in quantitative attribution stud-
Simulation of storms and jet streams in middle ies and in attempts to use the observational
latitudes is considered one of the strengths of record to constrain climate sensitivity. We do
atmospheric models because the dominant not know how much warming due to green-
scales involved are reasonably well resolved. As house gases has been cancelled by cooling due
a consequence, there is relatively high confi- to aerosols. Uncertainties related to clouds in-
dence in the models’ ability to simulate changes crease the difficulty in simulating the climatic
in these extratropical storms and jet streams as effects of aerosols, since these aerosols are
the climate changes. Deficiencies that still exist known to interact with clouds and potentially
may be due partly to insufficient resolution of can change cloud radiative properties and cloud
features such as fronts, to errors in the forcing cover.
terms from moist physics, or to inadequacies in
simulated interactions between the tropics and The possibility that natural variability has been
midlatitudes or between the stratosphere and the a significant contributor to the detailed time
troposphere. These deficiencies are still large evolution seen in the global temperature record
enough to impact ocean circulation and some is plausible but still difficult to address with
regional climate simulations and projections. models, given the large differences in charac-
teristics of the natural decadal variability be-
The quality of ocean climate simulations has tween models. While natural variability may
improved steadily in recent years, owing to bet- very well be relevant to observed variations on
ter numerical algorithms and more realistic as- the scale of 10 to 30 years, no models show any
sumptions concerning the mixing occurring on hint of generating large enough natural, un-
scales smaller than the models’ grid. Many of forced variability on the 100-year time scale to
the CMIP3 class of models are able to maintain compete with explanations that the observed
an overturning circulation in the Atlantic with century-long warming trend has been predomi-
roughly the observed strength without the arti- nantly forced.
ficial correction to air-sea fluxes commonly
used in previous generations of models, thus The observed southward displacement of the
providing a much better foundation for analysis Southern Hemisphere storm track and jet
of the circulation’s stability. Circulation in the stream in recent decades is reasonably well sim-
Southern Ocean, thought to be vitally important ulated in current models, which show that the
for oceanic uptake of carbon dioxide from the displacement is due partly to greenhouse gases
3
but also partly to the presence of the stratos- tendency toward too short a period. Bias in the
pheric ozone hole. Circulation changes in the Inter-Tropical Convergence Zone (ITCZ) in
Northern Hemisphere over the past decades coupled models is a major factor preventing fur-
have proven more difficult to capture in current ther improvement in these models. Projections
models, perhaps because of more complex infor future El Niño variability and the state of the
teractions between the stratosphere and tropo- Pacific Ocean are centrally important for re-
sphere in the Northern Hemisphere. gional climate change projections throughout
the tropics and in North America.
Observations of ocean heat uptake are begin-
ning to provide a direct test of aspects of the Other aspects of the tropical simulations in cur-
ocean circulation directly relevant to climate rent models remain inadequate. The Madden-
change simulations. Coupled models provide Julian Oscillation, a feature of the tropics in
reasonable simulations of observed heat uptake which precipitation is organized by large-scale
in the oceans but underestimate the observed eastward-propagating features with periods of
sea-level rise over the past decades. roughly 30 to 60 days, is a useful test of simu-
lation credibility. Model performance using this
Model simulations of trends in extreme weather measure is still unsatisfactory. The “double
typically produce global increases in extreme ITCZ–cold tongue bias,” in which water is ex-
precipitation and severe drought, with decreases cessively cold near the equator and precipitation
in extreme minimum temperatures and frost splits artificially into two zones straddling the
days, in general agreement with observations. equator, remains as a persistent bias in current
coupled atmosphere-ocean models. Projections
Simulations from different state-of-the-science of tropical climate change are affected adversely
models have not fully converged, however, since by these deficiencies in simulations of the or-
different groups approach uncertain model as- ganization of tropical convection. Models typi-
pects in distinctive ways. This absence of con- cally overpredict light precipitation and
vergence is one useful measure of the state of underpredict heavy precipitation in both the
climate simulation; convergence is to be ex- tropics and middle latitudes, creating potential
pected once all climate-relevant processes are biases when studying extreme events. Tropical
simulated in a convincing physically based cyclones are poorly resolved by the current gen-
manner. However, measuring the quality of cli- eration of global models, but recent results with
mate models so the metric used is directly rele- high-resolution atmosphere-only models and
vant to our confidence in the models’ dynamical downscaling provide optimism that
projections of future climate has proven diffi- the simulation of tropical cyclone climatology
cult. The most appropriate ways to translate will advance rapidly in coming years, as will
simulation strengths and weaknesses into con- our understanding of observed variations and
fidence in climate projections remain a subject trends.
of active research.
The quality of simulations of low-frequency
variability on decadal to multidecadal time
H ow well do climate models simulate scales varies regionally and also from model to
natural variability and how does variabil- model. On average, models do reasonably well
ity change over time? in the North Pacific and North Atlantic. In other
oceanic regions, lack of data contributes to un-
certainty in estimating simulation quality at
Simulation of climate variations also is de- these low frequencies. A dominant mode of
scribed in Chapter 5. Simulations of El Niño os- low-frequency variability in the atmosphere,
cillations, which have improved substantially in known as northern and southern annular modes,
recent years, provide a significant success story is very well captured in current models. These
for climate models. Most current models spon- modes involve north-south displacements of the
taneously generate El Niño–Southern Oscilla- extratropical storm track and have dominated
tion variability, albeit with varying degrees of observed atmospheric circulation trends in re-
realism. Oscillation spatial structure and dura- cent decades. Because of their ability to simu-
tion are impressive in a model subset but with a late annular modes, global climate models do
4
fairly well with interannual variability in polar the strengths and weaknesses of dynamical
regions of both hemispheres. They are less suc- modeling and statistical methods often are
cessful with daily polar-weather variability, al- complementary.
though finer-scale regional simulations do show
promise for improved global-model simulations Regional trends in extreme events are not al-
as their resolution increases. ways captured by current models, but it is diffi-
cult to assess the significance of these
discrepancies and to distinguish between model
H ow well do climate models simulate deficiencies and natural variability.
regional climate variability and change?
The use of climate model results to assess eco-
nomic, social, and environmental impacts is be-
Chapter 3 describes techniques to downscale coming more sophisticated, albeit slowly.
coarse-resolution global climate model output Simple methods requiring only mean changes
to higher resolution for regional applications. in temperature and precipitation to estimate im-
These downscaling methodologies fall prima- pacts remain popular, but an increasing number
rily into two categories. In the first, a higher- of studies are using more detailed information
resolution, limited-area numerical such as the entire distribution of daily or
meteorological model is driven by global cli- monthly values and extreme outcomes. The
mate model output at its lateral boundaries. mismatch between models’ spatial resolution vs
These dynamical downscaling strategies are the scale of impact-relevant climate features and
beneficial when supplied with appropriate sea- of impact models remains an impediment for
surface and atmospheric boundary conditions, certain applications. Chapter 7 provides several
but their value is limited by uncertainties in in- examples of applications using climate model
formation supplied by global models. Given the results and downscaling techniques.
value of multimodel ensembles for larger-scale
climate prediction, coordinated downscaling
clearly must be performed with a representative W hat are the tradeoffs to be made in fur-
set of global model simulations as input, rather ther climate model development (e.g.,
than focusing on results from one or two mod- between increasing spatial/temporal res-
els. Relatively few such multimodel dynami- olution and representing additional
cal downscaling studies have been performed physical/biological processes)?
to date.
In the second category, empirical relationships Chapter 6 is devoted to trends in climate model
between large- and small-scale observations are development. With increasing computer power
developed, then applied to global climate model and observational understanding, future models
output to provide regional detail. Statistical will include both higher resolution and more
techniques to produce appropriate small-scale processes.
structures from climate simulations are referred
to as “statistical downscaling.” They can be as Resolution increases most certainly will lead to
effective as high-resolution numerical simula- improved representations of atmospheric and
tions in providing climate change information oceanic general circulations. Ocean components
to regions unresolved by most current global of current climate models do not directly simu-
models. Because of the computational effi- late the oceans’ very energetic motions referred
ciency of these techniques, they can much more to as “mesoscale eddies.” Simulation of these
easily utilize a full suite of multimodel ensem- small-scale flow patterns requires horizontal
bles. The statistical methods, however, are com- grid sizes of 10 km or smaller. Current oceanic
pletely dependent on the accuracy of regional components of climate models are effectively
circulation patterns produced by global models. laminar rather than turbulent, and the effects of
Dynamical models, through higher resolution these eddies must be approximated by imper-
or better representation of important physical fect theories. As computer power increases, new
processes, often can improve the physical re- models that resolve these eddies will be incor-
alism of simulated regional circulation. Thus, porated into climate models to explore their im-
5
pact on decadal variability as well as heat and Inclusion of carbon-cycle processes and other
carbon uptake. Similarly, atmospheric general biogeochemical cycles is required to transform
circulation models will evolve to “cloud-re- physical climate models into full Earth system
solving models” (CRMs) with spatial resolu- models that incorporate feedbacks influencing
tions of less than a few kilometers. The hope is greenhouse gas and aerosol concentrations in
that CRMs will provide better results through the atmosphere. Land models that predict veg-
explicit simulation of many cloud properties etation patterns are being developed actively,
now poorly represented on subgrid scales of but the demands of these models on the quality
current atmospheric models. CRMs are not new of simulated precipitation patterns ensures that
frameworks but rather are based on models de- their evolution will be gradual and tied to im-
signed for mesoscale storm and cumulus con- provements in the simulation of regional cli-
vection simulations. mate. Uncertainties about carbon-feedback
processes in the ocean as well as on land, how-
Models of glacial ice are in their infancy. Gla- ever, must be reduced for more reliable future
cial models directly coupled to atmosphere- estimates of climate change.
ocean models typically account for only direct
melting and accumulation at the surface of ice
sheets and not the dynamic discharge due to gla-
cial flow. More-detailed current models typi-
cally generate discharges that change only over
centuries and millennia. Recent evidence for
rapid variations in this glacial outflow indicates
that more-realistic glacial models are needed to
estimate the evolution of future sea level.
6
1CH APTER
I ntroduction
The use of computers to simulate complex systems has grown in the past few decades to play a
central role in many areas of science. Climate modeling is one of the best examples of this trend
and one of the great success stories of scientific simulation. Building a laborator y analog of the
Earth’s climate system with all its complexity is impossible. Instead, the successes of climate mod-
eling allow us to address many questions about climate by experimenting with simulations— that
is, with mathematical models of the climate system. D espite the success of the climate modeling
enterprise, the complexity of our Earth imposes important limitations on existing climate mod-
els. This report aims to help the reader understand the valid uses, as well as the limitations, of cur-
rent climate models.
Climate modeling and forecasting grew from Niño variability or extratropical storms or At-
the desire to predict weather. The distinction be- lantic hurricanes, with an eye toward assessing
tween climate and weather is not precise. Oper- the ability of models to predict how variability
ational weather forecasting has focused might change as the climate evolves in coming
historically on time scales of a few days but decades and centuries.
more recently has been extended to months and
seasons in attempts to predict the evolution of El An important constraint on climate models not
Niño episodes. The goal of climate modeling imposed on weather-forecast models is the re-
can be thought of as the extension of forecasting quirement that the global system precisely and
to longer and longer time periods. The focus is accurately maintain the global energy balance
not on individual weather events, which are un- over very long periods of time. The Earth’s en-
predictable on long time scales, but on the sta- ergy balance (or “budget”) is defined as the dif-
tistics of these events and on the slow evolution ference between absorbed solar energy and
of oceans and ice sheets. Whether the forecast- emitted infrared radiation to space. It is affected
ing of individual El Niño episodes is considered by many factors, including the accumulation of
weather or climate is a matter of convention. For greenhouse gases, such as carbon dioxide, in the
the purpose of this report, we will consider El atmosphere. The decades-to-century changes in
Niño forecasting as weather and will not ad- the Earth’s energy budget, manifested as climate
dress it directly. On the climate side we are con- changes, are just a few percent of the average
cerned, for example, with the ability of models values of that budget’s largest terms. Many de-
to simulate the statistical characteristics of El cisions about model construction described in
7
The U.S. Climate Change Science Program Chapter 1 - Introduction
Chapter 2 are based on the need to properly came one of the most vigorous and longest-
and accurately simulate the long-term energy lived GCM development programs at the
balance. National Oceanic and Atmospheric Administra-
tion’s Geophysical Fluid Dynamics Laboratory
This report will focus primarily on comprehen- (GFDL) at Princeton University. The University
sive physical climate models used for the most of California at Los Angeles began producing
recent international Coupled Model Intercom- atmospheric general circulation models
parison Project (CMIP) coordinated experi- (AGCMs) beginning in 1961 under the leader-
ments (Meehl et al. 2006) sponsored by the ship of Yale Mintz and Akio Arakawa. This pro-
World Climate Research Programme (WCRP). gram influenced others in the 1960s and 1970s,
These coupled atmosphere-ocean general cir- leading to modeling programs found today at
culation models (AOGCMs) incorporate de- National Aeronautics and Space Administration
tailed representations of the atmosphere, land (NASA) laboratories and several universities.
surface, oceans, and sea ice. Where practical, At Lawrence Livermore National Laboratory,
we will emphasize and highlight results from Cecil E. Leith developed an early AGCM in
the three U.S. modeling projects that partici- 1964. The U.S. National Center for Atmospheric
pated in the CMIP experiments. Additionally, Research (NCAR) initiated AGCM develop-
this report examines the use of regional climate ment in 1964 under Akira Kasahara and Warren
models (RCMs) for obtaining higher-resolution Washington. Leith moved to NCAR in the late
details from AOGCM simulations over smaller 1960s and, in the early 1980s, oversaw con-
regions. Still, other types of climate models are struction of the Community Climate Model, a
being developed and applied to climate simula- predecessor to the present Community Climate
tion. The more-complete Earth system models, System Model (CCSM).
which build carbon-cycle and ecosystem
processes on top of AOGCMs, are used prima- Early weather models focused on fluid dynam-
rily for studies of future climate change and pa- ics rather than on radiative transfer and the at-
leoclimatology, neither of which is directly mosphere’s energy budget, which are centrally
relevant to this report. Another class of models important for climate simulations. Additions to
not discussed here but used extensively, partic- the original AGCMs used for weather analysis
ularly when computer resources are limited, is and prediction were needed to make climate
Earth system models of intermediate complex- simulations possible. Furthermore, because cli-
ity (EMICs). Although these models have many mate simulation focuses on time scales longer
more assumptions and simplifications than are than a season, oceans and sea ice must be in-
found in CMIP models (Claussen et al. 2002), cluded in the modeling system in addition to the
they are particularly useful in exploring a wide more rapidly evolving atmosphere. Thus, ocean
range of mechanisms and obtaining broad esti- and ice models have been coupled with atmos-
mates of future climate change projections that pheric models. The first ocean GCMs were de-
can be further refined with AOGCM experi- veloped at GFDL by Bryan and Cox in the
ments. 1960s and then coupled with the atmosphere by
Manabe and Bryan in the 1970s. Paralleling
1.1 BRIEF H ISTORY OF CLIMAT E events in the United States, the 1960s and 1970s
MODEL DEVELOPMEN T also were a period of climate- and weather-
model development throughout the world, with
As numerical weather prediction was develop- major centers emerging in Europe and Asia.
ing in the 1950s as one of the first computer ap- Representatives of these groups gathered in
plications, the possibility of also using Stockholm in August 1974, under the sponsor-
numerical simulation to study climate became ship of the Global Atmospheric Research Pro-
evident almost immediately. The feasibility of gramme to produce a seminal treatise on
generating stable integrations of atmospheric climate modeling (GARP 1975). This meeting
equations for arbitrarily long time periods was established collaborations that still promote in-
demonstrated by Norman Phillips in 1956. ternational cooperation today.
About that time, Joseph Smagorinsky started a
program in climate modeling that ultimately be-
8
The use of climate models in research on car- search Program (USGCRP), established in
bon dioxide and climate began in the early 1989, designated climate modeling and predic-
1970s. The important study, “Inadvertent Cli- tion as one of the four high-priority integrating
mate Modification” (SMIC 1971), endorsed the themes of the program (Our Changing Planet
use of GCM-based climate models to study the 1991). The combination of steadily increasing
possibility of anthropogenic climate change. computer power and research spurred by WCRP
With continued improvements in both climate and USGCRP has led to a steady improvement
observations and computer power, modeling in the completeness, accuracy, and resolution of
groups furthered their models through steady AOGCMS for climate simulation and predic-
but incremental improvements. By the tion. An often-used illustration from the Third
late1980s, several national and international or- IPCC Working Group 1 Scientific Assessment
ganizations formed to assess and expand scien- of Climate Change in 2001 depicts this evolu-
tific research related to global climate change. tion (see Fig. 1.1). Even more comprehensive
These developments spurred interest in acceler- climate models produced a series of coordinated
ating the development of improved climate numerical simulations for the third international
models. The primary focus of Working Group 1 Climate Model Intercomparison Project
of the United Nations Intergovernmental Panel (CMIP3), which were used extensively in re-
on Climate Change (IPCC), which began in search cited in the recent Fourth IPCC Assess-
1988, was the scientific inquiry into physical ment (IPCC 2007). Contributions came from
processes governing climate change. IPCC’s three groups in the United States (GFDL,
first Scientific Assessment (IPCC 1990) stated, NCAR, and the NASA Goddard Institute for
“Improved prediction of climate change de- Space Studies) and others in the United King-
pends on the development of climate models, dom, Germany, France, Japan, Australia,
which is the objective of the climate modeling Canada, Russia, China, Korea, and Norway.
programme of the World Climate Research Pro-
gramme.” The United States Global Change Re-
Development of Climate Models: Past, Present, and Future Figure 1.1. H istorical
Mid-1970s Mid-1980s Early 1990s Late 1990s Present Day Early 2000s?
Development of
Climate Models.
Atmosphere Atmosphere Atmosphere Atmosphere Atmosphere Atmosphere [Figure source: Climate
Change 2 0 0 1 :The Scientific
Land surface Land surface Land surface Land surface Land surface
Basis, Contribution of
Ocean and sea ice Ocean and sea ice Ocean and sea ice Ocean and sea ice W orking Group 1 to the
Assessment Report of the
Sulphate aerosol Sulphate aerosol Sulphate aerosol Intergovernmental Panel on
Nonsulphate Nonsulphate Climate Change, p. 48.
aerosol aerosol Used with permission
Carbon cycle Carbon cycle from IPCC .]
Dynamic
vegetation
Atmospheric
Ocean and Sulphur Nonsulphate chemistry
sea ice model cycle model aerosols
Land carbon
cycle model Carbon
Ocean carbon cycle model
cycle model
Dynamic Dynamic
vegetation vegetation
Atmospheric Atmospheric Atmospheric

chemistry chemistry chemistry
Adapted from IPCC 2001
9
1.2 CLIMAT E MODEL processes and their interactions. These mathe-

CON ST RU CT ION matical models are then translated into com-
puter codes executed on some of the most
Comprehensive climate models are constructed powerful computers in the world. Available
using expert judgments to satisfy many con- computer power helps determine the types of
straints and requirements. Overarching consid- approximations required. As a general rule,
erations are the accurate simulation of the most growth of computational resources allows mod-
important climate features and the scientific un- elers to formulate algorithms less dependent on
derstanding of the processes that control these approximations known to have limitations,
features. Typically, the basic requirement is that thereby producing simulations more solidly
models should simulate features important to founded on established physical principles.
humans, particularly surface variables such as These approximations are most often found in
temperature, precipitation, windiness, and “closure” or “parameterization” schemes that
storminess. This is a less-straightforward re- take into account unresolved motions and
quirement than it seems because a physically processes and are always required because cli-
based climate model also must simulate all mate simulations must be designed so they can
complex interactions in the coupled atmos- be completed and analyzed by scientists in a
phere–ocean–land surface–ice system mani- timely manner, even if run on the most power-
fested as climate variables of interest. For ful computers.
example, jet streams at altitudes of 10 km above
the surface must be simulated accurately if Climate models have shown steady improve-
models are to generate midlatitude weather with ment over time as computer power has in-
realistic characteristics. Midlatitude highs and creased, our understanding of physical
lows shown on surface weather maps are inti- processes of climatic relevance has grown,
mately associated with these high-altitude wind datasets useful for model evaluation have been
patterns. As another example, the basic temper- developed, and our computational algorithms
ature decrease from the equator to the poles can- have improved. Figure 1.2 shows one attempt at
not be simulated without taking into account the quantifying this change. It compares a particu-
poleward transport of heat in the oceans, some lar metric of climate model performance among
of this heat being carried by currents 2 or 3 km the CMIP1 (1995), CMIP2 (1997), and CMIP3
deep into the ocean interior. Thus, comprehen- (2004) ensembles of AOGCMs. This particular
sive models should produce correctly not just metric assesses model performance in simulat-
the means of variables of interest but also the ing the mean climate of the late 20th Century as
extremes and other measures of natural vari- measured by a basket of indicators focusing on
ability. Finally, our models should be capable of aspects of atmospheric climate for which ob-
simulating changes in statistics caused by rela- servational counterparts are deemed adequate.
tively small changes in the Earth’s energy Model ranking according to individual mem-
budget that result from natural and human ac- bers of this basket of indicators varies greatly, so
tions. this aggregate ranking depends on how different
indicators are weighted in relative importance.
Climate processes operate on time scales rang- Nevertheless, the conclusion that models have
ing from several hours to millennia and on spa- improved over time is not dependent on the rel-
tial scales ranging from a few centimeters to ative weighting factors, as nearly all models
thousands of kilometers. Principles of scale have improved in most respects. The construc-
analysis, fluid dynamical filtering, and numer- tion of metrics for evaluating climate models is
ical analysis are used for intelligent compro- itself a subject of intensive research and will be
mises and approximations to make possible the covered in more detail in Chapter 2.
formulation of mathematical representations of
10
18 15 10 12 4 14 7 5 11
CMIP-1
6 2 3 17 8 1 9 13 16
q e od ka j h m
CMIP-2
p l b c i f n g
LX W U C K IJ F OP N
CMIP-3
S T G H YDQ R M V
B
REA
X YU CD HJ G OV I
PICtrl
S LW T K FQR M P N
0.4 0.5 1 2 3 4 5 6 7
I2
Figure 1.2. Performance Index I2 for Individual Models (circles) and Model
Generations (rows).
Best performing models have low I2 values and are located toward the left. Circle sizes indicate the length
of the 95% confidence inter vals. Letters and numbers identify individual models; flux corrected models are
labeled in red. Grey circles show the average I2 of all models within one model group. Black circles indicate
the I2 of the multimodel mean taken over one model group. The green circle (REA) corresponds to the I2
of the N CEP/N C AR Reanalysis (Kalnay et al. 1996), conducted by the N ational W eather Ser vice’s N ational
Centers for Environmental Prediction and the N ational Center for Atmospheric Research. Last row
(PICTRL) shows I2 for the preindustrial control experiment of the CMIP3 project. [Adapted from Fig. 1 in
T. Reichler and J. Kim 2008: H ow well do coupled models simulate today’s climate? Bulletin American
M eteorological Society, 89(3), doi:10.1175/BAMS-89-3-303. Reproduced by permission of the American
Meteorological Society.]
Also shown in Fig. 1.2 is the same metric eval- adaptation or mitigation strategies is to take into
uated from climate simulation results obtained account, in a pertinently informed manner, the
by averaging over all AOGCMs in the CMIP1, products of distinct models built using different
CMIP2, and CMIP3 archives. The CMIP3 “en- expert judgments at centers around the world.
semble-mean” model performs better than any
individual model by this metric and by many 1.3 SU MMARY OF SAP 3.1
others. This kind of result has convinced the CH APT ERS
community of the value of a multimodel ap-
proach to climate change projection. Our un- The remaining sections of this report describe
derstanding of climate is still insufficient to climate model development, evaluation, and ap-
justify proclaiming any one model “best” or plications in more detail. Chapter 2 describes
even showing metrics of model performance the development and construction of models
that imply skill in predicting the future. More and how they are employed for climate research.
appropriate in any assessments focusing on Chapter 3 discusses regional climate models
11
and their use in “downscaling” global model re-

sults to specific geographic regions, particularly
North America. The concept of climate sensi-
tivity—the response of a surface temperature to
a specified change in the energy budget at the
top of the model’s atmosphere—is described in
Chapter 4. A survey of how well important cli-
mate features are simulated by modern models
is found in Chapter 5, while Chapter 6 depicts
near-term development priorities for future
model development. Finally, Chapter 7 illus-
trates a few examples of how climate model
simulations are used for practical applications.
A detailed Reference section follows Chapter 7.
12
2CH APTER
D escription of
Global Climate
Systems Models
M odern climate models are composed of a system of interacting model components, each of
which simulates a different part of the climate system. The individual parts often can be run in-
dependently for certain applications. N early all the Coupled Model Intercomparison Project 3
(CMIP3) class of models include four primary components: atmosphere, land surface, ocean, and
sea ice. The atmospheric and ocean components are known as “general circulation models” or
GCMs because they explicitly simulate the large-scale global circulation of the atmosphere and
ocean. Climate models sometimes are referred to as coupled atmosphere-ocean GCMs.This name
may be misleading because coupled GCMs can be employed to simulate aspects of weather and
ocean dynamics without being able to maintain a realistic climate projection over centuries of sim-
ulated time, as required of a climate model used for studying anthropogenic climate change.W hat
follows in this chapter is a description of a modern climate model’s major components and how
they are coupled and tested for climate simulation.
2.1 AT MOSPH ERIC GEN ERAL and momentum horizontally are relatively well
CIRCU LAT ION MODELS resolved by the grid in current atmospheric
models, but processes that redistribute these
Atmospheric general circulation models quantities vertically have a significant part
(AGCMs) are computer programs that evolve that is controlled by subgrid-scale parameteri-
the atmosphere’s three-dimensional state for- zations.
ward in time. This atmospheric state is de-
scribed by such variables as temperature, The model’s grid-scale evolution is determined
pressure, humidity, winds, and water and ice by equations describing the thermodynamics
condensate in clouds. These variables are de- and fluid dynamics of an ideal gas. The atmos-
fined on a spatial grid, with grid spacing deter- phere is a thin spherical shell of air that en-
mined in large part by available computational velops the Earth. For climate simulation,
resources. Some processes governing this at- emphasis is placed on the atmosphere’s lowest
mospheric state’s evolution are relatively well 20 to 30 km (i.e., the troposphere and the lower
resolved by model grids and some are not. The stratosphere). This layer contains over 95% of
latter are incorporated into models through the atmosphere’s mass and virtually all of its
approximations often referred to as parameter- water vapor, and it produces nearly all weather
izations. Processes that transport heat, water, although current research suggests possible in-
13
The U.S. Climate Change Science Program Chapter 2 - D escription of Global Climate System Models
teractions between this layer and higher atmos- All AGCMs must incorporate the effects of ra-
pheric levels (e.g., Pawson et al. 2000). Because diant-energy transfer. The radiative-transfer
of the disparity between scales of horizontal and code computes the absorption and emission of
vertical motions governing global and regional electromagnetic waves by air molecules and at-
climate, the two motions are treated differently mospheric particles. Atmospheric gases absorb
by model algorithms. The resulting set of equa- and emit radiation in “spectral lines” centered
tions is often referred to as the primitive equa- at discrete wavelengths, but the computational
tions (Haltiner and Williams 1980). costs are too high in a climate model to perform
this calculation for each individual spectral line.
Although nearly all AGCMs use this same set AGCMs use approximations, which differ
of primitive dynamical equations, they use dif- among models, to group bands of wavelengths
ferent numerical algorithms to solve them. In all together in a more efficient calculation. Most
cases, the atmosphere is divided into discrete models have separate radiation codes to treat
vertical layers, which are then overlaid with a solar (visible) radiation and the much-longer-
two-dimensional horizontal grid, producing a wavelength terrestrial (infrared) radiation. Ra-
three-dimensional mesh of grid elements. The diation calculation includes the effects of water
equations are solved as a function of time on vapor, carbon dioxide, ozone, and clouds. Mod-
this mesh. The portion of the model code gov- els used in climate change experiments also in-
erning the fluid dynamics explicitly simulated clude aerosols and additional trace gases such
on this mesh often is referred to as the model’s as methane, nitrous oxide, and the cloroflouro-
“dynamical core.” Even with the same numeri- carbons. Validation of AGCM radiation codes
cal approach, AGCMs differ in spatial resolu- often is done offline (separate from other
tions and configuration of model grids. Some AGCM components) by comparison with line-
models use a “spectral” representation of winds by-line model calculations that, in turn, are
and temperatures, in which these fields are writ- compared against laboratory and field observa-
ten as linear combinations of predefined pat- tions (e.g., Ellingson and Fouquart 1991;
terns on the sphere (spherical harmonics) and Clough, Iacono, and Moncet 1992; Collins et al.
are then mapped to a grid when local values are 2006b).
required. Some models have few layers above
the tropopause (the moving boundary between All GCMs use subgrid-scale parameterizations
the troposphere and stratosphere (e.g., GFDL to simulate processes that are too small or op-
2004)), while others have as many layers above erate on time scales too fast to be resolved on
the troposphere as in it (e.g., Schmidt et al. 2006). the model grid. The most important parameter-
izations are those involving cirrus and stratus
All AGCMs use a coordinate system in which cloud formation and dissipation, cumulus con-
the Earth’s surface is a coordinate surface, sim- vection (thunderstorms and fair-weather cumu-
plifying exchanges of heat, moisture, trace sub- lus clouds), and turbulence and subgrid-scale
stances, and momentum between the Earth’s mixing. For cloud calculations, most AGCMs
surface and the atmosphere. Numerical algo- treat ice and liquid water as atmospheric state
rithms of AGCMs should precisely conserve the variables. Some models also separate cloud par-
atmosphere’s mass and energy. Typical AGCMs ticles into ice crystals, snow, graupel (snow pel-
have spatial resolution of 200 km in the hori- lets), cloud water, and rainwater. Empirical
zontal and 20 levels in the volume below the al- relationships are used to calculate conversions
titude of 15 km. Because numerical errors often among different particle types. Representing
depend on flow patterns, there are no simple these processes on the scale of model grids is
ways to assess the accuracy of numerical dis- particularly difficult and involves calculation of
cretizations in AGCMs. Models use idealized fractional cloud cover within a grid box, which
cases testing the model’s long-term stability and greatly affects radiative transfer and model sen-
efficiency (e.g., Held and Suarez 1994), as well sitivity. Models either predict cloud amounts
as tests focusing on accuracy using short inte- from the instantaneous thermodynamical and
grations (e.g., Polvani, Scott, and Thomas 2005). hydrological state of a grid box or they treat
cloud fraction as a time-evolving model vari-
14
able. In higher-resolution models, one can at- in which the fluxes or second-order moments
tempt to explicitly simulate the size distribution are calculated prognostically (with memory in
of cloud particles and the “habit” or nonspheri- these higher-order moments from one time step
cal shape of ice particles, but no current global to the next). Turbulent fluxes near the surface
AGCMs attempt this. depend on surface conditions such as rough-
ness, soil moisture, and vegetation. In addition,
Cumulus convective transports, which are im- all models use diffusion schemes or dissipative
portant in the atmosphere but cannot be explic- numerical algorithms to simulate kinetic energy
itly resolved at GCM scale, are calculated using dissipation from turbulence far from the surface
convective parameterization algorithms. Most and to damp small-scale unresolved structures
current models use a cumulus mass flux scheme produced from resolved scales by turbulent at-
patterned after that proposed by Arakawa and mospheric flow.
Schubert (1974), in which convection’s upward
motion occurs in very narrow plumes that take The realization that a significant fraction of mo-
up a negligible fraction of a grid box’s area. mentum transfer between atmosphere and sur-
Schemes differ in techniques used to determine face takes place through nonturbulent pressure
the amount of mass flowing through these forces on small-scale “hills” has resulted in a
plumes and the manner in which air is entrained substantial effort to understand and model this
and detrained by the rising plume. Most models transfer (e.g., McFarlane 1987; Kim and Lee
do not calculate separately the area and vertical 2003). This process is often referred to as grav-
velocity of convection but try to predict only the ity wave drag because it is intimately related to
product of mass and area, or convective mass atmospheric wave generation. The variety of
flux. Prediction of convective velocities, how- gravity wave drag parameterizations is a signif-
ever, is needed for new models of interactions icant source of differences in mean wind fields
between aerosols and clouds. Most current generated by AGCMs. Accounting for both sur-
schemes do not account for differences between face-generated and convectively generated grav-
organized mesoscale convective systems and ity waves are difficult aspects of modeling the
simple plumes. The turbulent mixing rate of up- stratosphere and mesosphere (≥ 20 km altitude),
drafts and downdrafts with environments and since winds in those regions are affected
the phase changes of water vapor within con- strongly by transfer of momentum and energy
vective systems are treated with a mix of em- from these unresolved waves.
piricism and constraints based on the moist
thermodynamics of rising air parcels. Some Extensive field programs have been designed to
models also include a separate parameterization evaluate parameterizations in GCMs, ranging
of shallow, nonprecipitating convection (fair- from tests of gravity wave drag schemes
weather cumulus clouds). In short, clouds gen- [Mesoscale Alpine Program (called MAP), e.g.,
erated by cumulus convection in climate models Bougeault et al. 2001] to tests of radiative trans-
should be thought of as based in large part on fer and cloud parameterizations [Atmospheric
empirical relationships. Radiation Measurement Program (called
ARM), Ackerman and Stokes 2003]. Running
All AGCMs parameterize the turbulent trans- an AGCM coupled to a land model as a numer-
port of momentum, moisture, and energy in the ical weather prediction model for a few days—
atmospheric boundary layer near the surface. A starting with best estimates of the atmosphere
long-standing theoretical framework, Monin- and land’s instantaneous state at any given
Obukhov similarity theory, is used to calculate time—is a valuable test of the entire package of
the vertical distribution of turbulent fluxes and atmospheric parameterizations and dynamical
state variables in a thin (typically less than 10 core (e.g., Xie et al. 2004). Atmosphere-land
m) layer of air adjacent to the surface. Above models also are routinely tested by running
the surface layer, turbulent fluxes are calculated them with boundary conditions taken from ob-
based on closure assumptions that provide a served sea-surface temperatures and sea-ice dis-
complete set of equations for subgrid-scale vari- tributions (Gates 1992) and examining the
ations. Closure assumptions differ among resulting climate.
AGCMs; some models use high-order closures
15
2.2 OCEAN GEN ERAL to which they are coupled, typically on the order
CIRCU LAT ION MODELS of 100 km (~ 1º spacing in latitude and longi-
tude) for most of Earth. In many OGCMs the
Ocean general circulation models (OGCMs) north-south resolution is enhanced within 5º lat-
solve the primitive equations for global incom- itude of the equator to improve the ability to
pressible fluid flow analogous to the ideal-gas simulate important equatorial processes.
primitive equations solved by atmospheric OGCM grids usually are designed to avoid co-
GCMs. In climate models, OGCMs are coupled ordinate singularities caused by the convergence
to the atmosphere and ice models through the of meridians at the poles. For example, the
exchange of heat, salinity, and momentum at the CCSM OGCM grid is rotated to place its North
boundary among components. Like the atmos- Pole over a continent, while the GFDL models
phere, the ocean’s horizontal dimensions are use a grid with three poles, all of which are
much larger than its vertical dimension, result- placed over land (Murray 1996). Such a grid re-
ing in separation between processes that control sults in having all ocean grid points at numeri-
horizontal and vertical fluxes. With continents, cally viable locations.
enclosed basins, narrow straits, and submarine
basins and ridges, the ocean has a more com- Processes that control ocean mixing near the
plex three-dimensional boundary than does the surface are complex and take place on small
atmosphere.. Furthermore, the thermodynamics scales (order of centimeters). To parameterize
of sea water is very different from that of air, so turbulent mixing near the surface, the current
an empirical equation of state must be used in generation of OGCMs uses several different ap-
place of the ideal gas law. proaches (Large, McWilliams, and Doney
1994) similar to those developed for atmos-
An important distinction among ocean models pheric near-surface turbulence. Within the
is the choice of vertical discretization. Many ocean’s stratified, adiabatic interior, vertical
models use vertical levels that are fixed dis- mixing takes place on scales from meters to
tances below the surface (Z-level models) based kilometers (Fig. 2.1); the smaller scales also
on the early efforts of Bryan and Cox (1967) must be parameterized in ocean components.
and Bryan (1969a, b). The General Fluid Dy- Ocean mixing contributes to its heat uptake and
namics Laboratory (GFDL) and Community stratification, which in turn affects circulation
Climate System Model (CCSM) ocean compo- patterns over time scales of decades and longer.
nents fall into this category (Griffies et al. 2005; Experts generally feel (e.g., Schopf et al. 2003)
Maltrud et al. 1998). Two Goddard Institute for that subgrid-scale mixing parameterizations in
Space Studies (GISS) models (R and AOM) use OGCMs contribute significantly to uncertainty
a variant of this approach in which mass rather in estimates of the ocean’s contribution to cli-
than height is used as the vertical coordinate mate change.
(Russell, Miller, and Rind 1995; Russell et al.
2000). A more fundamental alternative uses Very energetic eddy motions occur in the ocean
density as a vertical coordinate. Motivating this on the scale of a few tens of kilometers. These
choice is the desire to control as precisely as so-called mesoscale eddies are not present in the
possible the exchange of heat between layers of ocean simulations of CMIP3 climate models.
differing density, which is very small in much of Ocean models used for climate simulation can-
the ocean yet centrally important for simulation not afford the computational cost of explicitly
of climate. The GISS EH model utilizes a hy- resolving ocean mesoscale eddies. Instead, they
brid scheme that transitions from a Z-coordinate must parameterize mixing by the eddies. Treat-
near the surface to density layers in the ocean ment of these mesoscale eddy effects is an im-
interior (Sun and Bleck 2001; Bleck 2002; Sun portant factor distinguishing one ocean model
and Hansen 2003). from another. Most real ocean mixing is along
rather than across surfaces of constant density.
Horizontal grids used by most ocean models in Development of parameterizations that account
the CMIP3 archive are comparable to or some- for this essential feature of mesoscale eddy mix-
what finer than grids in the atmospheric models ing (Gent and McWilliams 1990; Griffies 1998)
16
Heat
T+ T T–
Figure 2.1. Schematic Showing Interaction of a W ell-
Mixed Surface Layer with Stratified Interior in a Region
with a Strong Temperature Gradient.
Mixing (dashed lines) is occurring both across temperature (T)
gradients and along the temperature gradient with increasing depth.
This process is poorly observed and not well understood. It must be
parameterized in large-scale models. [Adapted from Fig. 1, p. 18, in
Coupling Process and M odel Studies of Ocean M ixing to Improve Climate
M odels— A Pilot Climate Process M odeling and Science Team , a U.S.
CLIVAR white paper by Schopf et al. (2003). Figure originated by John
Marshall, Massachusetts Institute of Technology.]
is a major advance in recent ocean and climate quires models to perform ad hoc exchanges of
modeling. Inclusion of higher-resolution, water between the isolated seas and the ocean
mesoscale eddy–resolving ocean models in fu- to simulate what in nature involves a channel or
ture climate models would reduce uncertainties strait. (The Strait of Gibraltar is an excellent ex-
associated with these parameterizations. ample.) Various modeling groups have chosen
different methods to handle water mixing be-
Other mixing processes that may be important tween smaller seas and larger ocean basins.
in the ocean include tidal mixing and turbulence
generated by interactions with the ocean’s bot- OGCM components of climate models are often
tom, both of which are included in some mod- evaluated in isolation—analogous to the evalu-
els. Lee, Rosati, and Spellman (2006) describe ation of AGCMs with prescribed ocean and sea-
some effects of tidal mixing in a climate model. ice boundary conditions—in addition to being
Some OGCMs also explicitly treat the bottom evaluated as components of fully coupled
boundary and sill overflows (Beckman and ocean-atmosphere GCMs. (Results of full
Dosher 1997; Roberts and Wood 1997; Griffies AOGCM evaluation are discussed in Chapter
et al. 2005). Furthermore, sunlight penetration 5.) Evaluation of ocean models in isolation re-
into the ocean is controlled by chlorophyll dis- quires input of boundary conditions at the air-
tributions (e.g., Paulson and Simpson 1977; sea interface. To compare simulations with
Morel and Antoine 1994; Ohlmann 2003), and observed data, boundary conditions or surface
the depth of penetration can affect surface tem- forcing are from the same period as the data.
peratures. All U.S. CMIP3 models include some These surface fluxes also have uncertainties
treatment of this effect, but they prescribe rather and, as a result, the evaluation of OGCMs with
than attempt to simulate the upper ocean biol- specified sea-surface boundary conditions must
ogy controlling water opacity. Finally, the in- take these uncertainties into account.
clusion of fresh water input by rivers is essential
to close the global hydrological cycle; it affects 2.3 LAN D-SU RFACE MODELS
ocean mixing locally and is handled by models
in a variety of ways. Interaction of Earth’s surface with its atmos-
phere is an integral aspect of the climate sys-
The relatively crude resolution of OGCMs used tem. Exchanges (fluxes) of mass and energy,
in climate models results in isolation of the water vapor, and momentum occur at the inter-
smaller seas from large ocean basins. This re- face. Feedbacks between atmosphere and sur-
17
face affecting these fluxes have important ef- Although these developments have increased
fects on the climate system (Seneviratne et al. the physical basis of land modeling, greater
2006). Modeling the processes taking place over complexity has at times contributed to more dif-
land is particularly challenging because the land ferences among climate models (Gates et al.
surface is very heterogeneous and biological 1999). However, the advent of systematic pro-
mechanisms in plants are important. Climate grams comparing land models, such as the Proj-
model simulations are very sensitive to the ect for Intercomparison of Land Surface
choice of land models (Irannejad, Henderson- Parameterization Schemes (PILPS, Henderson-
Sellers, and Sharmeen 2003). Sellers et al. 1995; Henderson-Sellers 2006) has
led gradually to more agreement with observa-
In the earliest global climate models, land-sur- tions and among land models (Overgaard, Ros-
face modeling occurred in large measure to pro- bjerg, and Butts 2006), in part because
vide a lower boundary to the atmosphere that additional observations have been used to con-
was consistent with energy, momentum, and strain their behavior. However, choices for
moisture balances (e.g., Manabe 1969). The adding processes and increasing realism have
land surface was represented by a balance varied among land-surface models (e.g., Ran-
among incoming and outgoing energy fluxes dall et al. 2007).
and a “bucket” that received precipitation from
the atmosphere and evaporated moisture into Figure 2.2 shows schematically the types of
the atmosphere, with a portion of the bucket’s physical processes included in typical land
water draining away from the model as a type models. Note that the schematic in the figure
of runoff. The bucket’s depth equaled soil field describes a land model used for both weather
capacity. Little attention was paid to the detailed forecasting and climate simulation, an indica-
set of biological, chemical, and physical tion of the increasing sophistication demanded
processes linked together in the climate system’s by both. The figure also hints at important bio-
terrestrial portion. From this simple starting physical and biogeochemical processes that
point, land surface modeling for climate simu- gradually have been added and continue to be
lation has increased markedly in sophistication, added to land models used for climate simula-
with increasing realism and inclusiveness of ter- tion, such as biophysical controls on transpira-
restrial surface and subsurface processes. tion and carbon uptake.
Figure 2.2. Schematic

of Physical Processes in
a Contemporary Land
Model.
[Adapted from Fig. 6 in F.
Chen and J. D udhia 2001:
Coupling an advanced land
surface–hydrology model
with the Penn State–N C AR
MM5 modeling system. Part I:
Model implementation and
sensitivity, M onthly W eather
Review, 129, 569–585.
Reproduced by permission of
the American Meteorological
Society.]
18
Some of the most extensive increases in complete though melt (e.g., Dickinson, Henderson-
plexity and sophistication have occurred with Sellers, and Kennedy 1993). Some recent land
vegetation modeling in land models. An early models for climate simulation include subgrid
generation of land models (Wilson et al. 1987; distributions of snow depth (Liston 2004) and
Sellers et al. 1986) introduced biophysical con- blowing (Essery and Pomeroy 2004). Snow
trols on plant transpiration by adding a vegeta- models now may use multiple layers to repre-
tion canopy over the surface, thereby sent fluxes through the snow (Oleson et al.
implementing vegetative control on the terres- 2004). Effort also has gone into including and
trial water cycle. These models included ex- improving effects of soil freezing and thawing
changes of energy and moisture among the (Koren et al. 1999; Boone et al. 2000; Warrach,
surface, canopy, and atmosphere, along with Mengelkamp, and Raschke 2001; Li and Koike
momentum loss to the surface. Further devel- 2003; Boisserie et al. 2006), although per-
opments included improved plant physiology mafrost modeling is more limited (Malevsky-
that allowed simulation of carbon dioxide fluxes Malevich et al. 1999; Yamaguchi, Noda, and
(e.g., Bonan 1995; Sellers et al. 1996). This Kitoh 2005).
method lets the model treat the flow of water
and carbon dioxide as an optimization problem, Vegetation interacts with snow by covering it,
balancing carbon uptake for photosynthesis thereby masking snow’s higher albedo (Betts
against water loss through transpiration. Im- and Ball 1997) and retarding spring snowmelt
provements also included implementation of (Sturm et al. 2005). The net effect is to main-
model parameters that could be calibrated with tain warmer temperatures than would occur
satellite observation (Sellers et al. 1996), without vegetation masking (Bonan, Pollard,
thereby allowing global-scale calibration. and Thompson 1992). Vegetation also traps
drifting snow (Sturm et al. 2001), insulating the
Continued development has included more re- soil from subfreezing winter air temperatures
alistic parameterization of roots (Arora and and potentially increasing nutrient release and
Boer 2003; Kleidon 2004) and the addition of enhancing vegetation growth (Sturm et al.
multiple canopy layers (e.g., Gu et al. 1999; 2001). Albedo masking is included in some
Baldocchi and Harley 1995; Wilson et al. 2003). land-surface models, but it requires accurate
The latter method, however, has not been used simulations of snow depth to produce accurate
in climate models because the added complex- simulation of surface-atmosphere energy ex-
ity of multicanopy models renders unambigu- changes (Strack, Pielke, and Adegoke 2003).
ous calibration very difficult. An important
ongoing advance is the incorporation of biolog- Time-evolving ice sheets and mountain glaciers
ical processes that produce carbon sources and are not included in most climate models. Ice
sinks through vegetation growth and decay and sheets once were thought to be too sluggish to
the cycling of carbon in the soil (e.g., Li et al. respond to climate change in less than a century.
2006), although considerable work is needed to However, observations via satellite altimetry,
determine observed magnitudes of carbon up- synthetic aperture radar interferometry, and
take and depletion. gravimetry all suggest rapid dynamic variability
of ice sheets, possibly in response to climatic
Most land models assume soil with properties warming (Rignot and Kanagaratnam 2006;
that correspond to inorganic soils, generally Velicogna and Wahr 2006). Most global climate
consistent with mixtures of loam, sand, and clay. models to date have been run with prescribed,
High-latitude regions, however, may have ex- immovable ice sheets. Several modeling groups
tensive zones of organic soils (peat bogs), and are now experimenting with the incorporation
some models have included organic soils topped of dynamic ice sheet models. Substantial phys-
by mosses, which has led to decreased soil heat ical, numerical, and computational improve-
flux and increased surface-sensible and latent- ments, however, are needed to provide reliable
heat fluxes (Beringer et al. 2001). projections of 21st Century ice sheet changes.
Among major challenges are incorporation of a
Climate models initially treated snow as a single unified treatment of stresses within ice sheets,
layer that could grow through snowfall or de- improved methods of downscaling atmospheric
19
fields to the finer ice sheet grid, realistic para- with patches of different land-use and vegeta-
meterizations of surface and subglacial hydrol- tion types. Although these patches may not in-
ogy (fast dynamic processes controlled largely teract directly with their neighbors, they are
by water pressure and extent at the base of the linked by their coupling to the grid box’s at-
ice sheet), and models of ice shelf interactions mospheric column. This coupling does not
with ocean circulation. Ocean models, which allow for possible small-scale circulations that
usually assume fixed topography, may need to might occur because of differences in surface-
be modified to include flow beneath advancing atmosphere energy exchanges among patches
and retreating ice. Meeting these challenges will (Segal and Arritt 1992; Segal et al. 1997). Under
require increased interaction between the most conditions, however, the imprint of such
glaciological and climate modeling communi- spatial heterogeneity on the overlying atmos-
ties, which until recently have been largely iso- pheric column appears to be limited to a few
lated from each another. meters above the surface (e.g., Gutowski, Ötles,
and Chen 1998).
The initial focus of land models was vertical
coupling of the surface with the overlying at- Vertical fluxes linking the surface, canopy, and
mosphere. However, horizontal water flow near-surface atmosphere generally assume some
through river routing has been available in some form of down-gradient diffusion, although
models for some time (e.g., Sausen, Schubert, counter-gradient fluxes can exist in this region
and Dümenil et al. 1994; Hagemann and Dü- much as in the overlying atmospheric boundary
menil 1998), with spatial resolution of routing layer. Some attempts have been made to replace
in climate models increasing in more recent ver- diffusion with more advanced Lagrangian ran-
sions (Ducharne et al. 2003). Freezing soil dom-walk approaches (Gu et al. 1999; Baldoc-
poses additional challenges for modeling runoff chi and Harley 1995; Wilson et al. 2003).
(Pitman et al. 1999), with more recent work
showing some skill in representing its effects Topographic variation within a grid box usually
(Luo et al. 2003; Rawlins et al. 2003; Niu and is ignored in land modeling. Nevertheless, im-
Yang 2006). plementing detailed river-routing schemes re-
quires accurate digital elevation models (e.g.,
Work also is under way to couple groundwater Hirano, Welch, and Lang 2003; Saraf et al.
models into land models (e.g., Gutowski et al. 2005). In addition, some soil water schemes in-
2002; York et al. 2002; Liang, Xie, and Huang clude effects of land slope on water distribution
2003; Maxwell and Miller 2005; Yeh and Eltahir (Choi, Kumar, and Liang 2007) and surface ra-
2005). Groundwater potentially introduces diative fluxes (Zhang et al. 2006).
longer time scales of interaction in the climate
system in places where it has contact with veg- Validation of land models, especially globally,
etation roots or emerges through the surface. remains a problem due to lack of measurements
for relevant quantities such as soil moisture and
Land models encompass spatial scales ranging energy, momentum, moisture flux, and carbon
from model grid-box size down to biophysical flux. The PILPS project (Henderson-Sellers et
and turbulence processes operating on scales al. 1995) has allowed detailed comparisons of
the size of leaves. Explicit representation of all multiple models with observations at points
these scales in a climate model is beyond the around the world having different climates, thus
scope of current computing systems and the ob- providing some constraint on the behavior of
serving systems that would be needed to pro- land models. Global participation in PILPS has
vide adequate model calibration for global and led to more understanding of differences among
regional climate. Model fluxes do not represent schemes and improvements. Compared to pre-
a single point but rather the behavior in a grid vious generations, the latest land surface mod-
box that may be many tens or hundreds of kilo- els exhibit relatively smaller differences from
meters across. Initially, these grid boxes were current observation-based estimates of the
treated as homogeneous units but, starting with global distribution of surface fluxes, but the re-
the pioneering work of Avissar and Pielke liability of such estimates remains elusive (Hen-
(1989), many land models have tiled a grid box derson-Sellers et al. 2003). River routing can
20
provide a diagnosis vs observations of a land of ice models. The EVP method explicitly
model’s spatially distributed behavior (Kattsov solves for the ice-stress tensor, while the VP so-
et al. 2000). Remote sensing has been useful for lution uses an implicit iterative approach. As ex-
calibrating models developed to exploit it but amples, the GFDL models (Delworth et al. 2006
generally has not been used for model valida- ) and Community Climate System Model, Ver-
tion. Regional observing networks that aspire to sion 3 (CCSM3) (Collins et al. 2006a) use the
give Earth system observations, such as some EVP rheology, while the GISS models use the
mesonets in the United States, offer promise of VP implementation.
data from spatially distributed observations of
important fields for land models. The thermodynamic portions of sea ice models
also vary. Earlier generations of climate models
Land modeling has developed in other disci- generally used the sea ice thermodynamics of
plines roughly concurrently with advances in Semtner (1976), which includes one snow layer
climate models. Applications are wide ranging and two ice layers with constant heat conduc-
and include detailed models used for planning tivities together with a simple parameterization
water resources (Andersson et al. 2006), man- of brine (salt) content. The GFDL climate mod-
aging ecosystems (e.g., Tenhunen et al. 1999), els continue to use this but also include the in-
estimating crop yields (e.g., Jones and Kiniry teractions between brine content and heat
1986; Hoogenboom, Jones, and Boote 1992), capacity (Winton 2000). The CCSM3 and GISS
simulating ice-sheet behavior (Peltier 2004), models use variations (Bitz and Lipscomb 1999,
and projecting land use such as transportation Briegleb et al. 2002) incorporating additional
planning (e.g., Schweitzer 2006). As suggested physical processes within the ice, such as the
by this list, widely disparate applications have melting of internal brine regions. Different
developed from differing scales of interest and models define snow and ice layers and ice cat-
focus. Development in some other applications egories differently, but all include an open water
has informed advances in land models for cli- category. Typically, ice models share the grid
mate simulation, as in representation of vegeta- structure of the underlying ocean model.
tion and hydrologic processes. Because land
models do not include all climate system fea- The albedo (proportion of incident sunlight re-
tures, they can be expected in future to engage flected from a surface) of snow and ice plays a
other disciplines and encompass a wider range significant role in the climate system. Sea-ice
of processes, especially as resolution increases. models parameterize the albedo using expres-
sions based on a mix of radiative transfer the-
2.4 SEA-ICE MODELS ory and empiricism. Figure 2.3 from Curry,
Schramm, and Ebert (1995) illustrates sea-ice
Most climate models include sea-ice compo- system interrelations and how the albedo is a
nents that have both dynamic and thermody- function of snow or ice thickness, ice extent,
namic elements. That is, models include the open water, and surface temperature, and other
physics governing ice movement as well as that factors. Models treat these factors in similar
related to heat and salt transfer within the ice. ways but vary on details. For example, the
While sea ice in the real world appears as ice CCSM3 sea-ice component does not include de-
floes on the scale of meters, in climate models pendence on solar elevation angle (Briegleb et
sea ice is treated as a continuum with an effec- al. 2002), but the GISS model does (Schmidt et
tive large-scale rheology describing the rela- al. 2006). Both models include the contribution
tionship between stress and flow. of melt ponds (Ebert and Curry 1993; Schramm
et al. 1997). The GFDL model follows Briegleb
Rheologies commonly in use are the standard et al. (2002) but accounts for different effects of
Hibler viscous-plastic (VP) rheology (Hibler the different wavelengths comprising sunlight
1979; Zhang and Rothrock 2000) and the more- (Delworth et al. 2006).
complex elastic-viscous-plastic (EVP) rheology
of Hunke and Dukowicz (1997), designed pri-
marily to improve the computational efficiency
21
Figure 2.3. Schematic

Diagram of Sea Ice–
Albedo Feedback
Mechanism. +/–
–
Arrow direction indicates
the interaction direction.
Surface +
The “+” signs indicate
Temperature Lead fraction Melt ponds
positive interaction (i.e.,
increase in the first
quantity leads to increase +
in the second quantity), – – – – –
and the “–” signs indicate
negative interaction (i.e.,
increase in the first +/–
Ice extent Ice thickness Snow cover
quantity leads to decrease
in the second quantity).
The “+/–” signs indicate
either that the interaction
sign is uncertain or that – + + –
the sign changes over the
annual cycle. [From Fig. 6 +
in J.A. Curry, J. Schramm,
and E.E. Ebert 1995: O n Surface
the sea ice albedo climate albedo
feedback mechanism, J.
Climate, 8, 240–247.
Reproduced by permission
of the American
2.5 COMPON EN T COU PLIN G 2.5.1 N OAA GFDL Model-

AN D COU PLED MODEL Development Path
EVALUAT ION
NOAA’s GFDL conducted a thorough restruc-
The climate system’s complexity and our in- turing of its atmospheric and climate models for
ability to resolve all relevant processes in mod- more than 5 years prior to its delivery of mod-
els result in a host of choices for development els to the CMIP3 database in 2004. This was
teams. Differing expertise, experience, and in- done partly in response to the need for modern-
terests result in distinct pathways for each cli- izing software engineering and partly in re-
mate model. While we eventually expect to see sponse to new ideas in modeling the
model convergence forced by increasing in- atmosphere, ocean, and sea ice. Differences be-
sights into the climate system’s workings, we tween the resulting models and the previous
are still far from that limit today in several im- generation of climate models at GFDL are var-
portant areas. Given this level of uncertainty, ied and substantial. Mapping out exactly why
multiple modeling approaches clearly are climate sensitivity and other considerations of
needed. Models vary in details primarily be- climate simulations differ between these two
cause development teams have different ideas generations of models would be very difficult
concerning underlying physical mechanisms and has not been attempted. Unlike the earlier
relevant to the system’s less-understood fea- generation, however, the new models do not use
tures. In the following, we describe some key flux adjustments; some other improvements are
aspects of model development by the three U.S. discussed below.
groups that contributed models to the IPCC
Fourth Assessment (IPCC 2007). Particular The new atmospheric models developed at
focus is on points most relevant for simulating GFDL for global warming studies are referred
the 20th Century global mean temperature and to as AM2.0 and AM2.1 (GFDL Atmospheric
on the model’s climate sensitivity. Model Development Team 2004). Key points of
departure from previous GFDL models are the
adoption of a new numerical core for solving
22
fluid dynamical equations for the atmosphere, tivity range from 4.0 to 4.5 K to between 2.5
the inclusion of liquid and ice concentrations as and 3.0 K, as discussed further in Chapter 4.
prognostic variables, and new parameterizations The change responsible for this reduction was
for moist convection and cloud formation. inclusion of a new model of mixing in the plan-
Much atmospheric development was based on etary boundary near the Earth’s surface. GFDL
running the model over observed sea-surface included the mixing model because it generated
temperature and sea-ice boundary conditions more-realistic boundary-layer depths and near-
from 1980 to 2000, with a focus on both the surface relative humidities. Sensitivity reduc-
mean climate and the atmospheric response to tion resulted from modifications to the
El Niño–Southern Oscillation (ENSO) variabil- low-level cloud field; the size of this reduction
ity in the tropical Pacific. Given the basic model was not anticipated.
configuration, several subgrid closures were
varied to optimize climate features. Modest im- Aerosol distributions used by the model were
provements in the midlatitude wind field were computed offline from the MOZART II model
obtained by adjusting the “orographic gravity as described in Horowitz et al. (2003). No at-
wave drag,” which accounts for the effects of tempt was made to simulate indirect aerosol ef-
force exerted on the atmosphere by unresolved fects (interactions between clouds and aerosols),
topographic features. Substantial improvements as confidence in the schemes tested was
in simulating tropical rainfall and its response deemed insufficient. In 20th Century simula-
to ENSO were the result of parameter opti- tions, solar variations followed the prescription
mization as well, especially the treatment of ver- of Lean, Beer, and Bradley (1995), while vol-
tical transport of horizontal momentum by canic forcing was based on Sato et al. (1993).
moist convection. Stratospheric ozone was prescribed, with the
Southern Hemisphere ozone hole prescribed in
The ocean model chosen for this development is particular, in 20th Century simulations. A new
the latest version of the modular ocean model detailed land-use history provided a time his-
(MOM) developed over several decades at tory of vegetation types.
GFDL. Notable new features in this version are
a grid structure better suited to simulating the Final tuning of the model’s global energy bal-
Arctic Ocean and a framework for subgrid-scale ance, using two parameters in the cloud predic-
mixing that avoids unphysical mixing among tion scheme, was conducted by examining
oceanic layers of differing densities (Gent and control simulations of the fully coupled model
McWilliams 1990; Griffies 1998). A new sea- using fixed 1860 and 1990 forcings (see box,
ice model includes an EVP large-scale effective Tuning the Global Mean Energy Balance). The
rheology that has proven itself in the past resulting model is described in Delworth et al.
decade in several models and multiple ice thick- (2006) and Gnanadesikan et al. (2006). IPCC-
nesses in each grid box. The land model chosen relevant runs of this model (CM2.0) were pro-
is relatively simple, with vertically resolved soil vided to the CMIP3–IPCC archive. Simulations
temperature but retaining the “bucket hydrol- of the 20th Century with time-varying forcings
ogy” from the earlier generation of models. provided to the database and described in Knut-
son et al. (2006) were the first of this kind gen-
The resulting climate model was studied, re- erated with this model. The model was not
structured, and tuned for an extended period, retuned, and no iteration of the aerosol or any
with particular interest in optimizing the struc- other time-varying forcings followed these ini-
ture and frequency of the model’s spontaneously tial simulations.
generated El Niño events, minimizing surface
temperature biases, and maintaining an Atlantic Model development proceeded in the interim,
overturning circulation of sufficient strength. and a new version emerged rather quickly in
During this development phase, climate sensi- which the atmospheric model’s numerical core
tivity was monitored by integrating the model was replaced by a “finite-volume” code (Lin
to equilibrium with doubled CO2 when coupled and Rood 1996). Treatment of wind fields near
to a “flux-adjusted” slab ocean model. A single the surface improved substantially, which in
model modification reduced the model’s sensi- turn resulted in enhanced extratropical ocean
23
Tuning the Global Mean Energy Balance
A procedure common to all comprehensive climate models is tuning the global mean energy bal-
ance. A climate model must be in balance at top of atmosphere (TO A) and globally averaged to
within a few tenths of a W /m2 in its control (pre-1860) climate if it is to avoid temperature drifts
in 20th and 21st centur y simulations that would obscure response to imposed changes in green-
house, aerosol, volcanic, and solar forcings. Especially because of difficulty in modeling clouds but
also even in clear sky, untuned models do not currently possess this level of accuracy in their ra-
diative fluxes. Untuned imbalances more typically range up to 5 W /m2. Parameters in the cloud
scheme are altered to create a balanced state, often taking care that individual components of this
balance— the absorbed solar flux and emitted infrared flux— are individually in agreement with ob-
servations, since these help ensure the correct distribution of heating between atmosphere and
ocean. This occasionally is referred to as “final tuning” the model to distinguish it from various
choices made for other reasons while the model is being configured.
The need for final tuning does not preclude the use of these models for global warming simulations
in which radiative forcing itself is on the order of several W /m2. Consider, for example, the Ra-
maswamy et al. (2001) study on the effects of modifying the “water vapor continuum” treatment in
a climate model.This is an aspect of the radiative transfer algorithm in which there is significant un-
certainty. W hile modifying continuum treatment can change the TO A balance by more than 1 W /m2,
the effect on climate sensitivity is found to be insignificant. The change in radiative transfer in this
instance alters the outgoing infrared flux by roughly 1%, and it affects the sensitivity (by changing
the flux derivative with respect to temperature) by roughly the same percentage.A sensitivity change
of this magnitude, say from 3 K to 3.03 K, is of little consequence given uncertainties in cloud feed-
backs. The strength of temperature-dependent feedbacks, not errors in mean fluxes per se, is of par-
ticular concern in estimating climatic responses.
circulation and temperatures. ENSO variability produce realistic simulations over a wide range
increased in this model to unrealistically large of spatial resolutions, enabling inexpensive sim-
values; however, the ocean code’s efficiency ulations lasting several millennia or detailed
also improved substantially. With retuning of studies of continental-scale dynamics, variabil-
the clouds for global energy balance, the new ity, and climate change. Twenty-six papers doc-
model CM2.1 was deemed to be an improved umenting all aspects of CCSM3 and runs
model over CM2.0 in several respects, warrant- performed with it were published in a special
ing the generation of a new set of database runs. issue of the Journal of Climate 19(11) (June
CM2.1, when run with a slab-ocean model, was 2006). The atmospheric component of CCSM3
found to have somewhat increased sensitivity. is a spectral model. Three different resolutions
However, transient climate sensitivity—the of CCSM3 are supported. The highest resolu-
global mean warming at the time of CO2 dou- tion is the configuration used for climate-
bling in a fully coupled model with 1% a year change simulations, with a T85 grid for
increase in CO2—actually is slightly smaller atmosphere and land and a grid with around 1º
than in CM2.0. Solar, aerosol, volcanic, and resolution for ocean and sea ice but finer merid-
greenhouse gas forcings are identical in the two ional resolution near the equator. The second
models. resolution is a T42 grid for atmosphere and land
with 1º ocean and sea-ice resolution. A lower-
2.5.2 Community Climate System resolution version, designed for paleoclimate
Model-Development Path studies, has T31 resolution for atmosphere and
land and a 3º version of ocean and sea ice.
CCSM3 was released to the climate community
in June 2004. CCSM3 is a coupled climate The new CCSM3 version incorporates several
model with components representing the at- significant improvements in physical parame-
mosphere, ocean, sea ice, and land surface con- terizations. Enhancements in model physics are
nected by a flux coupler. CCSM3 is designed to designed to reduce several systematic biases in
24
mean climate produced by previous CCSM ver- 2.5.3 GISS Development Path
sions. These enhancements include new treat-
ments of cloud processes, aerosol radiative The most recent version of the GISS atmos-
forcing, land-atmosphere fluxes, ocean mixed- pheric GCM, ModelE, resulted from a substan-
layer processes, and sea-ice dynamics. Signifi- tial reworking of the previous version, Model
cant improvements are shown in sea-ice II′. Although model physics has become more
thickness, polar radiation budgets, tropical sea- complex, execution by the user is simplified as
surface temperatures, and cloud radiative ef- a result of modern software engineering and im-
fects. CCSM3 produces stable climate proved model documentation embedded within
simulations of millennial duration without ad the code and accompanying web pages. The
hoc adjustments to fluxes exchanged among model, which can be downloaded from the
component models. Nonetheless, there are still GISS website by outside users, is designed to
systematic biases in ocean-atmosphere fluxes in run on myriad platforms ranging from laptops
coastal regions west of continents, the spectrum to a variety of multiprocessor computers, partly
of ENSO variability, spatial distribution of pre- because of NASA’s rapidly shifting computing
cipitation in tropical oceans, and continental environment. The most recent (post-AR4) ver-
precipitation and surface air temperatures. Work sion can be run on an arbitrarily large number of
is under way to produce the next version of processors.
CCSM, which will reduce these biases further,
and to extend CCSM to a more accurate and Historically, GISS has eschewed flux adjust-
comprehensive model of the complete Earth cli- ment. Nonetheless, the net energy flux at the top
mate system. of atmosphere (TOA) and surface has been re-
duced to near zero by adjusting threshold rela-
CCSM3’s climate sensitivity is weakly depend- tive humidity for water and ice cloud formation,
ent on the resolution used. Equilibrium temper- two parameters that otherwise are weakly con-
ature increase due to doubling carbon dioxide, strained by observations. Near-zero fluxes at
using a slab-ocean model, is 2.71°C, 2.47°C, these levels are necessary to minimize drift of
and 2.32°C, respectively, for the T85, T42, and either the ocean or the coupled climate.
T31 atmosphere resolutions. The transient cli-
mate temperature response to doubling carbon To assess climate-response sensitivity to treat-
dioxide in fully coupled integrations is much ment of the ocean, ModelE has been coupled to
less dependent on resolution, being 1.50°C, a slab-ocean model with prescribed horizontal
1.48°C, and 1.43°C, respectively, for the T85, heat transport, along with two ocean GCMs.
T42, and T31 atmosphere resolutions (Kiehl et One GCM, the Russell ocean (Russell, Miller,
al. 2006). and Rind 1995), has 13 vertical layers and hor-
izontal resolution of 4º latitude by 5º longitude
The following CCSM3 runs were submitted for and is mass conserving (rather than volume
evaluation for the IPCC Fourth Assessment Re- conserving like the GFDL MOM). Alterna-
port and to the Program for Climate Model Di- tively, ModelE is coupled to the Hybrid Coor-
agnosis and Intercomparison (called PCMDI) dinate Ocean Model (HYCOM), an isopycnal
for dissemination to the climate scientific com- model developed originally at the University of
munity: long, present day, and 1870 control Miami (Bleck et al. 1992). HYCOM has 2º lat-
runs; an ensemble of eight 20th Century runs; itude by 2º longitude resolution at the equator,
and smaller ensembles of future scenario runs with latitudinal spacing decreasing poleward
for the A2, A1B, and B1 scenarios and for the with the cosine of latitude. A separate rectilin-
20th Century commitment run where carbon ear grid is used in the Arctic to avoid polar sin-
dioxide levels were kept at their 2000 values. gularity and joins the spherical grid around
The control and 20th Century runs have been 60°N.
documented and analyzed in several papers in
the Journal of Climate special issue, and future Climate sensitivity to CO2 doubling depends
climate change projections using CCSM3 have upon the ocean model due to differences in sea
been documented by Meehl et al. (2006). ice. Climate sensitivity is 2.7°C for the slab-
ocean model and 2.9°C for the Russell ocean
25
GCM (Hansen et al. 2005). As at GFDL and Because of their uniform horizontal coverage,
CCSM, no effort is made to match a particular satellite retrievals are emphasized for model
sensitivity, nor is the sensitivity or forcing ad- evaluation like Earth Radiation Budget Experi-
justed to match 20th Century climate trends ment fluxes at TOA, Microwave Sounding Unit
(Hansen et al. 2007). Aerosol forcing is calcu- channels 2 (troposphere) and 4 (stratosphere)
lated from prescribed concentration, computed temperatures, and International Satellite Cloud
offline by a physical model of the aerosol life Climatology Project (ISCCP) diagnostics. Com-
cycle. In contrast to GFDL and NCAR models, parison to ISCCP is through a special algorithm
ModelE includes a representation of the aerosol that samples GCM output to mimic data collec-
indirect effect. Cloud droplet formation is re- tion by an orbiting satellite. For example, high
lated empirically to the availability of cloud clouds may include contributions from lower
condensation nuclei, which depends upon the levels in both the model and the downward-
prescribed aerosol concentration (Hansen et al. looking satellite instrument. This satellite per-
2005; Menon and Del Genio 2007). spective within the model allows a rigorous
comparison to observations. In addition to satel-
Flexibility is emphasized in model development lite retrievals, some GCM fields like zonal wind
(Schmidt et al. 2006). ModelE is designed for a are compared to in situ observations adjusted by
variety of applications ranging from simulation European Center for Medium Range Weather
of stratospheric dynamics and middle-atmos- Forecasts’ 40-year reanalysis data (Uppala et al.
phere response to solar forcing to projection of 2005). Surface air temperature is taken from the
21st Century trends in surface climate. Horizon- Climate Research Unit gridded global surface
tal resolution typically is 4º latitude by 5º lon- temperature dataset (Jones et al. 1999).
gitude, although twice that resolution is used
more often for studies of cloud processes. The 2.5.4 Common Problems
model top has been raised from 10 mb (as in the
previous Model II') to 0.1 mb, so the top has less The CCSM and GFDL development teams met
influence on stratospheric circulation. Coding several times to compare experiences and dis-
emphasizes “plug-and-play” structure, so the cuss common problems in the two models. A
model can be adapted easily for future needs such subject of considerable discussion and concern
as fully interactive carbon and nitrogen cycles. was the tendency for an overly strong “cold
tongue” to develop in the eastern equatorial Pa-
Model development is devoted to improving the cific Ocean and for associated errors to appear
realism of individual model parameterizations, in the pattern of precipitation in the Inter-Trop-
such as the planetary boundary layer or sea-ice ical Convergence Zone (often referred to as the
dynamics. Because of the variety of applica- “double-ITCZ problem”). Meeting attendees
tions, relatively little emphasis is placed on op- noted that the equilibrium climate sensitivities
timizing the simulation of specific phenomena of the two models to doubled atmospheric car-
such as El Niño or the Atlantic thermohaline bon dioxide (see Chapter 4) had converged from
circulation; as noted above, successful repro- earlier generations in which the NCAR model
duction of one phenomenon usually results in a was on the low end of the canonical sensitivity
suboptimal simulation of another. Nonetheless, range of 1.5 to 4.5 K, while the GFDL model
some effort was made to reduce biases in previ- was near the high end. This convergence in
ous model versions that emerged from the in- global mean sensitivity was considered coinci-
teraction of various model features such as dental because no specific actions were taken to
subtropical low clouds, tropical rainfall, and engineer convergence. It was not accompanied
variability of stratospheric winds. Some model by any noticeable convergence in cloud-feed-
adjustments were structural, as opposed to the back specifics or in the regional temperature
adjustment of a particular parameter—for exchanges that make up global mean values.
ample, introduction of a new planetary bound-
ary layer parameterization that reduced
unrealistic cloud formation in the lowest model
level (Schmidt et al. 2006).
26
2.6 REDU CT IVE VS H OLIST IC starting from some initial condition. They con-
EVALUAT ION OF MODELS sist of rules that generate the state of a variable
(e.g., temperature, wind, water vapor, clouds,
To evaluate models, appreciation of their struc- rainfall rate, water storage in the land, and land-
ture is necessary. For example, discussion of cli- surface temperature) from its preceding state
matic response to increasing greenhouse gases roughly a half-hour earlier. By this process a
is intimately related to the question of how in- model simulates the weather over the Earth. To
frared radiation escaping to space is controlled. change the way the model’s infrared radiation
When summarizing results from climate mod- reacts to increasing temperatures, the rules
els, modelers often speak and think in terms of would have to be modified.
a simple energy balance model in which the
global mean infrared energy escaping to space One goal of climate modeling is to decrease em-
has a simple dependence on global mean sur- piricism and base models as much as possible
face temperature. Water vapor or cloud feed- on well-established physical principles. This
backs often are incorporated into such global goal is pursued primarily by decomposing the
mean energy balance models with simple rela- climate system into a number of relatively sim-
tionships that can be tailored easily to generate ple processes and interactions. Modelers focus
a desired result. In contrast, Fig. 2.4 shows a on rules governing the evolution of these indi-
snapshot at an instant when infrared radiation is vidual processes rather than working with more
escaping to space in the kind of AGCM dis- holistic concepts such as global mean infrared
cussed in this report. Detailed distributions of radiation escaping to space, average summer-
clouds and water vapor simulated by the model time rainfall over Africa, and average winter-
and transported by the model’s evolving wind time surface pressure over the Arctic. These are
fields create complex patterns in space and time all outcomes of the model, determined by the
that, if the simulation is sufficiently realistic, re- set of reductive rules that govern the model’s
semble images seen from satellites viewing evolution.
Earth at infrared wavelengths.
Suppose the topic under study is how ocean
As described above, AGCMs evolve the state of temperatures affect rainfall over Africa. An em-
atmosphere and land system forward in time, pirical statistical model could be developed
Figure 2.4. A Snapshot

in Time of Infrared
Radiation Escaping to
Space in a Version of
Atmospheric Model
AM2 Constructed at
N OAA’s Geophysical
Fluid Dynamics
Laboratory (GFDL
2004).
The largest amount of energy
emitted is in the darkest
areas, and the least is in the
brightest areas. This version of
the atmospheric model has
higher resolution than that
used for simulations in the
CMIP3 archive (50 km rather
than 200 km), but, other than
resolution, it uses the same
numerical algorithm. The
resolution is typical in many
current studies with
atmosphere-only simulations.
27
using observations and standard statistical tech- triaux 2008). Aspects of observed climate that
niques in which the model is tuned to these ob- must be simulated to ensure reliable future pre-
servations. Alternatively, one can use an AGCM dictions are unclear. For example, models that
giving results like those pictured in Fig. 2.4. An simulate the most realistic present-day temper-
AGCM does not deal directly with high-level atures for North America may not generate the
climate output such as African rainfall averaged most reliable projections of future temperature
over some period. Rather, it attempts to simu- changes. Projected climate changes in North
late the climate system’s inner workings or dy- America may depend strongly on temperature
namics at a much finer level of granularity. To changes in the tropical Pacific Ocean and the
the extent that the simulation is successful and manner in which the jet stream responds to
convincing, the model can be analyzed and ma- them. The quality of a model’s simulation of air-
nipulated to uncover the detailed physical mech- sea coupling over the Pacific might be a more
anisms underlying the connection between relevant metric. However, metrics can provide
ocean temperatures and rainfall over Africa. The guidance about overall strengths and weak-
AGCM-simulated connection may or may not nesses of individual models, as well as the gen-
be as good as the fit obtained with the explicitly eral state of modeling.
tuned statistical model, but a reductive model
ideally provides a different level of confidence The use of metrics also can explain why the
in its explanatory and predictive power. See, for “best” climate model cannot be chosen at this
example, Hoerling et al. (2006) for an analysis time. In Fig. 2.5 below, each colored triangle
of African rainfall and ocean temperature rela- represents a different metric for which each
tionships in a set of AGCMs. model was evaluated (e.g., “ts” represents sur-
face temperature). The figure displays the rela-
Our confidence in the explanatory and predictive error value for a variety of metrics for each
tive power of climate models grows with their model, represented by a vertical column above
ability to simulate many climate system features each tick mark on the horizontal axis. Values
simultaneously with the same set of physically less than zero represent a better-than-average
based rules. When a model’s ability to simulate simulation of a particular field measured by the
the evolution of global mean temperature over metric, while values greater than zero show
the 20th Century is evaluated, it is important to models with errors greater than the average. The
try to make this evaluation in the context of the black triangles connected by the dashed line
model’s ability to spontaneously generate El represent the normalized sum from the errors of
Niño variability of the correct frequency and all 23 fields. The models were ranked from left
spatial structure, for example, and to capture the to right based on the value of this total error. As
effects of El Niño on rainfall and clouds. Sim- can be seen, models with the lowest total errors
ulation quality adds confidence in the reductive tend to score better than average in most indi-
rules being used to generate simultaneous sim- vidual metrics but not in all. For an individual
ulation of all these phenomena. application, the model with the lowest total er-
rors may not be the best choice.
A difficulty to which we will return frequently
in this report is that of relating climate-simula- 2.8 CLIMAT E SIMU LAT ION S
tion qualities to a level of confidence in the DISCU SSED IN T H IS REPORT
model’s ability to predict climate change.
Three types of climate simulation discussed in
2.7 U SE OF MODEL MET RICS this report are described below. They differ ac-
cording to which climate-forcing factors are
Recently, objective evaluation has exploded used as model input.
with the wide availability of model simulation
results in the CMIP3 database (Meehl et al. Control runs use constant forcing. The sun’s
2006). One important area of research is in the energy output and the atmospheric concentra-
design of metrics to test the ability of models to tions of carbon dioxide and other gases and
simulate well-observed climate features (Re- aerosols do not change in control runs. As with
ichler and Kim 2008; Gleckler, Taylor, and Dou- other types of climate simulation, day-night and
28
Figure 2.5. Model Metrics for 23 Different Climate Fields.

Values less than 0 indicate an error less than the average C MIP3 model, while values greater than 0 are
more than the average. The black triangles connected by the black line show a total score obtained by
averaging all 23 fields. Each tick mark represents a different model. [Figure adapted from P.J. Gleckler, K.E.
Taylor, and C . D outriaux 2008: Performance metrics for climate models. J. Geophysical Research , 113, D 06104,
doi:10.1029/2007JD 008972. Reproduced by permission of the American Geophysical U nion (AGU ).]
seasonal variations occur, along with internal Idealized climate simulations are aimed at un-
“oscillations” such as ENSO. Other than these derstanding important processes in models and
variations, the control run of a well-behaved cli- in the real world. They include experiments in
mate model is expected eventually to reach a which the amount of atmospheric carbon diox-
steady state. ide increases at precisely 1% per year (about
twice the current rate) or doubles instanta-
Values of control-run forcing factors often are neously. Carbon dioxide doubling experiments
set to match present-day conditions, and model typically are run until the simulated climate
output is compared with present-day observa- reaches a steady state of equilibrium with the
tions. Actually, today’s climate is affected not enhanced greenhouse effect. Until the mid-
only by current forcing but also by the history of 1990s, idealized simulations often were em-
forcing over time—in particular, past emissions ployed to assess possible future climate changes
of greenhouse gases. Nevertheless, present-day including human-induced global warming. Re-
control-run output and present-day observations cently, however, more realistic time-evolving
are expected to agree fairly closely if models are simulations (defined immediately below) have
reasonably accurate. We compare model control been used for making climate predictions. We
runs with observations in Chapter 5. discuss idealized simulations and their implica-
tions for climate sensitivity in Chapter 4.
29
Time-dependent climate-forcing simulations Time-evolving climate forcing also is used as

are the most realistic, especially for eras in input for modeling future climate change. This
which climate forcing is changing rapidly, such subject is discussed in CCSP Synthesis and As-
as the 20th and 21st centuries. Input for 20th Cen- sessment Product 3.2. Finally, we mention for
tury simulations includes observed time-vary- the record simulations of the distant past (vari-
ing values of solar energy output, atmospheric ous time periods ranging from early Earth up to
carbon dioxide, and other climate-relevant gases the 19th Century). These simulations are not
and aerosols, including those produced in vol- discussed in this report, but some of them have
canic eruptions. Each modeling group uses its been used to loosely “paleocalibrate” simula-
own best estimate of these factors. Significant tions of the more recent past and the future
uncertainties occur in many of them, especially (Hoffert and Covey 1992; Hansen et al. 2006;
atmospheric aerosols, so different models use Hegerl et al. 2006).
different input for their 20th Century simula-
tions. We discuss uncertainties in climate-
forcing factors in Chapter 4 and 20th Century
simulations in Chapter 5 after comparing con-
trol runs with observations.
30
3C H APTER
A dded Value of
Regional C limate
Model Simulations
3.1 T YPES OF DOW N SCALIN G SIMU LAT ION S
This section focuses on downscaling using three-dimensional models based on fundamental con-
ser vation laws [i.e., numerical models with foundations similar to general circulation models
(GC Ms)]. A later section of the chapter discusses an alternative method, statistical downscaling.
There are three primary approaches to numeri- scaling global simulations, most especially for
cal downscaling: studying climatic processes and interactions on
scales too fine for typical GCM resolutions.
• Limited-area models (Giorgi and Mearns
1991, 1999; McGregor 1997; Wang et al.
As limited-area models, RCMs cover only a
2004).
portion of the planet, typically a continental do-
• Stretched-grid models (e.g., Déqué and main or smaller. They require lateral boundary
Piedelievre 1995; Fox-Rabinovitz et al. conditions (LBCs), obtained from observations
2001, 2006). such as atmospheric analyses (e.g., Kanamitsu
et al. 2002; Uppala et al. 2005) or a global sim-
• Uniformly high resolution atmospheric
ulation. There has been limited two-way cou-
GCMs (AGCMs) (e.g., Brankovic and Gre-
pling wherein an RCM supplies part of its
gory 2001; May and Roeckner 2001; Duffy
output back to the parent GCM (Lorenz and
et al. 2003; Coppola and Giorgi 2005).
Jacob 2005). Simulations with observation-
based boundary conditions are used not only to
Limited-area models, also known as regional study fine-scale climatic behavior but also to
climate models (RCMs), have the most wide- help segregate GCM errors from those intrinsic
spread use. The third method sometimes is to the RCM when performing climate change
called “time-slice” climate simulation because simulations (Pan et al. 2001). RCMs also may
the AGCM simulates a portion of the period use grids nested inside a coarser RCM simula-
represented by the coarser-resolution parent tion to achieve higher resolution in subregions
GCM that supplies the model’s boundary con- (e.g., Liang, Kunkel, and Samel 2001; Hay et
ditions. All three methods use interactive land al. 2006).
models, but sea-surface temperatures and sea
ice generally are specified from observations or Stretched-grid models, like high-resolution
an atmosphere-ocean GCM (AOGCM). All AGCMs, are global simulations but with spatial
three also are used for purposes beyond down- resolution varying horizontally. The highest res-
31
The U.S. Climate Change Science Program C hapter 3 - Added Value of Regional C limate Model Simulations
olution may focus on one (e.g., Déqué and Leung et al. 2004; Plummer et al. 2006) and
Piedelievre 1995; Hope, Nicholls, and McGre- even as long as 140 years (McGregor 1999)
gor 2004) or a few regions (e.g., Fox-Rabi- with no serious drift away from reality. Even so,
novitz, Takacs, and Govindaraju 2002). In some the RCM, stretched-grid, and time-slice AGCM
sense, the uniformly high resolution AGCMs simulations typically last only months to a few
are the upper limit of stretched-grid simulations years. Vertical resolution usually does not change
in which the grid is uniformly high everywhere. with horizontal resolution, although Lindzen and
Fox-Rabinovitz (1989) and Fox-Rabinovitz and
Highest spatial resolutions are most often sev- Lindzen (1993) have expressed concerns about
eral tens of kilometers, although some (e.g., the adequacy of vertical resolution relative to
Grell et al. 2000a, b; Hay et al. 2006) have sim- horizontal resolution in climate models.
ulated climate with resolutions as small as a few
kilometers using multiple nested grids. Duffy et Higher resolution in RCMs and stretched-grid
al. (2003) have performed multiple AGCM models also must satisfy numerical constraints.
time-slice computations using the same model Stretched-grid models whose ratio of coarsest-
to simulate resolutions from 310 km down to 55 to-finest resolution exceeds a factor of roughly
km. Higher resolution generally yields im- 3 are likely to produce inaccurate simulation
proved climate simulation, especially for fields due to truncation error (Qian, Giorgi, and Fox-
such as precipitation that have high spatial vari- Rabinovitz 1999). Similarly, RCMs will suffer
ability. For example, some studies show that from incompletely simulated energy spectra and
higher resolution does not have a statistically thus loss of accuracy if their resolution is about
significant advantage in simulating large-scale 12 times or more finer than the resolution of the
circulation patterns but does yield better mon- LBC source, which may be coarser RCM grids
soon precipitation forecasts and interannual (Denis et al. 2002; Denis, Laprise, and Caya
variability (Mo et al. 2005) and precipitation in- 2003; Antic et al. 2004, 2006; Dimitrijevic and
tensity (Roads, Chen, and Kanamitsu 2003). Laprise 2005). In addition, these same studies
indicate that LBCs should be updated more fre-
Improvement in results, however, is not guaran- quently than twice per day.
teed: Hay et al. (2006) find deteriorating timing
and intensity of simulated precipitation vs ob- Additional factors also govern ingestion of
servations in their inner, high-resolution nests, LBCs by RCMs. LBCs are most often ingested
even though the inner nest improves topography in RCMs by damping the model’s state toward
resolution. Extratropical storm tracks in a time- LBC fields in a buffer zone surrounding the do-
slice AGCM may shift poleward relative to the main of interest (Davies 1976; Davies and
coarser parent GCM (Stratton 1999; Roeckner Turner 1977). If the buffer zone is only a few
et al. 2006) or to lower-resolution versions of grid points wide, the interior region may suffer
the same AGCM (Brankovic and Gregory phase errors in simulating synoptic-scale waves
2001); thus these AGCMs yield an altered cli- (storm systems), with resulting error in the over-
mate with the same sea-surface temperature dis- all regional simulation (Giorgi, Marinucci, and
tribution as the parent model. Bates 1993). Spurious reflections also may
occur in boundary regions (e.g., Miguez-
Spatial resolution affects the length of simula- Macho, Stenchikov, and Robock 2005). RCM
tion periods because higher resolution requires boundaries should be where the driving data are
shorter time steps for numerical stability and ac- of optimum accuracy (Liang, Kunkel, and
curacy. Required time steps scale with the in- Samel 2001), but placing the buffer zone in a
verse of resolution and can be much smaller region of rapidly varying topography can induce
than AOGCM time steps. Increases in resolu- surface-pressure errors. These errors result from
tion most often are applied to both horizontal mismatch between the smooth topography im-
directions, meaning that computational demand plicit in the coarse resolution driving the data
varies inversely with the cube of resolution. and the varying topography resolved by the
Several RCM simulations have lasted 20 to 30 model (Hong and Juang 1998). Domain size
years (Christensen, Carter, and Giorgi 2002; also may influence RCM results. If a domain is
32
too large, the model’s interior flow may drift slowly. Equally important, data for initial con-
from the large-scale flow of the driving dataset ditions often are lacking or have low spatial res-
(Jones, Murphy, and Noguer 1995). However, olution, so initial conditions may be only a poor
too small a domain overly constrains interior approximation of the model’s climatology.
dynamics, preventing the model from generat- Spinup is especially relevant for downscaling
ing appropriate response to interior mesoscale- because these models presumably are resolving
circulation and surface conditions (Seth and finer surface features than coarser models, with
Giorgi 1998). RCMs appear to perform well for the expectation that the downscaling models are
domains roughly the size of the contiguous providing added value through proper represen-
United States. Figure 3.1 shows that the daily, tation of these surface features. Deep-soil tem-
root-mean-square difference (RMSD) between perature and moisture, at depths of 1 to 2
simulated and observed (reanalysis) 500-hPa meters, may require several years of spinup.
heights generally is within observational noise However, these deep layers generally interact
levels (about 20 m). weakly with the rest of the model, so shorter
spinup times are used. For multiyear simula-
Because simulations from the downscaling tions, a period of 3 to 4 years appears to be the
models may be analyzed for periods as short as minimal requirement (Christensen 1999; Roads
a month, model spinup is important (e.g., Giorgi et al. 1999). This ensures that the upper meter of
and Bi 2000). During spinup, the model evolves soil has a climatology in further simulations that
to conditions representative of its own clima- is consistent with the evolving atmosphere.
tology, which may differ from the sources of ini-
tial conditions. The atmosphere spins up in a Many downscaling simulations, especially with
matter of days, so the key factor is spinup of soil RCMs, are for periods much shorter than 2
moisture and temperature, which evolve more years. Such simulations probably will not use
Figure 3.1. Daily Root-Mean-Square Differences (RMSD) in 500-hPa H eights Between Observations
(Reanalysis) and Seven Models Participating in the PIRCS 1a Experiment, for May 15 to July 15, 1988.
RMSD values were averaged over the simulation domain inside the boundar y-forcing zone. [Adapted from Fig. 4 in E.S. Takle et al. 1999:
Project to Intercompare Regional C limate Simulations (PIRC S): D escription and initial results. J. Geophysical Research, 104, 19443–
19461. U sed with permission of the American Geophysical U nion.]
33
multiyear spinup. Rather, these studies may whereas the Kain and Fritsch (1993) scheme is
focus on more rapidly evolving atmospheric be- heavily influenced by boundary-layer forcing.
havior governed by LBCs, including extreme As a result, the Grell scheme better simulates
periods such as drought (Takle et al. 1999) or the propagation of precipitation over the U.S.
flood (Giorgi et al. 1996; Liang, Kunkel, and Great Plains that is controlled by large-scale tro-
Samel 2001; Anderson, C. J., et al. 2003). Thus, pospheric forcing, while the Kain–Fritsch
they assume that interaction with the surface, scheme better simulates late-afternoon convec-
while not negligible, is not strong enough to tion peaks in the southeastern United States that
skew the atmospheric behavior studied. Alter- are governed by boundary-layer processes
natively, relatively short regional simulations (Liang et al. 2004). As a consequence, parame-
may specify, for sensitivity study, substantial terizations for regional simulation may differ
changes in surface evaporation (e.g., Paegle, from their GCM counterparts, especially for
Mo, and Nogués-Paegle 1996), soil moisture convection and cloud microphysics. As noted
(e.g., Xue et al. 2001), or horizontal moisture earlier, regional simulation in some cases may
flux at lateral boundaries (e.g., Qian, Tao, and have resolution of only a few kilometers, and
Lau 2004). the convection parameterization may be dis-
carded (Grell et al. 2000). A variety of parame-
3.1.1 Parameterization Issues terizations exist for each subgrid process, with
multiple choices often available in a single
Even with higher resolution than standard model (e.g., Grell, Dudhia, and Stanfler 1994;
GCMs, models simulating regional climate still Skamarock et al. 2005).
need parameterizations for subgrid-scale
processes, most notably boundary-layer dy- 3.1.2 Regional Simulation vs
namics, surface-atmosphere coupling, radiative Computational Costs
transfer, and cloud microphysics. Most regional
simulations also require a convection parame- The chief reason for performing regional simu-
terization, although a few have used sufficiently lation, whether by an RCM, a stretched-grid
fine grid spacing (a few kilometers) to allow ac- model, or a time-slice AGCM, is to resolve be-
ceptable simulation without it (e.g., Grell et al. havior considered important for a region’s cli-
2000). Often, these parameterizations are the mate that a global model does not resolve. Thus,
same or nearly the same as those used in GCMs. regional simulation should have clearly defined
All parameterizations, however, make assump- regional-scale (mesoscale) phenomena targeted
tions that they are representing the statistics of for simulation. These include tropical storms
subgrid processes. Implicitly or explicitly, they (e.g., Oouchi et al. 2006), effects of mountains
require that the grid box area in the real world (e.g., Leung and Wigmosta 1999; Grell et al.
has sufficient samples to justify stochastic mod- 2000; Zhu and Liang 2007), jet circulations
eling. For some parameterizations such as con- (e.g., Takle et al. 1999; Anderson et al. 2001;
vection, this assumption becomes doubtful Anderson, C. J., et al. 2003; Byerle and Paegle
when grid boxes are only a few kilometers in 2003; Pan et al. 2004), and regional ocean-land
size (Emanuel 1994). interaction (e.g., Kim et al. 2005; Diffenbaugh,
Snyder, and Sloan 2004). The most immediate
In addition, models simulating regional climate value of regional simulation, then, is to explore
may include circulation characteristics, such as how such phenomena operate in the climate sys-
rapid mesoscale circulations (jets) whose inter- tem, an understanding of which becomes a jus-
action with subgrid processes like convection tification for the expense of performing regional
and cloud cover differs from larger-scale circu- simulation. Phenomena and computational
lations resolved by typical GCMs. This factor is costs together influence the design of regional
part of a larger issue, that parameterizations simulations. Simulation periods and resolution
may have regime dependence, performing bet- are balanced between sufficient length and
ter for some conditions than for others. For ex- number of simulations for climate statistics vs
ample, the Grell (1993) convection scheme is computational cost. For RCMs and stretched-
responsive to large-scale tropospheric forcing, grid models, the sizes of regions targeted for
34
high-resolution simulation are determined in may capture much of the uncertainty in climate
part by where the phenomenon occurs. simulation, offering an opportunity for physi-
cally based analysis of climate changes and also
In the context of downscaling, regional simula- the uncertainty of the changes. Several regional
tion offers the potential to include phenomena models have performed simulations of climate
affecting regional climate change that are not change for parts of North America, but at pres-
explicitly resolved in the global simulation. ent no regional projections have used an en-
When incorporating boundary conditions cor- semble of regional models to simulate the same
responding to future climate, regional simula- time periods with the same boundary condi-
tion can then indicate how these phenomena tions. Such systematic evaluation has occurred
contribute to climate change. Results, of course, in Europe in the PRUDENCE (Christensen,
are dependent on the quality of the boundary- Carter, and Giorgi 2002) and ENSEMBLES
condition source (Pan et al. 2001; de Elía, (Hewitt and Griggs 2004) projects and is start-
Laprise, and Denis 2002), although use of mul- ing in North America with the North American
tiple sources of future climate may lessen this Regional Climate Change Assessment Program
vulnerability and offer opportunity for proba- (NARCCAP 2007).
bilistic estimates of regional climate change
(Raisanen and Palmer 2001; Giorgi and Mearns 3.2 EMPIRICAL DOW N SCALIN G
2003; Tebaldi et al. 2005). Results also depend
on the physical parameterizations used in the Empirical or statistical downscaling is an alter-
simulation (Yang and Arritt 2002; Vidale et al. native approach to obtaining regional-scale cli-
2003; Déqué et al. 2005; Liang et al. 2006). mate information (Kattenberg et al. 1996;
Hewitson and Crane 1996; Giorgi et al. 2001;
Advances in computing power suggest that typ- Wilby et al. 2004, and references therein). It
ical GCMs eventually will operate at resolutions uses statistical relationships to link resolved be-
of most current regional simulations (a few tens havior in GCMs with climate in a targeted area.
of kilometers), so understanding and modeling The targeted area’s size can be as small as a sin-
improvements gained for regional simulation gle point. As long as significant statistical rela-
can promote appropriate adaptation of GCMs to tionships occur, empirical downscaling can
higher resolution. For example, interaction be- yield regional information for any desired vari-
tween mesoscale jets and convection appears to able such as precipitation and temperature, as
require parameterized representation of con- well as variables not typically simulated in cli-
vective downdrafts and their influence on the mate models, such as zooplankton populations
jets (Anderson, Arritt, and Kain 2007), parame- (Heyen, Fock, and Greve 1998) and initiation of
terized behavior not required for resolutions that flowering (Maak and von Storch 1997). This ap-
do not resolve mesoscale circulations. proach encompasses a range of statistical tech-
niques from simple linear regression (e.g.,
Because of the variety of numerical techniques Wilby et al. 2000) to more-complex applica-
and parameterizations employed in regional tions such as those based on weather generators
simulation, many models and versions of mod- (Wilks and Wilby 1999), canonical correlation
els exist. Generally in side-by-side comparisons analysis (e.g., von Storch, Zorita, and Cubasch
(e.g., Takle et al. 1999; Anderson, C. J., et al. 1993), or artificial neural networks (e.g., Crane
2003; Fu et al. 2005; Frei et al. 2006; Rinke et and Hewitson 1998). Empirical downscaling
al. 2006), no single model appears best vs ob- can be very inexpensive compared to numerical
servations, with different models showing su- simulation when applied to just a few locations
perior performance depending on the field or when simple techniques are used. Lower
examined. Indeed, the best results for down- costs, together with flexibility in targeted vari-
scaling climate simulations and estimating cli- ables, have led to a wide variety of applications
mate-change uncertainty may come from for assessing impacts of climate change.
assessing an ensemble of simulations (Giorgi
and Bi 2000; Yang and Arritt 2002; Vidale et al. Some methods have been compared side by side
2003; Déqué et al. 2005). Such an ensemble (Wilby and Wigley 1997; Wilby et al. 1998;
35
Zorita and von Storch 1999; Widman, Brether- cially short-term variability such as extreme
ton, and Salathe 2003). These studies have winds and locally extreme temperature that
tended to show fairly good performance of rel- coarser-resolution models will smooth and thus
atively simple vs more-complex techniques and inhibit.
to highlight the importance of including mois-
ture and circulation variables when assessing Mean fields also appear to be simulated some-
climate change. Statistical downscaling and re- what better on average than are those in coarser
gional climate simulation also have been com- GCMs because spatial variations potentially are
pared (Kidson and Thompson 1998; Mearns et better resolved. Thus, Giorgi et al. (2001) report
al. 1999; Wilby et al. 2000; Hellstrom et al. typical errors in RCMs of less than 2˚C temper-
2001; Wood et al. 2004; Haylock et al. 2006), ature and 50% for precipitation in regions 105 to
with no approach distinctly better or worse than 106 km2. Large-scale circulation fields tend to
any other. Statistical methods, though compu- be well simulated, at least in the extratropics.
tationally efficient, are highly dependent on the
accuracy of regional temperature, humidity, and As alluded to above, regional-scale simulations
circulation patterns produced by their parent also have phenomenological value, simulating
global models. In contrast, regional climate sim- processes that GCMs either cannot resolve or
ulation, though computationally more demand- can resolve only poorly. These include internal
ing, can improve the physical realism of circulation features such as the nocturnal jet that
simulated regional climate through higher reso- imports substantial moisture to the center of the
lution and better representation of important re- United States and couples with convection (e.g.,
gional processes. The strengths and weaknesses Byerle and Paegle 2003; Anderson, Arritt, and
of statistical downscaling and regional model- Kain 2007). These processes often have sub-
ing thus are complementary. stantial diurnal variation and thus are important
to proper simulation of regional diurnal cycles
3.3 ST REN GT H S AN D of energy fluxes and precipitation. Some
LIMITAT ION S OF REGION AL processes require the resolution of surface fea-
MODELS tures too coarse for typical GCM resolution.
These include rapid topographic variation and
We focus here on numerical models simulating its influence on precipitation (e.g., Leung and
regional climate but do not discuss empirical Wigmosta 1999; Hay et al. 2006) and the cli-
downscaling because the wide range of appli- matic influences of bodies of water such as the
cations using the latter makes difficult a general Gulf of California (e.g., Anderson et al. 2001)
assessment of strengths and limitations. and the North American Great Lakes (Lofgren
2004) and their downstream influences. In ad-
The higher resolution in regional-scale simula- dition, regional simulations resolve land-surface
tions provides quantitative value to climate sim- features that may be important for climate-
ulation. With finer resolution, scientists can change impact assessments such as distributions
resolve mesoscale phenomena contributing to of crops and other vegetation (Mearns 2003;
intense precipitation, such as stronger upward Mearns et al. 2003), although care is needed to
motions (Jones, Murphy, and Noguer 1995) and obtain useful information at higher resolution
coupling between regional circulations and con- (Adams, McCarl, and Mearns 2003).
vection (e.g., Anderson, Arritt, and Kain 2007).
Time-slice AGCMs show intensified storm An important limitation for regional simulations
tracks relative to their parent model (Solman, is that they are dependent on boundary condi-
Nunez, and Rowntree 2003; Roeckner et al. tions supplied from some other source. This ap-
2006). Thus, although regional models may still plies to all three forms of numerical simulation
miss the most extreme precipitation (Gutowski (RCMs, stretched-grid models, and time-slice
et al. 2003, 2007a), they can give more intense AGCMs), since they all typically require input
events that will be smoothed in coarser-resolu- of sea-surface temperature and ocean ice. Some
tion GCMs. The higher resolution also includes RCM simulations have been coupled to a re-
other types of scale-dependent variability, espe- gional ocean-ice model, with mixed-layer ocean
36
(Lynch et al. 1995; Lynch, Maslanic, and Wu RCMs also may exhibit difficulty in outflow re-
2001) and a regional ocean-circulation model gions of domains, especially regions with rela-
(Rummukainen et al. 2004), but this is not com- tively strong cross-boundary flow, which may
mon. In addition, of course, RCMs require occur in extratropical domains covering a sin-
LBCs. Thus, regional simulations by these mod- gle continent or less. The difficulty appears to
els are dependent on the model quality or on ob- arise because storm systems may track across
servations supplying boundary conditions. This the RCM’s domain at a different speed from
is especially true for projections of future cli- their movement in the driving-data source, re-
mate, suggesting value in performing an en- sulting in a mismatch of circulations at bound-
semble of simulations using multiple aries where storms would be moving out of the
atmosphere-ocean global models to supply domain. Also, unresolved scales of behavior are
boundary conditions, thus including some of the always present, so regional simulations are still
uncertainty involved in constructing climate dependent on parameterization quality for the
models and projecting future changes in bound- scales explicitly resolved. Finally, higher com-
ary conditions. putational demand due to shorter time steps lim-
its the length of typical simulations to 2 to 3
Careful evaluation also is necessary to show dif- decades or less (e.g., Christensen, Carter, and
ferences, if any, between the regional simula- Giorgi 2002; NARCCAP 2007), with few en-
tion’s large-scale circulation and its driving semble simulations to date.
dataset. Generally, any tendency for the regional
simulation to alter biases in the parent GCM’s
large-scale circulation should be viewed with
caution (Jones, Murphy, and Noguer 1995). An
RCM normally should not be expected to cor-
rect large-scale circulation problems of the par-
ent model unless the physical basis for the
improvement is clearly understood. Clear phys-
ical reasons for the correction due to higher res-
olution, such as better rendition of physical
processes like topographic circulation (e.g.,
Leung and Qian 2003), surface-atmosphere in-
teraction (Han and Roads 2004), and convec-
tion (Liang et al. 2006) must be established.
Otherwise, the regional simulation may simply
have errors that counteract the parent GCM’s er-
rors, thus undermining confidence in projected
future climate.
37
38
4 CH APTER
M odel Climate
Sensitivity
The response of climate to a perturbation such as a change in carbon dioxide concentration, or

in the flux of energy from the sun, can be divided into two factors:“radiative forcing” due to the
perturbation in question and “climate sensitivity,” characterizing the response of the climate per
unit change in radiative forcing. Climate response is then the product of radiative forcing and cli-
mate sensitivity. This distinction is useful because of two approximations: radiative forcing often
can be thought of as independent of the resulting climate response, and climate sensitivity can
often be thought of as independent of the agent responsible for perturbation to the energy bal-
ance. W hen two or more perturbations are present simultaneously, their cumulative effect can be
approximated by adding their respective radiative forcings (H ansen et al. 2006).
Climate sensitivity as traditionally defined carbon dioxide would add energy to the surface
refers to the global mean temperature, but a and the troposphere at the rate of about 4 W/m2
model’s global mean temperature response is for the first few months after the doubling
very relevant to its regional temperature re- (Forster et al. 2007). Eventually, lower tropos-
sponses as well. This “pattern scaling” effect is pheric temperatures would increase (and cli-
discussed at the end of this chapter. mate would change in other ways) in response
to this forcing, Earth would radiate more energy
Radiative forcing typically is calculated by to space, and the imbalance would diminish as
changing the atmospheric composition or ex- the system returned to equilibrium.
ternal forcing and computing the net trapping
of heat that occurs before the climate system has 4.1 CH ARACT ERIZIN G CLIMAT E
had time to adjust.1 These direct heat-trapping RESPON SE
properties are well characterized for the most
significant greenhouse gases. As a result, un- 4.1.1 Equilibrium Sensitivity and
certainty in climate responses to greenhouse Transient Climate Response
gases typically is dominated by uncertainties in
climate sensitivity rather than in radiative forc- The idea of characterizing climate response
ing (Ramaswamy et al. 2001). For example, using a single number represented by climate
suddenly doubling the atmospheric amount of sensitivity appeared early in the development of
1
Because the stratosphere cools rapidly in response to increasing carbon dioxide and this cooling affects the net
warming of the lower atmosphere and surface, it has become standard to include the effects of this stratospheric cool-
ing in estimating radiative forcing due to carbon dioxide.
39
The U.S. Climate Change Science Program Chapter 4 - Model Climate Sensitivity
climate models (e.g., Schneider and Mass mates, where observations represent periods
1975). Today, two different numbers are in com- that are very long compared to the climate’s ad-
mon use. Both are based on changes in global justment time. The transient climate response is
and annual mean-surface or near-surface tem- more directly relevant to the attribution of re-
perature. Equilibrium sensitivity is defined as cent warming and projections for the next cen-
the long-term near-surface temperature increase tury. For example, Stott et al. (2006) show that
after atmospheric carbon dioxide has been dou- global mean warming due to well-mixed green-
bled from preindustrial levels but thereafter held house gases over the 20th Century, in the set of
constant until the Earth reaches a new steady models they consider, is closely proportional to
state, as described in the preceding paragraph. the model’s TCR. In the following, we discuss in-
Transient climate response or TCR is defined dividual feedback processes as well as these ad-
by assuming that carbon dioxide increases by ditional observational constraints on sensitivity.
1% per year and then recording the temperature
increase at the time carbon dioxide doubles Equilibrium warming in an AOGCM is difficult
(about 70 years after the increase begins). TCR to obtain because the deep ocean takes a great
depends on how quickly the climate adjusts to deal of time to respond to changes in climate
forcing, as well as on equilibrium sensitivity. forcing. To avoid unacceptably lengthy com-
The climate’s adjustment time itself depends on puter simulations, equilibrium warming usually
equilibrium sensitivity and on the rate and depth is estimated from a modified climate model in
to which heat is mixed into the ocean, because which the ocean component is replaced by a
the depth of heat penetration tends to be greater simplified, fast-responding “slab ocean model.”
in models with greater sensitivity (Hansen et al. This procedure makes the assumption that hor-
1985; Wigley and Schlesinger 1985). Account- izontal redistribution of heat in the ocean does
ing for ocean heat uptake complicates many at- not change as the climate responds to the per-
tempts at estimating sensitivity from observations, turbation. Current climate models generate a
as outlined below. range of equilibrium and transient climate sen-
sitivities. For the models in the CMIP3 archive
Equilibrium sensitivity depends on the strengths utilized in the Fourth Assessment of the IPCC,
of feedback processes involving water vapor, the range of equilibrium sensitivity is 2.1 to
clouds, and snow or ice extents (see, e.g., 4.4°C with a median of 3.2°C. This ensemble of
Hansen et al. 1984; Roe and Baker 2007). Small models was not constructed to systematically
changes in the strengths of feedback processes span the plausible range of uncertainty in cli-
can create large changes in sensitivity, making it mate sensitivity; rather, each development team
difficult to tightly constrain climate sensitivity simply provided its best attempt at climate sim-
by restricting the strength of each relevant feed- ulation. Complementary to this approach is one
back process. As a result, research aimed at con- in which a single climate model is modified in
straining climate sensitivity—and evaluating the a host of ways to explore more systematically
sensitivities generated by models—is not lim- the sensitivity variations associated with the
ited to studies of these individual feedback range of uncertainty in various key parameters.
processes. Studies of observed climate re- Results with a Hadley Centre model give a 5 to
sponses on short time scales (e.g., the response 95 percentile range of ~2 to 6°C for equilibrium
to volcanic eruptions or the 11-year solar cycle) sensitivity (Piani et al. 2005; Knutti et al. 2006).
and on long time scales (e.g., the climate of last
glacial maximum 20,000 years ago) also play Charney (1979) provided a range of equilibrium
central roles in the continuing effort to constrain sensitivities to CO2 doubling of 1.5 to 4.5°C,
sensitivity. The quantitative value of each of based on the two model simulations available at
these observational constraints is limited by the the time. Evidently, the range of model-implied
quality and length of relevant observational climate sensitivity has not contracted signifi-
records, as well as the necessity in several cases cantly over three decades. The current range,
to simultaneously restrict ocean heat uptake and however, is based on a much larger number of
equilibrium sensitivity. Equilibrium warming is models subjected to a far more comprehensive
directly relevant when considering paleocli- comparison to observations and containing
40
more detailed treatments of clouds and other Table 4.1

TCR Equilibrium
processes that are fundamental to climate sen- MODEL Equilibrium and
(ºC) Warming (ºC)*
sitivity. We understand in much more detail why Transient
models differ in their equilibrium climate sen- CSM1.4 1.4 2.0 Sensitivities in
Some U.S. Models
sitivities: the source of much of this spread lies
CCSM2 1.1 2.3 Contributing to
in differences in how clouds are modeled in CMIP3
AOGCMs. Questions remain as to whether or CCSM3 1.5 2.5
not the substantial spread among models is a
GFD L CM2.0 1.6 2.9
good indication of the uncertainty in climate
sensitivity, given all the constraints on this GFD L CM2.1 1.5 3.4
quantity of which we are aware. There also is a
GISS Model E 2.7 to 2.9
desire to know the prospects for constraining
equilibrium climate sensitivity more sharply in * Equilibrium warming was assessed by joining a
the near future. simplified slab ocean model to the atmosphere,
land, and sea-ice AOGCM components.
The variation among models is less for TCR [Sources of Information in table. First three
than for equilibrium warming, a consequence of lines – J.T. Kiehl et al. 2006:The climate
sensitivity of the Community Climate System
the interrelationship between the climate’s ad- Model: CCSM3. J. Climate, 19, 2584–2596. N ext
justment time and its sensitivity to forcing noted two lines – R.J. Stouffer et al. 2006: GFD L’s CM2
above (Covey et al. 2003). The full range for global coupled climate models. Part IV: Idealized
climate response. J. Climate, 19, 723–740. Last
TCR in the CMIP3 archive is 1.3 to 2.6°C, with line – J. H ansen et al. 2007: Climate simulations
a median of 1.6°C and 25 to 75% quartiles of for 1880–2003 with GISS ModelE. Climate
1.5 to 2.0°C (Randall et al. 2007). Systematic D ynamics, 29(7–8), 661–696.]
exploration of model input parameters in one

Hadley Centre model gives a range of 1.5 to
2.6°C (Collins, M., et al. 2006). modification to the cloud-prediction scheme al-
ters climate sensitivity. In the standard version
The equilibrium and transient sensitivities in of the model, the effective size of cloud drops
some models developed by U.S. centers con- was fixed. In two other versions, this cloud-drop
tributing to CMIP3 are listed in Table 4.1. In the size was tied to the total amount of liquid-water
last column, the larger of the two GISS ModelE cloud through two different empirical relation-
values is obtained using a full ocean model in ships. The equilibrium sensitivity ranged from
which the circulation is allowed to adjust. All 1.9 to 5.5°C in these three models. In general,
other values of equilibrium warming in the table the nonlinear dependence of equilibrium sensi-
are obtained with the ocean component replaced tivity on the strength of feedback processes al-
by a slab ocean model. The close agreement in lows relatively small changes in feedbacks to
transient climate sensitivity among models in generate large changes in sensitivity (see, e.g.,
this subset should not be overinterpreted, given Hansen et al. 1984; Roe and Baker 2007).
the larger range among the full set of CMIP3
models. Studies of the CCSM family of models provide
another example of this problem. Kiehl et al.
Climate sensitivity is not a model input. It (2006) found that a variety of factors is respon-
emerges from explicitly resolved physics, sub- sible for differences in climate sensitivity
grid-scale parameterizations, and numerical ap- among the models of this family. However, the
proximations used by the models—many of lower TCR of CCSM2 (relative to CSM1.4 and
which differ from model to model—particularly CCSM3), evident in Table 4.1, results primarily
those related to clouds and ocean mixing. The from a single change in the model’s algorithm
climate sensitivity of a model can be changed for simulating convective clouds. Table 4.2
by modifying parameters that are poorly con- shows how equilibrium sensitivity varied dur-
strained by observations or theory. Influential ing development of the most recent GFDL mod-
early papers by Senior and Mitchell (1993, els. The dramatic drop in sensitivity between
1996) demonstrated how a seemingly minor model versions p10 and p12.5.1 was unex-
41
mation, particularly from the peak of the last Ice

Table 4.2
Equilibrium Global Equilibrium Age some 20,000 years ago; aspects of the sea-
MODEL VERSION
Mean N ear-Surface Warming (ºC)* sonal cycle; and, needless to say, the magnitude
W arming Due to of observed warming over the past century.
Doubled p7 3.87
Atmospheric 4.1.2.1 VO LC AN IC ERUPTIO N S
Carbon Dioxide p9 4.28
from Intermediate Volcanoes provide a rapid change in radiative
(“p”) Model forcing due to the scattering and absorption of
p10 4.58
Versions Leading to solar radiation by stratospheric volcanic aerosol.
GFDL’s CM2.0 and Of special importance, recovery time after the
CM2.1 p12.5.1 2.56
eruption contains information about climate
sensitivity that is independent of uncertainties
p12.7 2.65
in the magnitude of the radiative forcing per-
turbation (e.g., Lindzen and Giannitsis 1998).
p12.10b 2.87
Larger climate sensitivity implies weaker restor-
p12b 2.83
ing forces on Earth’s temperature, and, there-
fore, a slower relaxation back toward the
CM 2.0 2.90 unperturbed climate. However, this time scale
also is affected by the pathways through which
CM 2.1 3.43 heat anomalies propagate into the ocean depths,
with deeper penetration increasing the relax-
ation time. Several modeling studies have con-
* Equilibrium warming was assessed by joining a firmed that this relaxation time after an eruption
simplified slab ocean model to the atmosphere,
increases as climate sensitivity increases in
land, and sea-ice AOGCM components.
GCMs when holding the ocean model fixed
[Source of information for table: Personal
communication with Thomas Knutson, N O AA (Soden et al. 2002; Yokohata et al. 2005), en-
GFD L laborator y.] couraging the use of volcanic responses to con-
strain sensitivity. On the other hand, Boer,
Stowasser, and Hamilton (2007) study two mod-
els with differing climate sensitivity and differ-
pected. It followed a reformulation of the ent ocean models; they highlight the difficulty
model’s treatment of processes in the lower at- in determining which model has the higher sen-
mospheric boundary layer, which, in turn, af- sitivity from the surface-temperature responses
fected how low-level clouds in the model to volcanic forcing in isolation, without quanti-
respond to climate change. tative information on ocean heat uptake.
4.1.2 Observational Constraints on Some studies have argued that observations of

Sensitivity responses to volcanoes imply that models are
overestimating climate sensitivity (e.g., Dou-
Climate models in isolation have not yet con- glass and Knox 2005; Lindzen and Giannitsis
verged on a robust value of climate sensitivity. 1998). These studies argue that observed relax-
Furthermore, the actual climate sensitivity in ation times are shorter than those expected if
nature might not be found in the models’ range climate sensitivity is as large as in typical
of sensitivities, since all the models may share AOGCMs. Studies that directly examine the
common deficiencies. However, observations volcanic responses in AOGCMs, however, find
can be combined with models to constrain cli- no such gross disagreement with observations
mate sensitivity. The observational constraints (Wigley et al. 2005; Boer, Stowasser, and
include the response to volcanic eruptions; as- Hamilton 2007; Frame et al. 2005) consistent
pects of the internal variability of climate that with earlier studies (e.g., Hansen et al. 1996;
provide information on the strength of climatic Santer et al. 2001). They nevertheless consis-
“restoring forces”; the response to the 11-year tently suggest (Frame et al. 2005; Yokohata et
cycle in solar irradiance; paleoclimatic infor- al. 2005) that climate sensitivities as large as
42
6ºC are inconsistent with observed relaxation sitivity. Total solar irradiance is known to vary
times. It is important to note that these “obser- by roughly 0.1% over this cycle (Frölich 2002).
vational” studies of climate sensitivity that do The expected response in global mean temper-
not utilize GCMs still make use of models, but ature is only ~0.1ºC, so the technique is limited
they use simple energy balance “box” models in value by the quality and length of the obser-
rather than GCMs. The value of these studies vational record, both of which restrict our abil-
depends on the relevance of the simple models ity to isolate this small signal. Recent results
as well as on the techniques for estimating pa- show promise in more cleanly identifying the
rameters in models that control climate sensi- climatic response to this cyclic perturbation
tivity. From these analyses, one can infer that (Camp and Tung 2007). Since ultraviolet wave-
further research isolating changes in ocean heat lengths play a disproportionately larger role in
content after eruptions, such as that of Church, these cyclic variations, detailed representations
White, and Arblaster (2005), will be needed to of the stratosphere and mesosphere, where ul-
strengthen constraints on climate sensitivity traviolet radiation is absorbed, along with ozone
provided by responses to volcanic eruptions. chemistry are required for quantitative analysis
of climatic response to the solar cycle (e.g.,
4.1.2.2 N ATURAL C LIMATE VARIABILITY
Shindell et al. 2006). Solar variations also have
Natural variability of climate also provides a been invoked repeatedly to explain early 20th
way of estimating the strength of the restoring Century warming and to connect the Little Ice
forces that determine climate sensitivity. Just as Age to the Maunder Minimum in sunspot num-
investigators learn something about sensitivity ber. While these connections may very well
by watching the climate recover from a volcanic have a valid basis, using them to constrain cli-
eruption, they can hope to obtain similar infor- mate sensitivity remains difficult as long as
mation by watching the climate relax from an variations in insolation on time scales longer
unforced period of unusual global warmth or than the 11-yr cycle are not better quantified.
cold. This approach to constraining the response To illustrate the difficulty, we note the substan-
to a perturbation by examining the character of tial reduction in estimated insolation variations
a system’s natural variability, discussed by Leith in the 20th Century between the Third and
(1975) in the context of climate sensitivity, is Fourth IPCC Assessments (Forster et al. 2007).
referred to as “fluctuation-dissipation” analysis Further analyses of responses to the sunspot
in other branches of physics. In the case of equi- cycle in models and observations seem likelier
librium statistical mechanics, this relationship to lead to stronger constraints on climate sensi-
between characteristics of natural variability tivity in the near term.
and response to an external force has been
4.1.2.4 G LACIAL-IN TERGLACIAL VARIATIO N S
placed on a firm theoretical footing, but appli-
cation to the climate is more heuristic, gener- The glacial-interglacial fluctuations of the Pleis-
ally depending on approximation of the climate tocene (the Ice Ages) are thought to be forced
system by a linear stochastically forced model. by changes in the Earth’s orbit on time scales of
The power of the approach is illustrated by Grit- 20,000 years and longer—the astronomical the-
sun and Branstator (2007) in a study of the ex- ory of the Ice Ages. Since this theory assumes
tratropical atmosphere’s response to a that the mean temperature of the Earth can be
perturbation in tropical heating. A recent attempt altered by changing the distribution of the in-
to apply this approach to climate sensitivity can coming solar flux without changing its global
be found in Schwartz (2007). This technique de- mean, it suggests important limitations to sim-
serves more attention, with careful analysis of ple models based solely on global mean radia-
uncertainties. Its value likely will be determined tive forcing. For the limited purpose of
by its ability to infer an AOGCM’s sensitivity constraining climate sensitivity, we need not un-
from an analysis of its internal variability. derstand how glacial-interglacial variations of
ice sheets and of carbon dioxide are forced by
4.1.2.3 SO LAR VARIATIO N S
changes in the Earth’s orbit. Since we have
The 11-year solar cycle has potential for pro- knowledge from ice cores of greenhouse gas
viding very useful information on climate sen- concentrations at the peak of the last major gla-
43
cial advance 20,000 years ago as well as con- ity in models is found, investigators can then ex-
siderable information on the extent of conti- amine its value in observations and hope
nental ice sheets, one may ask if climate models thereby to constrain climate sensitivity. Knutti
can simulate the ocean-surface temperatures in- et al. (2006) use a neural network to look for as-
ferred from a variety of proxies, given these pects of the seasonal cycle with this predictive
greenhouse gas concentrations and ice sheets capability, with some success. Their study fa-
(Manabe and Broccoli 1985). A logical as- vors sensitivity in the middle of the typical
sumption is that models that are more sensitive model range (near 3˚C).
to doubling of carbon dioxide would also simu-
late larger cooling during the low carbon diox- The work of Qu and Hall (2006) provides an es-
ide levels 20,000 years ago. Crucifix (2006) pecially straightforward example of this ap-
describes some of the difficulties with this sim- proach. They do not address climate sensitivity
ple picture. Annan and Hargreaves (2006) argue directly but only the strength of one feedback
that the tropics and Antarctica are regions where mechanism that contributes to sensitivity: snow-
this connection may be the strongest. Model re- albedo feedback (the decrease in reflection of
sults generated in the Paleoclimate Modelling solar radiation by snow as the snowcover re-
Intercomparison Project (Braconnet et al. treats in a warming climate). Qu and Hall
2007a, b; Crucifix et al. 2006)) provide a valu- demonstrate that the strength of this feedback
able resource for analyzing these relationships. in models is strongly correlated to the seasonal
Despite these complications, several studies cycle of the snow cover simulated by the mod-
agree that past climates are difficult to recon- els. Comparison of observed and simulated sea-
cile with the low end of the equilibrium-sensi- sonal cycles of snow cover then suggest which
tivity range generated by models (e.g., Hansen model simulations of snow albedo feedback are
et al. 1993; Covey, Sloan, and Hoffert 1996). the most reliable. These studies suggest that de-
Models of the last glacial maximum also pro- tailed comparisons of modeled and observed
vide some of the strongest evidence that climate seasonal cycles should provide valuable infor-
sensitivity is very unlikely to be larger than 6°C mation on climate sensitivity in the future.
(Annan et al. 2005; Annan and Hargreaves
2006). As paleoclimatic reconstructions for this The observed 20th Century warming is a funda-
period improve, these simulations will become mental constraint on climate models, but it is
of greater quantitative value. Uncertainty in Ice less useful than one might think in constraining
Age aerosol concentrations may be the most dif- sensitivity because of the large uncertainty in
ficult obstacle to overcome. forcing due to anthropogenic aerosols. Twenti-
eth Century simulations are important in
4.1.2.5 SEASO N AL VARIATIO N
demonstrating the consistency of certain com-
The seasonal cycle is a familiar forced climate binations of sensitivity, aerosol forcing, and
response to changes in the Earth-sun geometry ocean-heat uptake, but they do not provide a
and, therefore, should yield information on cli- sharp constraint on sensitivity in isolation
mate sensitivity. Although the seasonal cycles (Kiehl 2007). Further discussion of 20th Century
of global (Lindzen 1994) and hemispheric simulations can be found in Chapter 5.
(Covey et al. 2000) mean temperature are not
themselves strongly related to equilibrium cli- Rather than focusing on one particular observa-
mate sensitivity, regional variations and other tional constraint or on models in isolation, at-
aspects of the seasonal cycle may constrain sen- tempts to combine some or all of these
sitivity. Knutti et al. (2006) provide an example observational constraints with model simula-
of a methodology using ensembles of climate tions are recognized as the most productive ap-
model simulations to search for variables, or proaches to constraining climate sensitivity
combinations of variables, that correlate with (Bierbaum et al. 2003; Randall et al. 2007; Stott
climate sensitivity (see also Shukla et al. 2006). and Forest 2007). As an example, while model
If such a variable that predicts climate sensitiv- ensembles in which parameters are varied sys-
2
Estimating the probability of very high climate sensitivities above the high end of the CMIP3 model range, even if
these probabilities are low, can be relevant for analyses of unlikely but potentially catastrophic climate change. It is
not within the scope of this report to attempt to quantify these probabilities.
44
tematically can include models with sensitivi- that models disagree in their estimates of equi-
ties larger than 6ºC (Stainforth et al. 2005; Roe librium climate sensitivity; which (if any) mod-
and Baker 2007), these very high values can be els give accurate cloud simulations remains
excluded with high confidence through compar- unclear (Randall et al. 2007) as debate over spe-
isons with observations of volcanic relaxation cific processes continues (Spencer et al. 2007)
times and simulations of the last glacial maxi-
mum. As summarized by Randall et al. (2007) in Examples of competing hypotheses concerning
the Fourth IPCC Assessment, these multicon- high clouds (for which the infrared trapping ef-
straint studies are broadly consistent with the fects are large) are the IRIS hypothesis of
spread of sensitivity in the CMIP3 models.2 Lindzen, Chou, and Hou (2001) and the FAT
(Fixed Anvil Temperature Hypothesis) of Hart-
4.2 FEEDBACKS mann and Larsson (2002). The IRIS hypothesis
asserts that warmer temperatures cause the area
Better understanding of Earth’s climate sensi- coverage of clouds in the tropical upper tropo-
tivity, with potential reduction in its uncertainty, sphere to decrease, a negative feedback since
will require better understanding of a variety of these clouds are infrared absorbers. The FAT hy-
climate feedback processes (Bony et al. 2006). pothesis asserts that the altitude of these tropi-
We discuss some of these processes in more de- cal high clouds tends to increase with warming,
tail below. minimizing the temperature change at the cloud
tops—a positive feedback since the lack of
4.2.1 Cloud Feedbacks warming at cloud top prevents the increase in
outgoing radiation needed to balance the heat
Clouds reflect solar radiation to space, cooling trapping of greenhouse gases. Observational
the Earth-atmosphere system. Clouds also trap studies aimed at evaluating these mechanisms
infrared radiation, keeping the Earth warm. The are difficult because clouds in the tropics are
integrated net effect of clouds on climate de- strongly forced by circulations that are, in turn,
pends on their height, location, microphysical driven by temperature gradients and not by the
structure, and evolution through the seasonal local temperature in isolation. These circulation
and diurnal cycles. Cloud feedback refers to effects must be eliminated to isolate effects rel-
changes in cloud amounts and properties that evant to global warming. Very high resolution
can either amplify or moderate a climate simulations in localized regions have some po-
change. Differences in cloud feedbacks in cli- tential to address these questions. The FAT hy-
mate models have been identified repeatedly as pothesis, in particular, has received some
the leading source of spread in model-derived support from high-resolution modeling (Kuang
estimates of climate sensitivity (beginning with and Hartmann 2007).
Cess et al. 1990). The fidelity of cloud feed-
backs in climate models therefore is important Although these studies focus on high clouds, the
to the reliability of their prediction of future cli- intermodel differences in model responses of
mate change. low-level clouds are responsible for most of the
spread of cloud feedback values in climate mod-
Soden and Held (2006) evaluated cloud feed- els (Bony et al. 2006). While tempting, assum-
backs in 12 CMIP3 AOGCMs and found ing that this implies that low-cloud feedbacks
weakly to strongly positive cloud feedback in are more uncertain than high-cloud feedbacks
the various models. The highest values of cloud probably is premature. The strengths and weak-
feedback raise the equilibrium climate sensitiv- nesses of cloud-cover simulations for present-
ity (for CO2 doubling) from values of about 2 K day climate are described in Chapter 5.
to roughly 4 K. In comparison with the earlier
studies of Cess (1990) and Colman (2003), the As discussed in Chapter 6, a new class of much
spread of cloud feedbacks among GCMs has higher resolution global atmospheric simula-
become somewhat smaller over the years but is tions promises fundamental improvements in
still very substantial. Indeed, intermodel differ- cloud simulation. Using the surrogate climate
ences in cloud feedback are the primary reason change framework of Cess (1990) in which
45
ocean temperatures are warmed uniformly, quantity has been found between satellite ob-
Miura et al. (2005) carried out experiments servations and climate models constrained by
using a global model with 7-km resolution, ob- the observed ocean-surface temperatures
taining results suggestive of negative cloud (Soden 2000). These studies increase confi-
feedback outside the tropics, and Wyant et al. dence in the models’ vapor distributions more
(2006) describe results from a multigrid tech- generally, but column water vapor is dominated
nique in which high-resolution cloud models are by changes in the lower troposphere, whereas
embedded in each grid box of a traditional water-vapor feedback is strongest in the upper
GCM. Much work will be required with these troposphere where most outgoing terrestrial ra-
new types of models before they can be given diation to space originates. The results of Soden
substantial weight in discussions of the most and Held (2000) imply that at least half the
probable value for cloud feedback, but they sug- global water-vapor feedback arises from the
gest that real-world feedback is less positive tropical upper troposphere in models in which
than the typical CMIP3 AGCMs and that mid- relative humidity changes are small. Studies of
latitude cloud feedbacks may be more impor- vapor trends in this region are therefore of cen-
tant than hitherto assumed. Results from this tral importance. Soden et al. (2005) present
new generation of models will be of consider- analysis of radiance measurements, implying
able interest in the coming years. that relative humidity has remained unchanged
in the upper tropical troposphere over the past
Several questions remain to be answered about few years, which, combined with temperature
cloud feedbacks in GCMs. Physical mecha- measurements, provides evidence that water
nisms underlying cloud feedbacks in different vapor in this region is increasing.
models must be better characterized. How best
to judge the importance of model biases in sim- Observations of interannual variability in water
ulations of current climate and in simulations of vapor can help to judge the quality of model
cloud changes in different modes of observed simulations. Soden et al. (2002) concluded that
variability is not clear. In particular, how to a GCM appropriately simulates water-vapor
translate these biases into levels of confidence variations in the tropical upper troposphere dur-
in simulations of cloud feedback processes in ing cooling associated with the Pinatubo vol-
climate change scenarios is unclear. New satel- canic eruption. Minschwaner, Essler, and
lite products such as those from active radar and Sawaengphokhai (2006) compared the interan-
lidar systems should play a central role in cloud nual variability of humidity measured in the
research in coming years by providing more highest altitudes of the tropical troposphere with
comprehensive space-time cloud datasets. CMIP3 20th Century simulations. Both models
and observations show a small negative corre-
4.2.2 W ater-Vapor Feedbacks lation between relative humidity and tropical
temperatures, due in large part to lower relative
Analysis of radiative feedbacks in the CMIP3 humidity in warm El Niño years and higher rel-
models (Soden and Held 2006) reaffirms that ative humidity in cold La Niña years. However,
water-vapor feedback—the increase in heat there is a suggestion that the magnitude of this
trapping due to the increase in water vapor as covariation is underestimated in most models.
the lower atmosphere warms—is fundamental There also is a tendency for models with larger
to the models’ climate sensitivity. The strength interannual variations in relative humidity to
of their water-vapor feedback typically is close produce larger reductions in this region in re-
in magnitude to but slightly weaker than that ob- sponse to global warming, suggesting that this
tained by assuming that relative humidity re- deficiency in interannual variability might be
mains unchanged as the atmosphere warms. relevant for climate sensitivity. (This is another
example, analogous to the Qu and Hall (2006)
A trend toward increasing column water vapor analysis of snow feedback, in which the strength
in the atmosphere consistent with model pre- of a feedback in models is correlated with a
dictions has been documented from microwave more readily observed aspect of climatic vari-
satellite measurements (Trenberth, Fasullo, and ability.) In short, the study of Minschwaner,
Smith 2005), and excellent agreement for this Essler, and Sawaengphokhai (2006) suggests
46
that water-vapor feedback in the very highest in the literature is complex because of differing
levels of the tropical troposphere may be over- calculations in different papers. An important
estimated in models, but it does not imply that objective for the climate modeling community
a significant correction is needed to the overall is to improve the consistency of its reporting of
magnitude of the feedback. radiative forcing in models.
Positive water-vapor feedback, resulting from 4.3.1 Greenhouse Gases

increases in vapor that keep the relative humid-
ity from changing substantially as the climate Greenhouse gases like carbon dioxide and
warms, has been present in all GCMs since the methane have atmospheric lifetimes that are
first simulations of greenhouse gas–induced long, compared to the time required for these
warming (Manabe and Wetherald 1975). It rep- gases to be thoroughly mixed throughout the at-
resents perhaps the single most robust aspect of mosphere. Trends in concentration, consistent
global warming simulations. Despite the fact throughout the world, have been measured rou-
that the distribution of water vapor in the at- tinely since the International Geophysical Year
mosphere is complex, we are aware of no ob- in 1958. Measurements of gas bubbles trapped
servational or modeling evidence that casts in ice cores give the concentration prior to that
doubt of any significance on this basic result, date (with less time resolution). Nevertheless,
and we consider the increase in equilibrium sen- the associated radiative forcing varies somewhat
sitivity to roughly 2ºC from this feedback to be among climate models because GCM radiative
a solid starting point from which the more un- calculations must be computationally efficient,
certain cloud feedbacks then operate. necessitating approximations that make them
less accurate than the best laboratory spectro-
4.3 T W EN T IET H CEN T U RY scopic data and radiation algorithms. Using
RADIAT IVE FORCIN G changes in well-mixed greenhouse gases meas-
ured between 1860 and 2000, Collins et al.
Radiative forcing is defined as a change that af- (2006b) compared the radiative forcing of cli-
fects the Earth’s radiation balance at the top of mate models (including CCSM, GFDL, and
the tropopause between absorbed energy re- GISS) with line-by-line (LBL) calculations in
ceived in the form of solar energy and emitted which fewer approximations are made. The me-
infrared energy to space, typically expressed in dian LBL forcing at the top of the model by
terms of changes to the equilibrium preindus- greenhouse gases is 2.1 W/m2, and the corre-
trial climate. Uncertainties in 20th Century ra- sponding median among the climate models is
diative forcing limit the precision with which higher by only 0.1 W/m2. However, the standard
climate sensitivity can be inferred from ob- deviation among model estimates is 0.30 W/m2
served temperature changes. In this section, we (compared to 0.13 for the LBL calculations).
briefly discuss the extent to which models pro- Based on these most-recent comparisons with
vide consistent and reliable estimates of radia- LBL computations, we can reasonably assume
tive forcing over the 20th Century. Further that radiative forcing due to carbon dioxide dou-
information is provided by Forster et al. (2007). bling in individual climate models may be in
error by roughly 10%.
Radiative forcing in models can be quantified
in different ways, as outlined by Hansen et al. 4.3.2 Other Forcings
(2005). For example, the radiative forcing for
the idealized case of CO2 doubling can be com- While increases in the concentration of green-
puted by (1) holding all atmospheric and sur- house gases provide the largest radiative forc-
face temperatures fixed, (2) allowing the ing during the 20th Century, other smaller
stratospheric temperatures to adjust to the new forcings must be considered to quantitatively
CO2 levels, (3) fixing surface temperatures over model the observed change in surface air tem-
both land and ocean and allowing the atmos- perature. The burning of fossil fuels that release
phere to equilibrate, or (4) fixing ocean tem- greenhouse gases into the atmosphere also pro-
peratures only and allowing the atmosphere and duces an increase in atmospheric aerosols
land to equilibrate. Comparing model forcings (small liquid droplets or solid particles that are
47
temporarily suspended in the atmosphere). the IPCC AR4 models, although, among the
Aerosols cool the planet by reflecting sunlight U.S. CMIP3 models, it was included in GISS
back to space. In addition, among other forcings ModelE where increased cloud cover due to
are changes in land use that alter the reflectivity aerosols results in a 20th Century forcing of –
of the Earth’s surface, as well as variations in 0.8 W/m2 (Hansen et al. 2007).
sunlight impinging on the Earth.
4.3.2.2 VARIABILITY O F SO LAR IRRAD IAN CE AN D
4.3.2.1 A ERO SO LS VO LC AN IC A ERO SO LS
Aerosols have short lifetimes (on the order of a Other climate forcings include variability of
week) that prevent them from dispersing uni- solar irradiance and volcanic aerosols. Satellites
formly throughout the atmosphere, in contrast provide the only direct measurements of these
to well-mixed greenhouse gases. Consequently, quantities at the top of the atmosphere. Satellite
aerosol concentrations have large spatial varia- measurements of solar irradiance are available
tions that depend on the size and location of from the late 1970s and now span about 3 of the
sources as well as changing weather that dis- sun’s 11-year magnetic or sunspot cycles. Ex-
perses and transports the aerosol particles. tracting a long-term trend from this relatively
Satellites can provide the global spatial cover- brief record (Wilson et al. 2003) is difficult.
age needed to observe these variations, but Prior to the satellite era, solar variations are in-
satellite instruments cannot distinguish between ferred using records of sunspot area and number
natural and anthropogenic contributions to total and cosmic ray–generated isotopes in ice cores
aerosol forcing. The anthropogenic component (Foukal et al. 2006), which are converted into
can be estimated using physical models of irradiance variations using empirical relations
aerosol creation and dispersal constrained by The U.S. CMIP3 models all use the solar re-
available observations. construction by Lean, Beer, and Bradley (1995)
with subsequent updates.
Satellites increasingly are used to provide ob-
servational estimates of the “direct effect” of Volcanic aerosols prior to the satellite era are in-
aerosols on the scattering and absorption of ra- ferred from surface estimates of aerosol optical
diation. These estimates range from –0.35 +/– depth. The radiative calculation requires aerosol
0.25 W/m2 (Chung et al. 2005) to –0.5 +/– 0.33 amount and particle size, which is inferred
W/m2 (Yu et al. 2006) to –0.8 +/– 0.1 W/m2 using empirical relationships with optical depth
(Bellouin et al. 2005). The fact that two of these derived from recent eruptions. The GFDL and
three estimates do not overlap suggests incom- GISS models use updated versions of the Sato
plete uncertainty analysis in these studies. In et al. (1993) eruption history, while the CCSM
particular, each calculation must decide how to uses Ammann et al. (2003). As with solar vari-
extract the anthropogenic fraction of aerosol. ability, different reconstructions of volcanic
Global direct forcing by aerosols is estimated forcing differ substantially (see, e.g., Lindzen
by the IPCC AR4 as –0.2 +/– 0.2 W/m2, ac- and Giannitsis 1998). Land-use changes also are
cording to models, and –0.5 +/– 0.4 W/m2, uncertain, and they can be of considerable sig-
based upon satellite estimates and models. This nificance locally. Global models, however, typ-
central estimate is smaller in magnitude than the ically show very modest global responses, as
2001 IPCC estimate of –0.9 +/– 0.5 W/m2. discussed in Hegerl et al. (2007).
In addition to their direct radiative forcing, Studies attributing 20th Century global warming
aerosols also act as cloud condensation nuclei. to various natural and human-induced forcing
Through this and other mechanisms, they alter changes clearly are hindered by these uncer-
the radiative forcing of clouds (Twomey 1977; tainties in radiative forcing, especially in the
Albrecht 1989; Ackerman et al. 2004). Complex solar and aerosol components. The trend in total
interactions among aerosols and cloud physics solar irradiance during the last few decades (av-
make this “aerosol indirect effect” very difficult eraging over the sun’s 11-year cycle) apparently
to measure, and model estimates of it vary is negative and thus cannot explain recent global
widely. This effect was generally omitted from warming (Lockwood and Fröhlich 2007). The
48
connection between solar energy output found in general that models with lower ocean-
changes and the warming earlier in the 20th Cen- uptake efficiency had lower climate sensitivity,
tury is more uncertain. With the solar recon- as expected (Hansen et al. 1985; Wigley and
structions assumed in the CMIP3 models, much Schlesinger 1985). Uptake efficiency can be
of the early 20th Century warming is driven by thought of as the amount of heat the ocean ab-
solar variations, but uncertainties in these re- sorbs through mixing relative to the change in
constructions do not allow confident attribution surface temperature (e.g., to reproduce the ob-
statements concerning this early-century warm- served 20th Century warming despite a high cli-
ing. The large uncertainties in aerosol forcing mate sensitivity, a model needs large heat export
are a more important reason that the observed to the deep ocean). Comparing the current gen-
late 20th Century warming cannot be used to eration of AOGCMs with the previous genera-
provide a sharp constraint on climate sensitivity. tion, however, Kiehl et al. (2006) found that the
We do not have good estimates of the fraction of atmospheric component of the models is the pri-
greenhouse gas forcing that has been offset by mary reason for different transient climate sen-
aerosols. sitivities, and the ocean component’s ability to
uptake heat is of secondary importance. Ocean
4.4 OCEAN H EAT U PTAKE AN D heat-uptake efficiency values calculated in this
CLIMAT E SEN SIT IVIT Y study differ substantially from those in Raper et
al. (2002).
As noted above, the rate of heat uptake by the
ocean is a primary factor determining transient Despite these complexities, modern ocean
climate response (TCR): the larger the heat up- GCMs are able to transport both heat
take by the oceans, the smaller the initial re- (AchutaRao et al. 2006) and passive tracers
sponse of Earth’s surface temperature to such as chlorofluorocarbons and radiocarbon
radiative forcing (e.g., Sun and Hansen 2003). (Gent et al. 2006; Dutay et al. 2002) consistent
Studies show (e.g., Völker, Wallace, and Wolf- with the limited observations available for these
Gladrow 2002) that CO2 uptake by the ocean quantities. Better observations in the future—
also is linked to certain factors that control heat particularly of the enhanced ocean warming ex-
uptake, albeit not in a simple fashion. In an pected from the anthropogenic greenhouse
AOGCM, the ocean component’s ability to take effect—should provide stronger constraints on
up heat depends on vertical mixing of heat and modeled ocean transports.
salt and how the model transports heat between
low latitudes (where heat is taken up by the 4.5 IMPACT OF CLIMAT E
ocean) and high latitudes (where heat is given SEN SIT IVIT Y ON U SIN G MODEL
up by the ocean). The models make use of sev- PROJECT ION S OF FU T U RE
eral subgrid-scale parameterizations (see Chap- CLIMAT ES
ter 2), which have their own uncertainties. Thus,
as part of understanding a model’s climate-sen- This chapter has emphasized the global and an-
sitivity value, we must assess its ability to rep- nual mean of surface temperature change even
resent the ocean’s mixing processes and the though practical applications of climate change
transport of its heat, as well as feedbacks among science involve particular seasons and loca-
the ocean, ice, and atmosphere. tions. The underlying assumption is that local
climate impacts scale with changes in global
The reasons for differing model estimates of mean surface temperature (Santer et al. 1990).
ocean uptake are incompletely understood. As- In that case, time histories of global mean tem-
sessments typically compare runs of the same perature—obtained from a simple model of
model or output from different AOGCMs. global mean temperature, run under a variety of
Raper, Gregory, and Stouffer (2002) examined forcing scenarios—could be combined with a
climate sensitivity and ocean heat uptake in a single AOGCM-produced map of climate
suite of then-current AOGCMs. They calculated change normalized to the global mean surface
the ratio of the change in heat flux (from the temperature change. In that way, the regional
surface to the deep ocean) to the change in tem- changes expected for many different climate-
perature (Gregory and Mitchell 1997) and forcing scenarios could be obtained from just
49
one AOGCM simulation using one idealized

forcing scenario such as atmospheric CO2 dou-
bling (Oglesby and Saltzman 1992) or 1% per
year increasing CO2 (Mitchell et al. 1999). This
“pattern scaling” assumption also permits the
gauging of effects on regional climate change
that arise from different estimates of global cli-
mate sensitivity. For example, if an AOGCM
with TCR = 1.5 K predicts temperature and pre-
cipitation changes ΔT and ΔP as a function of
season and location in a 21st Century climate
simulation, and if investigators believe that TCR
= 1.0 K is a better estimate of the real world’s
climate sensitivity, then, under the pattern-scal-
ing assumption, they would reduce the local ΔT
and ΔP values by 50%.
Although it introduces its own uncertainties, the

pattern-scaling assumption increasingly is used
in climate impact assessments (e.g., Mitchell
2003; Ruosteenoja, Tuomenvirta, and Jylha
2007). For example, the annual mean tempera-
ture change averaged over the central United
States during the 21st Century for any of the
projections in the IPCC Special Report on
Emissions Scenarios shows that about 75% of
the variance among the CMIP3 models is ex-
plained by their differing global mean warming
(B. Wyman, personal communication). (The
central United States is defined in this context
following Table 11.1 in Christensen et al. 2007.)
Precipitation patterns, in contrast, do not scale
as well as temperature patterns due to sharp
variations between locally decreasing and lo-
cally increasing precipitation in conjunction
with global warming.
50
5CH APTER
M odel Simulation
of Major Climate Features
Although a typical use of atmosphere-ocean general circulation model (AO GCM) output for cli-
mate impact assessment focuses on one particular region such as a river basin or one of the 50
United States, knowing model simulation overall accuracy on continental to global scales is im-
portant. Fidelity in simulating climate on the largest scales is a necessary condition for credible
predictions of future climate on smaller scales. Model developers devote great effort to assess-
ing the level of agreement between simulated and observed large-scale climate, both for the pres-
ent day and for the two centuries since the Industrial Era began. Unlike physical theories of such
fundamentally simple systems as the hydrogen atom, AO GCMs cannot promise precise accuracy
for every simulated variable on all relevant space and time scales. N evertheless, before applying a
model to a practical question, users should demand reasonable overall agreement with observa-
tions, with the definition of “reasonable” in part subjective and dependent on the problem at
hand. H ere we provide an overview of how well modern AO GCMs satisfy this criterion.
5.1 MEAN SU RFACE ing” of the models’ energy balance as described

T EMPERAT U RE AN D in Chapter 2 and by itself is not a stringent test
PRECIPITAT ION of model quality. More relevant is consideration
of space and time variations about the global an-
Simulations of monthly near-surface air tem- nual mean (including the seasonal cycle). The
perature and precipitation provide a standard overall correlation pattern between simulations
starting point for model evaluation since these and observations typically is 95 to 98%, and
fields are central to many applications. The two variation magnitudes typically agree within
fields also illustrate the difficulty in designing ±25% (Covey et al. 2003). This level of success
appropriate metrics for measuring model quality. has been retained in the latest generation of
models that allow ocean and atmosphere to ex-
By most measures, modern AOGCMs simulate change heat and water without artificial adjust-
the basic structure of monthly mean near-sur- ments (Randall et al. 2007). Nevertheless, as
face temperatures quite well. The globally aver- shown below, local errors in surface tempera-
aged annual mean value generally lies within ture that are clearly outside the bounds of ob-
the observed range (~286 to 287 K) of modern servational uncertainty persist in the latest
and preindustrial values; this agreement, how- generation of models.
ever, is in part a consequence of the “final tun-
51
The U.S. Climate Change Science Program Chapter 5 - Model Simulation of Major Climate Features
AOGCM simulations are considerably less ac- For illustration, we show examples from two of
curate for monthly mean precipitation than for the U.S. models discussed in Chapter 4. In Fig.
temperature. The space-time correlation be- 5.1 (Delworth et al. 2006) and Fig. 5.2 (Collins
tween models and observations typically is only et al. 2006a), simulated and observed maps of
about 50 to 60% (Covey et al. 2003). As we dis- surface temperature and even precipitation ap-
cuss below, these poor correlations originate pear rather similar at first glance. Constructing
mainly in the tropics, where precipitation varies simulated-minus-observed difference maps,
greatly over relatively small ranges of latitude however, reveals monthly and seasonal mean
and longitude. Strong horizontal gradients in the temperature and precipitation errors up to 10°C
field lead to a significant drop in correlations and 7 mm/day, respectively, at some points.
with observations, even with only slight shifts CCSM3 temperature-difference maps exhibit
in the modeled precipitation distribution. These the largest errors in the Arctic (note scale
modest correlations are relevant for precipita- change in Fig. 5.2d), where continental winter-
tion at a particular location, but AOGCMs gen- time near-surface temperature is overestimated.
erally reproduce the observed broad patterns of AOGCMs find this quantity particularly diffi-
precipitation amount and year-to-year variabil- cult to simulate because, for land areas near the
ity (see Fig. 5.1 and Dai 2006). One prominent poles in winter, models must resolve a strong
error is that models without flux adjustment temperature inversion above the surface (warm
typically fail to simulate the observed north- air overlying cold air). For precipitation, GFDL
west-to-southeast orientation of a large region difference maps reveal significant widespread
of particularly heavy cloudiness and precipita- errors in the tropics, most notably in the ITCZ
tion in the southwest Pacific Ocean. Instead, region discussed above and in the Amazon
these models tend to rotate this convergence River basin, where precipitation is underesti-
zone into an east-west orientation, producing an mated by several millimeters per day. Similar
unrealistic pair of distinct, parallel convection precipitation errors appear in CCSM3 results
bands straddling the equator instead of a con- (e.g., a 28% underestimate of Amazon annual
tinuous Inter-Tropical Convergence Zone mean). AOGCM precipitation errors have seri-
(ITCZ). The double-ITCZ error has been frus- ous implications for Earth system models with
tratingly persistent in climate models despite interactive vegetation, because such models use
much effort to correct it. simulated precipitation to calculate plant growth
(see Chapter 6). Errors of this magnitude would
Another discrepancy between models and ob- produce an unrealistic distribution of vegetation
servations appears in the average day-night in an Earth system model, for example, by spu-
cycle of precipitation. While the model’s diurnal riously deforesting the Amazon basin.
temperature cycle exhibits general agreement
with observations, simulated cloud formation In summary, modern AOGCMs generally simu-
and precipitation tend to start too early in the late continental and larger-scale mean surface
day. Also, when precipitation is sorted into light, temperature and precipitation with considerable
moderate, and heavy categories, models repro- accuracy, but the models often are not reliable
duce the observed extent of moderate precipi- for smaller regions, particularly for precipita-
tation (10 to 20 mm/day) but underestimate that tion.
of heavy precipitation and overestimate the ex-
tent of light precipitation (Dai 2006). Additional
model errors appear when precipitation is stud-
ied in detail for particular regions [e.g., within
the United States (Ruiz-Barradas and Nigam
2006)].
52
O bserved
(a)
80°N
40°N
0°
40°S
80°S
150°E 110°W 10°W 90°E
CM2.0 CM2.1
(b) (d)
80°N
40°N
0°
40°S
80°S
150°E 110°W 10°W 90°E 150°E 110°W 10°W 90°E
0 1 2 3 4 5 10 15 20
CM2.0 minus observed CM2.1 minus observed

(c) (e)
80°N
40°N
0°
40°S
80°S
150°E 110°W 10°W 90°E 150°E 110°W 10°W 90°E
–7 –5 –3 –1.75 –1.25 –0.75 –0.25 0.25 0.75 1.25 1.75 3 5 7
Figure 5.1a–e. Observed and GFDL Model-Simulated Precipitation (mm/day).

O bserved image from P. X ie and P.A. Arkin 1997: Global precipitation:A 17-year monthly analysis based on gauge observations, satellite
estimates, and numerical model outputs. Bulletin American M eteorological Society, 78, 2539–2558. [O ther images from Fig. 17 in T.L.
D elworth et al. 2006: GFD L’s CM2 global coupled climate models. Part I: Formulation and simulation characteristics. J. Climate, 19, 643–
674. Images reproduced with permission of the American Meteorological Society.]
53
CCSM3 (yrs 400–499) DJF

2- meter Temp (land) mean = 276.91 K
Min = 236.83 Max = 305.62
310
305
300
295
290
285
280
275
270
260
250
240
230
220
210
WILLMOTT
2- meter Temp (land) mean = 276.31 K
Min = 220.88 Max = 305.82
310
305
300
295
290
285
280
275
270
260
250
240
230
220
210
CCSM3 – WILLMOTT
mean = 0.84 r mse = 3.58 K
Min = –16.12 Max = 20.98
12
10
8
6
4
2
0
–2
–4
–6
–8
–10
–12
–14
Figure 5.2a–c. CCSM3 December-January-February Simulated (top panel), Observed (middle panel), and
Simulated-Minus-Observed (bottom panel) N ear-Surface Air Temperature for Land Areas (°C).
N ote change in scale from 5.2a to 5.2c. [Figures from W. Collins et al. 2006:The Community Climate System Model Version 3 (CCSM3).
J. Climate, 19(11), 2122–2143. Reproduced with permission of the American Meteorological Society.]
54
Figure 5.2d. CCSM3 Annual Mean Simulated-Minus-Observed Sea Surface

Temperature (°C).
[Figure from W. Collins et al. 2006:The Community Climate System Model Version 3 (CCSM3). J. Climate,
19(11), 2122–2143. Reproduced with permission of the American Meteorological Society.]
5.2 T W EN T IET H CEN T U RY temperature and salinity, are not known for
T REN DS 1860. The spread among individual simulations
from the same model (the dotted-line curves)
Modern AOGCMs are able to simulate not only thus indicates uncertainty in model-simulated
the time-average climate but also changes temperature arising from lack of knowledge
(trends) in climate over the past 140 years. For about initial conditions.
example, Fig. 5.3 shows results from the three
U.S. models and the “average” CMIP3 model. These results demonstrate that modern climate
Plotted in the figure are curves of globally av- models exhibit agreement with observed global
eraged annual mean near-surface temperature mean near-surface temperature trends to within
from model simulations and the observational observational uncertainty, despite imprecise ini-
value as determined from the U.K. Climatic Re- tial conditions and uncertain climate forcing
search Unit (CRU) gridded observational data- and heat uptake by the deep ocean (Min and
base. Two curves are plotted for the CMIP3 Hense 2006). Models achieve this agreement
models. The first shows the average over all only if they include anthropogenic emissions of
CMIP3 models, and the second, the average greenhouse gases and aerosols. No plausible
over only CMIP3 models that included the ef- combination of natural climate-forcing factors
fects of volcanic eruptions. Results from indi- allows models to explain the global warming
vidual U.S. models are shown for separate observed over the last several decades. Indirect
ensemble members (dotted lines) and for the av- solar effects [e.g., involving cosmic rays and
erage over all ensemble members (continuous clouds (Svensmark 2007)] are not generally in-
lines). Individual members of a particular model cluded in AOGCM simulations. These effects
ensemble differ from each other because they have been proposed occasionally as causes of
were run from different initial conditions. Pre- global warming, although over the past 20 years
cise initial conditions, especially deep-ocean their trends would, if anything, lead to cooling
55
(Lockwood and Fröhlich 2007). Unless the choices. In contrast to simple energy-balance
models grossly underestimate the climate sys- models that predict only the global mean tem-
tem’s natural internally generated variability or perature using a limited representation of cli-
are all missing a large unknown forcing agent, mate physics, an AOGCM’s climate sensitivity
the conclusion is that most recent warming is is difficult to specify a priori. More fundamen-
anthropogenic (IPCC 2007b). tally, AOGCMs, unlike simpler climate models,
have far fewer adjustable parameters than the
Nevertheless, total climate forcing during the number of observations available for model
20th Century is not accurately known, especially evaluation (Randall et al. 2007). Thus, an
the aerosol component (see Chapter 2). Aerosol AOGCM’s multidimensional output can be
forcing used in these simulations, however, is compared to observations independent of this
derived from aerosol parameterizations con- adjustment (e.g., using observed trends in re-
strained by satellite and ground-based measure- gional temperature). Agreement between mod-
ments of the aerosols themselves and was not eled and observed trends has been described for
designed to obtain a fit to observed global mean temperature trends on each inhabited continent
temperature trends. The observed trend in sur- (Min and Hense 2007); for trends in climate ex-
face temperature can result from models with tremes, such as heat-wave frequency and frost-
different aerosol forcing (Schwartz 2007). Thus, day occurrence (Tebaldi et al. 2006); and for
20th Century temperature records cannot distin- trends in surface pressure and Arctic sea ice (see
guish models that would warm by differing Chapter 9 in IPCC 2007), all of which comple-
amounts for the same total forcing. ment comparisons between modeled and ob-
served time-averaged climate discussed in the
Note that climate sensitivity is not prescribed in following sections.
AOGCMs. Instead, this sensitivity emerges as a
result of a variety of lower-level modeling
Figure 5.3a. Simulation Global Warming relative to 1900 for gfdl_cm2_1

of 20th Century (10 years r unning average smoothed)
Globally Averaged 1
IPCC Mean Volc
Surface Temperature IPCC Mean
from GFDL CM2.1. r un1
r un2
“CRU” is the value based on r un3
Mean
the Climate Research Unit CRU
0.5
gridded observational
dataset, “IPCC Mean” is the
average value of all CMIP3
models, and “IPCC Mean
tas
Volc” is the average of all

0
CMIP3 models that included
volcanic forcing. Individual
realizations of the CMIP3 20th
Century experiment are
denoted by the dotted
curves labeled “run(1–3),” -0.5
and the ensemble mean is
marked “Mean.”
00
80
00
0
90
80
40
90
20
60
70
50
30
70
10
6
20
18
18
18
18
19
19
19
19
19
19
19
19
19
19
Time
56
Global Warming relative to 1900 for giss_model_e_r

Figure 5.3b. Simulation
(10 years r unning average smoothed)
1 of 20th Century
IPCC Mean Volc
IPCC Mean
Globally Averaged
r un1 Surface Temperature
r un2
r un3
from GISS Model E-R.
r un4 Curve labels are the same as
0.5 r un5
r un6 in Fig. 5.3a.
r un7
r un8
r un9
tas
Mean
CRU
0
-0.5
00
80
00
0
90
80
40
90
20
60
70
50
30
70
10
6
20
18
18
18
18
19
19
19
19
19
19
19
19
19
19
Time
Global Warming relative to 1900 for ncar_ccsm3_0 Figure 5.3c. Simulation

(10 years r unning average smoothed)
1
of 20th Century
IPCC Mean Volc Globally Averaged
IPCC Mean
r un1 Surface Temperature
r un2 from CCSM3.
r un3
r un4 Curve labels are the same as
0.5 r un5
r un6
in Fig. 5.3a.
r un7
r un9
Mean
tas
CRU
-0.5
00
80
00
0
90
80
40
90
20
60
70
50
30
70
10
6
20
18
18
18
18
19
19
19
19
19
19
19
19
19
19
Time
Global Warming Relative to 1900 for American Models Figure 5.3d.

(10 years r unning average smoothed) Comparison of
1
IPCC Mean Volc Simulations of 20th
IPCC Mean
gfdl_cm2_1 Century Globally
giss_ model_e_r Averaged Surface
ncar_ccsm3_0 Temperature from the
CRU
0.5 Three U.S. CMIP3
Models.
Model curves represent
tas
ensemble means for CCSM3

(ncar_ccsm3_0), GISS Model
0
E-R (giss_e_r), and GFD L
CM2.1 (gfdl_cm2_1). “CRU,”
“IPCC Mean,” and “IPCC
Mean Volc” labels are the
-0.5 same as in Fig. 5.3a.
00
80
00
0
90
80
40
90
20
60
70
50
30
70
10
6
20
18
18
18
18
19
19
19
19
19
19
19
19
19
19
Time
57
As an example of 20th Century temperature gional trends than in the global average, where
trends on continental-to-global spatial scales uncorrelated fluctuations in different regions
and multidecadal time scales, Fig. 5.4 shows tend to cancel. For both 20th Century warming
global maps for different time periods between periods, the model simulates, but underesti-
1880 and 2003 as observed and simulated by mates, the high-latitude amplification of global
GISS ModelE (Hansen et al. 2006; also see warming. Additional discrepancies between
Knutson et al. 2006). The figure shows general AOGCMs and observations appear at smaller
agreement between model and observations not scales. For example, model-simulated trends do
only for the overall period but also for segments not consistently match the observed lack of 20th
1880 to 1940 and 1979 to 2003, which encom- Century warming in the central United States
pass periods of early and late 20th Century (Kunkel et al. 2006).
warming. For 1940 to 1979, the model simu-
lates only a small change in global mean tem- 5.2.1 Trends in Vertical Temperature
perature in agreement with observations, but it
fails to simulate the strong north polar cooling While models simulate the 20th Century warm-
observed for this period. As a result, the modeling observed at the surface, agreement is less
simulated global mean-temperature change obvious with tropospheric observations from
(upper right corner of each frame) is slightly satellites and weather balloons. This issue was
positive rather than slightly negative as ob- the focus of CCSP SAP 1.1 (CCSP 2006). Since
served. Part of this discrepancy may result from 1979 (beginning of the satellite record), glob-
chaotic fluctuations within observed climate ally averaged warming in the troposphere ac-
that the model cannot synchronize correctly due cording to climate models is within the range of
to inprecise knowledge of the initial conditions available observations. Within the tropics, the
in the 19th Century period. These chaotic fluc- model-simulated troposphere warms more rap-
tuations generally are more important in re- idly than observed (see CCSP 2006, Fig. 5.4 F–
Surface Temperature Change Based on Local Linear Trends (°C)

1880-2003 1880-1940 1940-1979 1979-2003
O bser vatio ns .60 .24 .06 .39
All Forcings 55 26 07 32
–2 –1 –.6 –.1 .1 .3 .6 1 1.5 3
Figure 5.4. N ear-Surface Temperature Changes as Observed (top panels) and as

Simulated by GISS ModelE (bottom panels) for Selected Time Periods Between 1880
and 2003.
N umbers above upper right panel corners are global means. [Images from Fig. 9 in J. H ansen et al. 2007:
Climate simulations for 1880–2003 with GISS ModelE. Climate D ynamics, 29(7–8), 661–696. Reproduced
with kind permission of Springer Science and Business Media.]
58
G). SAP 1.1 noted, however, that “Large struc- plications for many aspects of model projec-
tural uncertainties in the observations . . . make tions in the tropics.
it difficult to reach more definitive conclusions
regarding the significance and importance of 5.2.2 Model Simulation of Observed
model-data discrepancies” (CCSP 2006, p. 112 Climate Variability
and Section 5.4).
The following sections discuss a number of spe-
Research since publication of SAP1.1 has con- cific climate phenomena directly or indirectly
tinued to highlight uncertainties implicit in related to near-surface temperature, precipita-
measuring the difference between surface and tion, and sea level. Numerous studies of climate
lower-atmospheric warming. For example, change have focused on one or two of these phe-
Thorne et al. (2007) found that the tropical at- nomena, so a great deal of information (and oc-
mosphere-to-surface warming ratio in both ob- casional debate) has accumulated for each of
servations and model simulations is sensitive to them. Here we attempt to summarize the points
the time period analyzed. Meanwhile, debate that would best give users of AOGCM model
continues over the best way to process data from output a general sense of model reliability or
satellites (Christy et al. 2007) and weather bal- unreliability. Although the following sections
loons (Christy and Spencer 2005). AOGCMs individually note different types of climate vari-
continue to differ from most published obser- ation, the reader should recognize that the total
vations on the ratio of atmosphere-to-surface amount of natural climate variability forms
warming in the tropics since the beginning of background “noise” that must be correctly as-
satellite observations (e.g., as shown by Thorne sessed to identify the “signal” of anthropogenic
et al. 2007, Fig. 3), with the ratio being larger in climate change. Natural variability in turn sep-
the models than is seen in decadal observational arates into an externally forced part (e.g., from
trends. solar energy output and volcanic eruptions) and
internally generated variability just as weather
Paradoxically, trends are more consistent be- varies on shorter time scales because of the sys-
tween models and observations on interannual tem’s intrinsic chaotic character. As noted
time scales. AOGCM simulation of tropical at- above, long-term trends in both solar and vol-
mospheric warming involves mainly subgrid- canic forcing during the past few decades have
scale parameterizations. As discussed in had a cooling rather than warming effect. It fol-
Chapter 2, these are not as trustworthy as ex- lows that if global warming during this period is
plicitly computed processes, but internal vari- not anthropogenic, then the climate system’s in-
ability [primarily due to El Niño–Southern ternal variation is the most likely alternative ex-
Oscillation (ENSO)] provides a useful test of planation.
the models’ ability to redistribute heat realisti-
cally. AOGCMs simulate very well the portion Control runs of AOGCMs (in which no changes
of tropical temperature trends due to interannual in external climate forcing are included) provide
variability (Santer et al. 2005). In addition, ex- estimates of the level of internally generated cli-
plaining how atmospheric water vapor increases mate variability. Control runs generally obtain
coincidentally with surface temperature is dif- realistic near-surface temperature variability on
ficult (Trenberth, Fasullo, and Smith 2005; San- annual-to-decadal time scales, although they
ter et al. 2007; Wentz et al. 2007) unless lower typically underestimate variability in areas of
tropospheric temperature also increases coinci- the Pacific and Indian Ocean where ENSO and
dentally with surface temperature. While defi- the Pacific Decadal Oscillation (PDO) (see
ciencies in model subgrid-scale parameter- below) predominate (Stouffer, Hegerl, and Tett
izations are certainly possible, trends in poorly 2000). Unfortunately, the longest time periods
documented forcing agents (see Chapter 4) may that are directly relevant to separating natural
prove important in explaining the discrepancy from anthropogenic climate change are the least
over the longer time scales. Future research is observed. Assessing variations of surface tem-
required to resolve the issue because tropos- perature for time periods longer than 50 to 100
pheric observations at face value suggest a trend years depends on paleodata such as ice-core
toward greater tropical instability, which has im- composition and tree-ring thickness. Interpreta-
59
tion of these data is made difficult by sparse ge- tracted before computing variances. In each
ographical coverage and also is complicated by case, eddy statistics are compared to estimates
natural variations in external climate forcing. of observed statistics obtained from
NCEP/NCAR Reanalysis (B.Wyman, personal
5.2.2.1 EX TRA-T RO PIC AL STO RMS
communication). When analyzing eddy statis-
Climate models have developed from numeri- tics, the data are typically filtered to retain only
cal weather-prediction models whose perform- those time scales, roughly 2 to 10 days, associ-
ance has been judged primarily on their ability ated with midlatitude weather systems. The two
to forecast midlatitude weather. The success of quantities chosen here, however, are sufficiently
forecast models in their simulation of midlati- dominated by these time scales that they are rel-
tude cyclones and anticyclones has resulted in atively insensitive to the monthly filtering used
continuous growth in the value of numerical here. In winter, Northern Hemisphere storms
weather prediction. The ability of GCMs to gen- are organized into two major oceanic storm
erate realistic statistics of midlatitude weather tracks over the Pacific and Atlantic oceans. His-
also has been central in climate model develop- torically, atmospheric models of horizontal res-
ment. This is true not only because midlatitude olutions of 200 to 300 km typically are capable
weather is important in its own right, but also of simulating midlatitude storm tracks with re-
because these storms are the primary mecha- alism comparable to that shown in the figure.
nism by which heat, momentum, and water Eddy amplitudes often are a bit weak and often
vapor are transported by the atmosphere, mak- displaced slightly equatorward. In spectral mod-
ing their simulation crucial for simulation of els with resolution coarser than 200 to 300 km,
global climate. Indeed, a defining feature of at- simulation of midlatitude storm tracks typically
mospheric general circulation models deteriorates significantly (see, e.g., Boyle
(AGCMs) is that they compute midlatitude eddy 1993). General improvements in most models
statistics and associated eddy fluxes through ex- in the CMIP3 database over previous genera-
plicit computation of the life cycles of individ- tions of models, as described in Chapter 1, are
ual weather systems and not through some thought to be partly related to the fact that most
turbulence or parameterization theory. Comput- of these models now have grid sizes of 100 to
ing the evolution of individual eddies may seem 300 km or smaller. Although even-finer resolu-
very inefficient when primary interest is in tion results in better simulations of midlatitude-
long-term eddy statistics, but the community storm structure, including that of warm and cold
clearly has judged for decades that explicit eddy fronts and interactions among these storms and
simulation in climate models is far superior to coastlines and mountain ranges, improvements
attempts to develop closure theories for eddy in midlatitude climate on large scales tend to be
statistics. The latter theories typically form the less dramatic and systematic. Other factors be-
basis for Earth system models of intermediate sides horizontal resolution are considered im-
complexity (EMICs), which are far more effi- portant for details of storm track structure. Such
cient computationally than GCMs but provide factors include distribution of tropical rainfall,
less convincing simulations. which is sensitive to parameterization schemes
used for moist convection, and interactions be-
Two figures illustrate the quality of simulated tween stratosphere and troposphere, which are
midlatitude eddy statistics from coupled sensitive to vertical resolution. Roeckner et al.
AOGCMs used in IPCC AR4. Shown for the (2006), for example, illustrate the importance
GFDL CM2.1 in Fig. 5.5a is wintertime vari- of vertical resolution for midlatitude circulation
ance of the north-south velocity component at and storm track simulation.
300 hPa (in the upper troposphere). This quan-
tity represents the magnitude of variability in Lucarini et al. (2006) provide a more detailed
the upper troposphere associated with day-to- look at the ability of CMIP3 models to simulate
day weather. In Fig. 5.5b, the wintertime pole- the space-time spectra of observed eddy statis-
ward eddy heat flux or covariance between tics. These authors view the deficiencies noted,
temperature and north-south velocity is shown which vary in detail from model to model, as
at 850 mb (in the lower troposphere). For these serious limitations to model credibility. As in-
calculations, the monthly means were sub- dicated in Chapter 1, however, our ability is lim-
60
ited in translating measures of model biases into tations in climate projection credibility, note
useful measures of model credibility for 21st that the Atlantic storm track, as indicated by the
Century projections, and the implications of maximum velocity variance in Fig. 5.5a, follows
these biases in eddy space-time spectra are not a latitude circle too closely and the observed
self-evident. Indeed, in the context of simulating storm track has more of a southwest-northeast
eddy characteristics generated in complex tur- tilt. This particular deficiency is common in
bulent flows in the laboratory (e.g., Pitsch CMIP3 models (van Ulden and van Oldenborgh
2006), the quality of atmospheric simulations, 2006) and is related to difficulty in simulating
based closely on fluid dynamical first princi- the blocking phenomenon in the North Atlantic
ples, probably should be thought of as one of with correct frequency and amplitude. Van
the most impressive characteristics of current Ulden and van Oldenborgh make the case that
models. As an example of a significant model this bias is significant for the quality of regional
deficiency that plausibly can be linked to limi- climate projections over Europe.
D JF 300 hPa v' v' ( m2s–2) Figure 5.5a.Top:

CM2.1_AllForc_H 2 (1981 – 2000) Variance of N orth-
max = 337.5 avg = 88.05
South Velocity at 300
min = 2.989 std = 72.62
400 hPa as Simulated by
80°N GFDL CM2.1 Model in
350
Years 1981 to 2000 of
40°N
300 One Realization of
250 20C3M Simulation, as
Contributed to the
0° 200
CMIP3 Database.
150 Units are m2/s2. Middle:
40°S 100 Same quantity as obtained
from N CEP/N C AR
50
Reanalysis (Kalnay et al.
80°S
0 1996). Bottom: Model minus
50°E 150°E 110°W 10°W observations.
max = 358.4 avg = 95.3
min = 5.79 N CEP (1981 – 2000) std = 78.99
400
80°N
350
40°N 300
250
0° 200
150
40°S 100
50
80°S
0
50°E 150°E 110°W 10°W
max = 85.53 avg = –7.245
min = –118 CM2.1_AllForc_H 2 – N CEP r ms = 24.45
80°N 125
100
40°N 75
50
25
0° 0
–25
–50
40°S
–75
–100
80°S –125
50°E 150°E 110°W 10°W
61
Figure 5.5b.Top:
Covariance of N orth- D JF 850 hPa v' T ' ( m° K s–1)
South Velocity and CM2.1_AllForc_H 2 (1981 – 2000)
max = 32.27 avg = 2.3
Temperature at 850 min = 14.85 std = 8.24
hPa as Simulated by 40
GFDL CM2.1 Model in 80°N 35
30
Years 1981 to 2000 of 25
One Realization of 40°N 20
15
20C3M Simulation, as 10
Contributed to the 5
0° 0
CMIP3 Database. –5
–10
Units are K-m/s. Middle: –15
Same quantity as obtained 40°S –20
from N CEP/N C AR –25
–30
Reanalysis (Kalnay et al. 80°S –35
1996). Bottom: Model minus –40
observations. 50°E 150°E 110°W 10°W
max = 32.38 avg = 1.398
min = 17.59
N CEP (1981 – 2000) std = 7.577
40
80°N 35
30
25
40°N 20
15
10
5
0° 0
–5
–10
–15
40°S –20
–25
–30
80°S –35
–40
50°E 150°E 110°W 10°W
max = 18.4 avg = 0.9021
min = –11.78 CM2.1_AllForc_H 2 – N CEP r ms = 2.588
80°N 21
18
12
40°N 9
6
3
0
0° –3
–6
–9
40°S –12
–15
–18
80°S –21
50°E 150°E 110°W 10°W
5.2.2.2 T RO PIC AL STO RMS in simulating the effects of El Niño on Atlantic

Tropical storms (hurricanes in the Atlantic and storm frequency.
typhoons in the Pacific and Indian oceans) are
too small to be simulated reliably in the class of Simulations with atmospheric models are
global-climate models currently used for cli- steadily moving to higher resolutions (e.g.,
mate projections. There is hope for simulating Bengtsson, Hodges, and Esch 2007). The recent
regional climate aspects that control the gene- 20-km–resolution simulation with an atmos-
sis of tropical depressions, however. Vitart and pheric model over prescribed ocean tempera-
Anderson (2001), for example, identified trop- tures by Oouchi et al. (2006) is indicative of the
ical storm-like vortices in simulations with kinds of modeling that will be brought to bear
models of this type, demonstrating some skill on this problem in the next few years. Experi-
62
ence with tropical storm forecasting suggests ture from the Gulf of Mexico brings an annual
that this resolution should be adequate for de- peak in rainfall. Thus, the climate in these re-
scribing many aspects of the evolution of ma- gions also is described as monsoonal.
ture tropical storms and possibly the generation
of storms from incipient disturbances, but prob- Because of the Asian monsoon’s geographical
ably not tropical storm intensity. A promising extent, measures of the fidelity of Asian mon-
alternative approach is described by Knutson et soonal simulations can differ depending on spe-
al. (2007), in which a regional model of com- cific regional focus and the metrics being used.
parable resolution (18 km) is used in a down- Kripalani et al. (2007) judged that 3/4 of the 18
scaling framework (see Chapter 3) to simulate analyzed coupled models match the timing and
the Atlantic hurricane season. Given observed magnitude of the summertime peak in precipi-
year-to-year variations in the large-scale atmos- tation over East Asia between 100 and 145°E
phere structure over the Atlantic Ocean, the and 20 to 40°N evident in the NOAA-NCEP
model is capable of simulating year-to-year Climate Prediction Center’s Merged Analysis of
variations in hurricane frequency over a 30-year Precipitation (CMAP, Xie, and Arkin 1997).
period with a correlation of 0.7 to 0.8. It also However, only half of these models were able to
captures the observed trend toward greater hur- reproduce the gross observed spatial distribu-
ricane frequency in the Atlantic during this pe- tion of monsoon rainfall and its migration along
riod. These results suggest that downscaling the coast of China toward the Korean peninsula
using models of this resolution may be able to and Japan. Considering a broader range of lon-
provide a convincing capability for tropical gitude (40 to 180°E) that includes the Indian
storm frequency projections into the future, al- subcontinent, Annamalai, Hamilton, and Sper-
though these projections still will rely on the ber (2007) found that 6 of 18 AOGCMs signif-
quality of global model projections for changes icantly correlated with the observed spatial
in sea-surface temperature, atmospheric stabil- pattern of CMAP precipitation from June
ity, and vertical shear. through September. (These six models also pro-
duced relatively realistic simulation of ENSO
5.2.2.3 M O N SO O N S
variability, which is known to influence inter-
A monsoonal circulation is distinguished by its annual variations in the Asian summer mon-
seasonal reversal after the sun crosses the equa- soon.) Kitoh and Uchiyama (2006) computed
tor into the new summer hemisphere. Rain is the spatial correlation and root-mean-square
most plentiful in, if not entirely restricted to, error of simulated precipitation over a similar
summer within monsoonal climates, when con- region and found, for example, the GFDL mod-
tinental rainfall is supplied mainly by evapora- els in the top tercile with a spatial correlation
tion from the nearby ocean. This limits the reach exceeding 0.8.
of monsoon rains to the distance over which
moisture can be transported onshore (Privé and During boreal winter, Asian surface winds are
Plumb 2007). Variations in the monsoon’s spa- directed offshore: from the northeast over India
tial extent from year to year determine which and the northwest over East Asia. Hori and
inland regions experience drought. Ueda (2006) provide correlations between ob-
served spatial distributions of surface pressures
Over a billion people are dependent on the ar- and 850-mb zonal winds during the East Asian
rival of monsoon rains for water and irrigation winter monsoon with winds and pressures sim-
for agriculture. The Asian monsoon during bo- ulated by nine CMIP3 models. Correlations for
real summer is the most prominent example of zonal winds, for example, vary from 0.96 to
a monsoon circulation dominating global rain- 0.75. Monsoonal simulations in these models
fall during this season. However, the summer clearly vary considerably in quality, more so
rainfall maximum and seasonal reversal of perhaps than other circulation features. Ob-
winds also indicate monsoon circulations in served year-to-year variability of the West
West Africa and the Amazon basin. In addition, African monsoon is related to remote ocean
during boreal summer, air flows off the eastern temperatures in the North and South Atlantic
Pacific Ocean toward Mexico and the American and Indian oceans (Rowell et al. 1992; Zhang
Southwest while, over the Great Plains, mois- and Delworth 2006) as well as to temperatures
63
in the nearby Gulf of Guinea. Cook and Vizy Great Plains and Mexico was summarized by
(2006) found that slightly more than half of 18 Ruiz-Barradas and Nigam (2006). Models gen-
analyzed coupled models reproduced the ob- erally have more difficulty in simulating sum-
served precipitation maximum over land from mer rainfall in the Great Plains than winter
June through August. Of these models, only six rainfall, and this disparity probably should be
(including GISS ModelE-H and both GFDL thought of as reflecting the quality of future
models) reproduced the observed anticorrela- rainfall projections as well. Strengths and weak-
tion between Gulf of Guinea ocean temperature nesses vary considerably across the models. As
and Sahel rainfall. an example, GISS ModelE-H closely matches
the annual precipitation cycle over the Great
The late 20th Century Sahel drought was a dra- Plains and Mexico and is one of two models to
matic change in the Earth’s hydrological cycle simulate interannual precipitation variations
that plausibly must be simulated by climate significantly correlated with observed variabil-
models if we are to have any confidence in their ity during the second half of the 20th Century.
ability to project future climate in this region.
Atmospheric models, when run over observed Initial monsoon evaluations simulated by the
oceanic temperatures, simulate this drought rea- most recent generation of climate models have
sonably well (Hoerling et al. 2006). In these emphasized the seasonal time scale. However,
models, the drought is at least partly forced by subseasonal variations, such as break periods
warming of the Northern Hemisphere oceans, when the monsoon rains are interrupted tem-
particularly the North Atlantic, with respect to porarily, are crucial to forecasting the mon-
Southern Hemisphere oceans, especially the In- soon’s impact on water supply. Simulating the
dian Ocean and Gulf of Guinea. Although the diurnal cycle and the local hour of rainfall also
consensus is that these variations in ocean tem- is important to partitioning rainfall between
perature gradients are at least partly due to nat- runoff and transpiration, and these are impor-
ural variability, they may have been partly tant topics for future model evaluation. Trans-
anthropogenically forced. Analysis of CMIP3 ports of moisture by regional circulations
simulations of the 20th Century by Biasutti and beneath model resolution (such as low-level jets
Giannini (2006), supporting the earlier model- along the Rockies and Andes and tropical cy-
ing study of Rotstayn and Lohmann (2002), clones) contribute to the onshore transport of
suggests that aerosol forcing in these models moisture. In general, models show some success
played a part in generating this drought by cool- at simulating gross seasonal features of various
ing the North Atlantic with respect to other monsoon circulations, but studies are limited on
ocean basins. A small number of coupled mod- variations of the smaller spatial and time scales
els simulate droughts of the observed magni- important to specific watersheds and hydrolog-
tude, including GFDL models (Held and Soden ical projections.
2006), but why some models are more realistic
5.2.2.4 M AD D EN -JULIAN O SCILLATIO N S
in this regard than others is not understood.
The Madden-Julian Oscillation (MJO) consists
Rainfall over the Sahel and Amazon are anti- of large-scale eastward-propagating patterns in
correlated: when the Gulf of Guinea warms, humidity, temperature, and atmospheric circu-
rainfall generally is reduced over the Sahel but lation that strengthen and weaken tropical rain-
increases over South America. Amazon rainfall fall as they propagate around the Earth in
also depends on the eastern equatorial Pacific, roughly 30 to 60 days. This pattern often domi-
and, during an El Niño, rainfall is reduced in the nates tropical precipitation variability on time
Nordeste region of the Amazon. Li et al. (2006) scales longer than a few days and less than a
compare the hydrological cycle of 11 CGCMs season, creating such phenomena as 1- to 2-
over the Amazon during the late 20th and 21st week breaks in Asian monsoonal rainfall and
centuries. Based on a comparison to CMAP weeks with enhanced hurricane activity in the
rainfall, the GISS ModelE-R is among the best. eastern North Pacific and the Gulf of Mexico.
Inadequate prediction of the evolution of these
The ability of climate models to simulate North- propagating structures is considered a main im-
ern Hemisphere summer rainfall over the U.S. pediment to more useful extended-range
64
weather forecasts in the tropics, and improved ficiency but is a result of the phenomenon’s
simulation of this phenomenon is considered an complexity, given the long list of factors thought
important metric for the credibility of climate to be significant. In several multimodel studies
models in the tropics. such as Lin et al. (2006), a few models do per-
form well. However, without a clearer under-
Nearly all models capture the pattern’s essential standing of how these factors combine to
feature, with large-scale eastward propagation generate the observed characteristics of MJO,
and with roughly the correct vertical structure. maintaining a good simulation when the model
But propagation often is too rapid and ampli- is modified for other reasons is difficult, as is
tudes too weak. Recent surveys of model per- applying the understanding gained from one
formance indicate that simulations of MJO model’s successful simulation to other models.
remain inadequate. For example, Lin et al. Whether models with superior MJO simulations
(2006), in a study of many CMIP3 models, con- should be given extra weight in multimodel
clude that “… current GCMs still have signifi- studies of tropical climate change is unclear.
cant problems and display a wide range of skill
5.2.2.5 EL N IÑ O –SO UTH ERN O SCILLATIO N
in simulating the tropical intraseasonal vari-
ability,” while Zhang et al. (2005) in another By the mid-20th Century, scientists recognized
multimodel comparison study, state that “… that a local anomaly. in rainfall and oceanic up-
commendable progress has been made in MJO welling near the coast of Peru was in fact part of
simulations in the past decade, but the models a disruption to atmospheric and ocean circula-
still suffer from severe deficiencies ….” As an tions across the entire Pacific basin. During El
example of recent work, Boyle et al. (2008) at- Niño, atmospheric mass migrates west of the
tempted, with limited success, to determine dateline as part of the Southern Oscillation, re-
whether two U.S. CMIP3 models could main- ducing surface pressure and drawing rainfall
tain a preexisting strong MJO pattern when ini- into the central and eastern Pacific (Rasmussen
tialized with observations [from the Tropical and Wallace 1983). Together, El Niño and the
Ocean Global Atmosphere–Coupled Ocean At- Southern Oscillation, abbreviated in combina-
mosphere Response Experiment (called TOGA- tion as ENSO, are the largest source of tropical
COARE) field experiment]. variability observed during recent decades. Be-
cause of the Earth’s rotation, easterly winds
The difficulty in simulating MJO is related to along the equator cool the surface by raising
the phenomenon’s multiscale nature: the propa- cold water from below, which offsets heating by
gating pattern itself is large enough to be re- sunlight absorption (e.g., Clement et al. 1996).
solvable by climate models, but the convection Cold water is especially close to the surface in
and rainfall modulated by this pattern, which the east Pacific, while warm water extends
feed back on the large-scale environment, occur deeper in the west Pacific so upwelling has lit-
on much smaller, unresolved scales. In addition tle effect on surface temperature there. The
to this dependence on parameterization of trop- westward increase in temperature along the
ical convection, a long list of other effects has equator is associated with a decrease in atmos-
been shown by models and observational stud- pheric pressure, reinforcing the easterly trade
ies to be important for MJO. These effects in- winds. El Niño occurs when easterly trade
clude the pattern of evaporation generated as winds slacken, reducing upwelling and warm-
MJO propagates through convecting regions, ing the ocean surface in the central and east Pa-
feedback from cloud-radiative interactions, in- cific.
traseasonal ocean temperature changes, the di-
urnal cycle of convection over the ocean, and Changes along the equatorial Pacific have been
the vertical structure of latent heating , espe- linked to global disruptions of climate (Ro-
cially the proportion of shallow cumulus con- pelewski and Halpert 1987). During an El Niño
gestus clouds and deep convective cores in event, the Asian monsoon typically is weakened,
different phases of oscillation (Lin et al. 2004)]. along with rainfall over eastern Africa, while
precipitation increases over the American
A picture seems to be emerging that simulation Southwest. El Niño raises the surface tempera-
difficulty may not be due to a single model de- ture as far poleward as Canada, while changes
65
in the north Pacific Ocean are linked to decadal the equator in oceanic components of the seven
variations in ENSO (Trenberth and Hurrell American coupled models whose output was
1994). In many regions far from eastern equa- submitted to CMIP3.
torial Pacific, accurate projections of climate
change in the 21st Century depend upon the ac- Along the equator, oceanic waves that adjust the
curate projection of changes to El Niño. More- equatorial temperature and currents to changes
over, the demonstration that ENSO alters in the wind are confined tightly to within a few
climate across the globe indicates that even degrees of latitude. To simulate this adjustment,
changes to the time-averaged equatorial Pacific the ocean state is calculated at points as closely
during the 21st Century will influence climate spaced as 0.27 degrees of latitude in the NCAR
far beyond the tropical ocean. For example, CCSM3. NCAR PCM has a half-degree resolu-
long-term warming of the eastern equatorial Pa- tion, while both GFDL models have equatorial
cific relative to the surrounding ocean will favor resolution of a third of a degree. This degree of
a weaker Asian monsoon year after year, even detail is a substantial improvement compared to
in the absence of changes to the size and fre- previous generations of models. In contrast, the
quency of El Niño events. GISS AOM and ModelE-R calculate equatorial
temperatures at grid points separated by four de-
In general, coupled models developed for grees of latitude. This is broad compared to the
CMIP3 are far more realistic than those of a latitudinal extent of cold temperatures observed
decade ago, when ENSO variability was com- within the eastern Pacific. The cooling effect of
paratively weak and some models lapsed into upwelling is spread over a larger area, so the
permanent El Niño states (Neelin et al. 1992). amplitude of the resulting surface temperature
Even compared to models assessed more re- fluctuation is weakened. In fact, both the GISS
cently by the El Niño Simulation Intercompar- AOM models and ModelE-R have unrealistic
ison Project (called ENSIP) and CMIP2 (Latif ENSO variations that are much smaller than ob-
et al. 2001; AchutaRao and Sperber 2002), served (Hansen et al. 2007). This minimizes the
ENSO variability of ocean surface temperature influence of their simulated El Niño and La
is more realistic in CMIP3 simulations, al- Niña events on climate outside the equatorial
though sea-level pressure and precipitation Pacific, and we will not discuss these two mod-
anomalies show little recent improvement els further in this section.
(AchutaRao and Sperber 2006). Part of this
progress is the result of increased resolution of In comparison to previous generations of global
equatorial ocean circulation that has accompa- models, where ENSO variability was typically
nied increases in computing speed. Table 5.1 weak (Neelin et al. 1992), the AR4 coupled
shows horizontal and vertical resolution near models generally simulate El Niño near the ob-
served amplitude or even above (AchutaRao
Table 5.1. Spacing Ver t ical and Sperber 2006). The latter study compared
of Grid Points at MODEL Longit ude Lat it ude
Levels sea-surface temperature (SST) variability within
the Equator in the
American GFD L CM2.0 1 1/3 50 the tropical Pacific, calculated under preindus-
Coupled Models trial conditions. Despite its comparatively low
Developed for GFD L CM2.1 1 1/3 50 two-degree latitudinal grid spacing, the GISS
AR4* ModelE-H (among American models) most
GISS AO M 5 4 13
closely matches observed SST variability since
GISS ModelE-H 2 2 16 the mid-19th Century, according to the HadISST
v1.1 dataset (Rayner et al. 2003). The NCAR
GISS ModelE-R 5 4 13
PCM also exhibits El Niño warming close to the
N C AR CCSM3 1.125 0.27 27 observed magnitude. This comparison is based
on spatial averages within three longitudinal
N C AR PCM 0.94 0.5 32 bands, and GISS ModelE-H, along with NCAR
* Except for GISS models, spacing of grid points
models, exhibits its largest variability in the
generally increases away from the equator outside eastern band as observed. However, GISS Mod-
the EN SO domain, so resolution is highest at the elE-H underestimates variability since 1950,
equator.
when the NCAR CCSM3 is closest to observa-
66
tions (Joseph and Nigam 2006). Although the tongue is observed to warm during boreal
fidelity of each model’s ENSO variability de- spring and cool again late in the calendar year.
pends on the specific dataset and period of com- GFDL CM2.1 and NCAR PCM1 have the
parison (c.f. Capotondi, Wittenberg, and Masina weakest seasonal cycle among American mod-
2006; Merryfield 2006; van Oldenborgh, Philip, els, while GISS ModelE-H, GFDL 2.0, and
and Collins 2005), the general consensus is that NCAR CCSM3 are closest to the observed am-
GISS ModelE-H, both NCAR models, and plitude (Guilyardi 2006). Among the worldwide
GFDL CM2.0 have roughly the correct ampli- suite of CMIP3 models, amplitude of the sea-
tude, while variability is too large by roughly sonal cycle of equatorial ocean temperature
one-third in GFDL CM2.1. Most models (in- generally varies inversely with the ENSO
cluding GISS ModelE-H and both NCAR mod- cycle’s strength.
els but excluding GFDL models) exhibit the
largest variability in the eastern band of longi- Several studies have compared mechanisms
tude, but none of the CMIP3 models matches generating ENSO variability in CMIP3 models
the observed variability at the South American to those inferred from observations (e.g., van
coast where El Niño was identified originally Oldenborgh, Philip, and Collins 2005; Guilyardi
(AchutaRao and Sperber 2006; Capotondi, Wit- 2006; Merryfield 2006; Capotondi, Wittenberg,
tenberg, and Masina 2006). This possibly is be- and Masina 2006). Models must simulate the
cause the longitudinal spacing of model grids is change in ocean upwelling driven by changes in
too large to resolve coastal upwelling and its in- surface winds, which in turn are driven by re-
terruption during El Niño (Philander and gional contrasts in ocean temperature. In gen-
Pacanowski 1981). Biases in atmospheric mod- eral, GFDL2.1 is ranked consistently among
els (e.g., underestimating persistent stratus American models as providing the most realis-
cloud decks along the coast) also may con- tic simulation of El Niño. This is not based pri-
tribute (Mechoso et al. 1995). marily on its surface-temperature variability
(which is slightly too large) but on its faithful
El Niño occurs every few years, albeit irregu- simulation of the observed relationship between
larly. The spectrum of anomalous ocean tem- ocean temperature and surface wind, along with
perature shows a broad peak between 2 and 7 wind-driven ocean response. While SST vari-
years, and multidecadal variations occur in ability in CMIP3 models is controlled by anom-
event frequency and amplitude. Almost all AR4 alies of either upwelling rate or temperature,
models have spectral peaks within this range of these processes alternate in importance over
time scales. Interannual power is distributed several decades within GFDL CM2.1 as ob-
broadly within the American models, as observed (Guilyardi 2006). Since the 1970s the
served, with the exception of NCAR CCSM3, upwelling temperature, rather than the rate, has
which exhibits strong biennial oscillations been the predominant driver of SST variability
(Guilyardi 2006). (Wang 1995). A confident prediction of future
El Niño amplitude requires both the upwelling
Although models generally simulate the ob- rate and temperature, along with their relative
served magnitude and frequency of events, re- amplitude, to be simulated correctly. This re-
producing their seasonality is more elusive. mains a challenge.
Anomalous warming typically peaks late in the
calendar year, as originally noted by South El Niño events are related to climate anomalies
American fisherman. Among American mod- throughout the globe. Models with more realis-
els, this seasonal dependence is simulated only tic ENSO variability generally exhibit an anti-
by NCAR CCSM3 (Joseph and Nigam 2006). correlation with the strength of the Asian
Warming in GFDL CM2.1 and GISS ModelE- summer monsoon (e.g., Annamalai, Hamilton,
H is nearly uniform throughout the year, while and Spencer 2007), while 21st Century changes
warming in NCAR PCM is largest in Decem- to Amazon rainfall have been shown to depend
ber but exhibits a secondary peak in early sum- on projected trends in the tropical Pacific (Li et
mer. The mean seasonal cycle along the al. 2006). El Niño has a long-established rela-
equatorial Pacific also remains a challenge for tion to North American climate (Horel and Wal-
the models. Each year, the east Pacific cold lace 1981), assessed in CMIP3 models by
67
Joseph and Nigam (2006). This relation is change the coupling. Nonetheless, improved
strongest during boreal winter, when tropical simulations of the ENSO cycle compared to
anomalies are largest. Anomalous circulations previous generations (AchutaRao and Sperber
driven by rainfall over the warming equatorial 2006) suggest that additional realism can be ex-
Central Pacific radiate atmospheric distur- pected in the future.
bances into midlatitudes amplified within the
5.2.2.6 A N N ULAR M O D ES
north Pacific storm track (Sardeshmukh and
Hoskins 1988; Held, Lyons, and Nigam 1989; The primary mode of Arctic interannual vari-
Trenberth et al. 1998). To simulate ENSO’s inability is the Arctic Oscillation (Thompson and
fluence on North America, models must repre- Wallace 1998), which also is referred to as the
sent realistic rainfall anomalies in the correct northern annular mode (NAM) and is related to
season so the connection is amplified by win- the North Atlantic Oscillation (Hurrell 1995).
tertime storm tracks. The connection between The primary mode of Antarctic interannual vari-
equatorial Pacific and North American climate ability is the southern annular mode (SAM)
is simulated most accurately by the NCAR (Thompson and Wallace 2000), also known as
PCM model (Joseph and Nigam 2006). In Antarctic Oscillation. The variability modes are
GFDL CM2.1, North American anomalies are particularly important for attributing and pro-
too large, consistent with the model’s excessive jecting climate change; observed circulation
El Niño variability within the equatorial Pacific. changes in the past few decades (especially in
The connection between the two regions is re- the Southern Hemisphere) and model-projected
alistic if the model’s tropical amplitude is ac- changes in future circulation strongly resemble
counted for. In the GISS model, anomalous these structures.
rainfall during ENSO is small, consistent with
the weak tropical wind stress anomaly cited Coupled climate models have shown skill in
above. The influence of El Niño over North simulating NAM (Fyfe, Boer, and Flato 1999;
America is nearly negligible in this model. The Shindell et al. 1999; Miller, Schmidt, and Shin-
weak rainfall anomaly presumably is a result of dell 2006). In some cases, too much variability
unrealistic coupling between atmospheric and in the simulation of sea-level pressure is asso-
ocean physics. When SST instead is prescribed ciated with NAM (Miller, Schmidt, and Shin-
in this model, rainfall calculated by the GISS dell 2006). Global climate models also
ModelE AGCM over the American Southwest realistically simulate SAM (Fyfe, Boer, and
is significantly correlated with El Niño as ob- Flato 1999; Cai, Whetton, and Karoly 2003;
served. Miller, Schmidt, and Shindell 2006), although
some details of SAM (e.g., amplitude and zonal
Realistic simulation of El Niño and its global structure) show disagreement among global cli-
influence remains a challenge for coupled mod- mate model simulations and reanalysis data
els because of myriad contributing processes (Raphael and Holland 2006; Miller, Schmidt,
and their changing importance in the observa- and Shindell 2006).
tional record. Key aspects of coupling between
ocean and atmosphere—the relation between In response to increasing concentrations of
SST and wind stress anomalies, for example— greenhouse gases and tropospheric sulfate
are the result of complicated interactions among aerosols in the 20th Century, the multimodel av-
resolved model circulations, along with para- erage exhibits a positive trend in the annular
meterizations of ocean and atmospheric bound- mode index in both hemispheres, with decreas-
ary layers and moist convection. Simple models ing sea-level pressure over the poles and a com-
identify parameters controlling the magnitude pensating increase in midlatitudes most
and frequency of El Niño, such as the wind apparent in the Southern Hemisphere (Miller,
anomaly resulting from a change in SST (e.g., Schmidt, and Shindell 2006). A variety of mod-
Zebiak and Cane 1987; Fedorov and Philander eling studies also have shown that trends in
2000), offering guidance to improve the realism stratospheric climate can affect the tropos-
of fully coupled GCMs. However, in a GCM, phere’s annular modes (Shindell et al. 1999). In-
the coupling strength is emergent rather than deed, an important result from atmospheric
prescribed, and it is often unclear a priori how to modeling in recent years is the realization that
68
the stratospheric ozone hole has contributed sig- SST are not seen in the models (e.g., Alexander
nificantly to observed trends in surface winds et al. 2006).
and sea-level pressure distribution in the South-
ern Hemisphere (Thompson and Solomon One of the most difficult areas to simulate is the
2002; Gillett and Thompson 2003). The mod- Indian Ocean because of the competing effects
els, however, may not be trustworthy in their of warm water inflow through the Indonesian
simulation of the relative magnitude of green- archipelago, ENSO, and monsoons. The
house gas and stratospheric ozone effects on the processes interact to varying degrees, challeng-
annular mode. They also may underestimate the ing a model’s ability to simulate all system as-
coupling of stratospheric changes due to vol- pects with observed relative emphasis. An index
canic aerosols with annular surface variations used to understand variability is the Indian
(Miller, Schmidt, and Shindell 2006; Arblaster Ocean Dipole pattern that combines informa-
and Meehl 2006). tion about SST and wind stress fields (Saji et al.
1999). While most models evaluated by Saji,
5.2.2.7 O TH ER M O D ES O F M ULTID EC ADAL
Xie, and Yamagata (2005) were able to simulate
VARIABILITY
the Indian Ocean’s response to local atmos-
In the Arctic during the last century, two long- pheric forcing in short time periods (semian-
period warm events occurred, one between 1920 nual), longer-period events such as the ocean’s
and 1950 and another beginning in the late response to ENSO changes in the Pacific were
1970s. Wang et al. (2007) evaluated a set of not simulated well.
CMIP3 models for their ability to reproduce the
amplitudes of air temperature variability of this 5.2.3 Polar Climates
character. As examples, CCSM3 and GFDL-
CM2 models contain variance similar to that ob- Changes in polar snow and ice cover affect the
served in the Arctic region. Earth’s albedo and thus the amount of insola-
tion heating the planet (e.g., Holland and Bitz
Multidecadal variability in the North Atlantic is 2003; Hall 2004; Dethloff et al. 2006). Melting
characterized by the Atlantic Multidecadal Os- glaciers and ice sheets in Greenland and western
cillation (AMO) index, which represents a spa- Antarctica could produce substantial sea-level
tial average of SST (Enfield, Mestas-Nuñez, rise (Arendt et al. 2002; Braithwaite and Raper
and Trimble 2001). Kravtsov and Spannagle 2002; Alley et al. 2005). Polar regions thus re-
(2007) analyzed SST from a set of current gen- quire accurate simulation for projecting future
eration climate models. Their analysis attempts climate change and its impacts.
to separate variability associated with internal
ocean fluctuations from that associated with Polar regions present unique environments and,
changes by anthropogenic contributions. By consequently, challenges for climate modeling.
isolating the multidecadal period of several re- Key processes include sea ice, seasonally frozen
gions in the ensemble SST series through sta- ground, and permafrost (Lawrence and Slater
tistical methods, they found that models obtain 2005; Yamaguchi, Noda, and Kitoh 2005).
the observed magnitude of the AMO (Kravtsov Processes also include seasonal snow cover
and Spannagle 2007). (Slater et al. 2001), which can have significant
subgrid heterogeneity (Liston 2004), and clear-
In the midlatitude Pacific region, decadal vari- sky precipitation, especially in the Antarctic
ability generally is underrepresented in the (King and Turner 1997; Guo, Bromwich, and
ocean (e.g., volume transports as described by Cassano 2003). Polar regions test the ability of
Zhang and McPhaden 2006), with some mod- models to handle extreme geophysical behavior
els approaching amplitudes seen in observa- such as longwave radiation in clear, cold envi-
tions. Examination of complicated feedbacks ronments (Hines et al. 1999; Chiacchio, Fran-
between atmosphere and ocean at decadal and cis, and Stackhouse 2002; Pavolonis, Key, and
longer scales shows that, while climate models Cassano 2004) and cloud microphysics in the
generally reproduce the SST pattern related to relatively clean polar atmosphere (Curry et al.
the Pacific Decadal Oscillation (PDO), ob- 1996; Pinto, Curry, and Intrieri 2001; Morrison
served correlations between PDO and tropical and Pinto 2005). In addition, polar atmospheric
69
boundary layers can be very stable (Duynkerke lated turbulent heat flux simulated by models
and de Roode 2001; Tjernström, Zagar, and was many times larger than the observed turbu-
Svensson 2004; Mirocha, Kosovic, and Curry lent heat flux (Fig. 5.6).
2005), and their simulation remains an impor-
tant area for model improvement. In global models, polar climate may be affected
by errors in simulating other planetary regions,
For polar regions, much of simulated-variability but much of the difference from observations
assessment has focused on primary modes of and the uncertainty about projected climate
polar interannual variability, along with the change stem from current limitations in polar
northern and southern annular modes. Less at- simulation. These limitations include missing or
tention has been given to the ability of global incompletely represented processes and poor
climate-system models to simulate shorter-du- resolution of spatial distributions.
ration climate and weather variability in polar
regions. Uotila et al. (2007) and Cassano et al. As with other regions, model resolution affects
(2007) evaluated the ability of an ensemble of simulation of important processes. In polar re-
15 global climate-system models to simulate gions, surface distributions of snow depth vary
daily variability in sea-level pressure in the markedly, especially when snow drifting occurs.
Antarctic and Arctic. In both polar regions, they Improved snow models are needed to represent
found that the ensemble was not able to repro- such spatial heterogeneity (e.g., Liston 2004),
duce many features of daily synoptic climatol- which will continue to involve scales smaller
ogy, with only a small subset of models than resolved for the foreseeable future. Frozen
accurately simulating the frequency of primary ground, whether seasonally frozen or occurring
synoptic weather patterns identified in global as permafrost, presents additional challenges.
reanalysis datasets. U.S. models discussed in de- Models for permafrost and seasonal soil freez-
tail in Chapter 2 of this report spanned the same ing and thawing are being implemented in land
range of accuracy as non-U.S. models, with surface models (see Chapter 2). Modeling soil
GFDL and CCSM models part of a small, ac- freeze and thaw continues to be a challenging
curate subset. More encouraging results were problem as characteristics of energy and water
obtained by Vavrus et al. (2006), who assessed flowing through soil affect temperature
the ability of seven global climate models to changes. Such fluxes are poorly understood (Ya-
simulate extreme cold-air outbreaks in the maguchi, Noda, and Kitoh 2005).
Northern Hemisphere.
Frozen soil affects surface and subsurface hy-
Attention also has been given to the ability of drology, which influences the surface water’s
regional climate models to simulate polar cli- spatial distribution with attendant effects on
mate. In particular, the Arctic Regional Climate other parts of the polar climate system such as
Model Intercomparison Project (ARCMIP) en- carbon cycling (e.g., Gorham 1991; Aurela,
gaged a suite of Arctic regional atmospheric Laurila, Tuovinen 2004), surface temperature
models to simulate a common domain and pe- (Krinner 2003), and atmospheric circulation
riod over the western Arctic (Curry and Lynch (Gutowski et al. 2007). The flow of fresh water
2002). Rinke et al. (2006) evaluated spatial and into polar oceans potentially alters their circu-
temporal patterns simulated by eight ARCMIP lation, too. Surface hydrology modeling typi-
models and found that the model ensemble cally includes, at best, limited representation of
agreed well with global reanalyses, despite subsurface water reservoirs (aquifers) and hor-
some large errors for individual models. Tjern- izontal flow of water both at and below the sur-
strom et al. (2005) evaluated near-surface prop- face. These features limit the ability of climate
erties simulated by six ARCMP models. In models to represent changes in polar hydrology,
general, surface pressure, air temperature, hu- especially in the Arctic.
midity, and wind speed all were well simulated,
as were radiative fluxes and turbulent momen- Vegetation has been changing in the Arctic
tum flux. The research group also found that (Callaghan et al. 2004), and projected warming,
turbulent heat flux was poorly simulated and which may be largest in regions where snow and
that, over an entire annual cycle, the accumu- ice cover retreat, may produce further changes
70
in vegetation (e.g., Lawrence and Slater 2005). observations. Observations of ice extent were
Current models use static distributions of vege- fewer before that. Other quantities that might be
tation, but dynamic vegetation models will be evaluated include ice thickness, but, due to lim-
needed to account for changes in land-atmos- ited observations, comparisons with models are
phere interactions influenced by vegetation. difficult and will not be discussed further here.
A key concern in climate simulations is how The seasonal pattern in ice growth and decay in
projected anthropogenic warming may alter polar regions for all the models is reasonable
land ice sheets, whose melting could raise sea (Holland and Raphael 2006; see Fig. 5.7). How-
levels substantially. At present, climate models ever, a large amount of variability between mod-
do not include ice-sheet dynamics (see Chapter els occurs in their representation of sea-ice
2), and thus cannot account directly for ways in extent in both Northern and Southern hemi-
which ice sheets might change, possibly chang- spheres. Generally, models do better in simulat-
ing heat absorption from the sun and atmos- ing the Arctic than the Antarctic region, as
pheric circulation in the vicinity of ice sheets. shown with Fig. 5.8. An example of the com-
plex nature of reproducing the ice field is given
Distributions of snow, ice sheets, surface water, in Parkinson, Vinnikov, and Cavalieri (2006a,b),
frozen ground, and vegetation have important which found that all models showed an ice-free
spatial variation on scales smaller than the res- region in winter to the west of Norway, as seen
olutions of typical contemporary climate mod- in observational data, but all also produced too
els. This need for finer resolution may be much ice north of Norway. The authors suggest
satisfied by regional models simulating just a that this is because the North Atlantic Current is
polar region. Because both northern and south- not being simulated correctly. In a qualitative
ern polar regions are within circumpolar at- comparison, Hudson Bay is ice covered in win-
mospheric circulations (cf. Giorgi and Bi 2000 ter in all models correctly reproducing the ob-
and Gutowski et al. 2007b), their coupling with servations. The set of models having the most
other regions is more limited than in the case of fidelity in the Arctic is not the same as the set
midlatitude regions, which could allow polar- having the most fidelity in the Antarctic. This
specific models that focus on Antarctic and Arc- difference may be due to distinctive ice regimes
tic processes, in part, to improve modeling of in the north and south or to differences in sim-
surface-atmosphere exchange processes (Fig. ulations of oceanic or meteorological circula-
5.6). Although each process above has been tions in those regions.
simulated in finer-scale, stand-alone models,
their interactions as part of a climate system Holland and Raphael (2006) examined carefully
also need to be simulated and understood. the variability in Southern Ocean sea-ice extent.
As an indicator of ice response to large-scale at-
5.2.3.1 SEA ICE
mospheric events, they compared data from a
Sea ice plays a critical role in the exchange of set of IPCC AR4 climate models to the atmos-
heat, mass, and momentum between ocean and pheric index SAM for the April–June (AMJ) pe-
atmosphere, and any errors in the sea-ice sys- riod (see Table 5.2). The models show that ice
tem will contribute to errors in other compo- variability does respond modestly to large-scale
nents. Two recent papers (Holland and Raphael atmosphere forcing but less than the limited ob-
2006; Parkinson, Vinnikov, and Cavalieri 2006a, servations show. Table 5.2 uses the U.S. models
b) quantify how current models simulate the cli- to examine whether models exhibit the observed
mate system’s sea-ice process. Very limited ob- out-of-phase buildup of ice between the Atlantic
servations make any evaluation of sea ice and Pacific sectors (referred to as the Antarctic
difficult. The primary observation available is Dipole).
sea-ice areal concentration. In some compar-
isons, sea-ice extent (the area where local ice
concentration is greater than 15%) is used. For
the past few decades, satellites have made it
possible to produce a more complete dataset of
71
Figure 5.6. Cumulative 50

Fluxes of Surface
Sensible H eat (top
panel) and Latent H eat 0
(bottom) at the
Accum. Sens. H eat Flux (MJ m–2)

SH EBA Site.
D ata are from six models –50
simulating a western Arctic
domain for Sept. 1997
through Sept. 1998 for
ARCMIP. SH EBA –100
observations are gray shaded
regions; model results are
shown by the individual
–150
curves identified in the key at ARCSYM
the lower left of the upper CO AMPS
panel. [Figure adapted from H IRH AM
Fig. 10(c and d) in M. –200 PMMS
Tjernstrom et al. 2005: RC A
Modelling the Arctic REMO
boundary layer :An evaluation
of six ARCMIP regional-scale –250
models with data from the 1 Sep 1 D ec 1 Mar 1 Jun 1 Sep
SH EBA project. Boundary-
Layer M eteorology, 117, 200
337–381. Reproduced with
kind permission of Springer
Science and Business Media.]
150
Accum. Sens. H eat Flux (MJ m–2)
100
50
–250
1 Sep 1 D ec 1 Mar 1 Jun 1 Sep
D ate
Table 5.2. Correlations AMJ SAM and High-Pass AMJ SAM and Det rended
of the Leading Mode of Filt ered Fields Fields
Sea-Ice Variability and
Southern Annular O bservations 0.47 0.47
Mode (SAM) for
CCSM3 0.40 0.44
Observations and
Model Simulations GFD L-CM2.1 0.39 0.19
GISS-ER 0.30 0.20
The leading mode of sea-ice variability represents a shift of ice from the Atlantic to the Pacific
sector. Bold values are significant at the 95% level, accounting for autocorrelation of the time series.
[Table modified from Table 1, p. 19, in M.M. H olland and M.N . Raphael 2006:Twentieth Century
simulations of the Southern H emisphere climate in coupled models. Part II: Sea ice conditions
and variability. Climate D ynamics, 26, 229–245. Reproduced with kind permission of Springer
Science and Business Media.]
72
Figure 5.7. Annual Cycle

SH Total Extent of Southern H emisphere
25 Ice Extent.
It is defined as the area of ice
with concentrations greater
20 than 15%. O bservations are
identified by the black curve
labeled “O bs,” while the results
from individual models are
15 identified by the six colored
10 6 km3
curves. [From Fig. 1 in M.M.

obs H olland and M.N . Raphael 2006:
10 ccsm Twentieth Century simulations
giss_model_e_r of the Southern H emisphere
gfd_cm2_1 climate in coupled models. Part
csiro_mk3_0 II: Sea ice conditions and
5 variability. Climate D ynamics, 26,
ukmo_hodcm3
miroc3_2_hires 229–245. Reproduced with kind
permission of Springer Science
0 and Business Media.]
J F M A M J J A S O N D
Month
10 Figure 5.8. Difference

(a) N orther n H emisphere
Between Modeled 1979
to 2004 Monthly Average
Sea-Ice Extents and
Ice Extension D ifferences (106 km2)
5 Satellite-Based
Observations (modeled
minus observed).
D ata are shown for each of 11
major GCMs for both (a)
0 N orthern H emisphere and (b)
Southern H emisphere. [From
H adCM3 Fig. 4 in C .L. Parkinson, K.Y.
Vinnnikov, and D.J. Cavalieri
H adGEM1 2006: Correction to evaluation
–5 of the simulation of the annual
ECH AM5
cycle of Arctic and Antarctic.
CGCM3 J. Geophysical Research, 111,
C07012. Reproduced by
CSIRO Mk3 permission of the American
–10 Geophysical Union (AGU).]
J F M A M J J A S O N D MIRO C3
10 BCCR BCM2
(b) Souther n H emisphere
GISS ER
IPSL CM4
Ice Extension D ifferences (106 km2)
5 IN M CM3
GFD L CM2.1
–5
–10
J F M A M J J A S O N D
73
5.2.4 Ocean Structure and face, errors in SSTs typically originate with de-
Circulation ficiencies in both atmospheric and ocean model
components. In general, more recent model ver-
Unlike the atmosphere, the amount of observa- sions show improvement over previous models
tional data available to evaluate ocean simula- when simulated SST fields are compared to ob-
tions is very limited for long time periods. servations. Figure 5.9 (Delworth et al. 2006)
Nevertheless, sufficient data exist to identify a shows comparisons of simulated and observed
set of ocean characteristics or metrics to evalu- mean SST fields of both the older GFDL
ate ocean models for their climate simulation CM2.0 and newer CM2.1 averaged over a 100-
properties. The most important is sea-surface year period. The new model reduced a cold bias
temperature, but other quantities that serve as in the Northern Hemisphere from earlier simu-
good indicators of ocean realism in climate lations, resulting in both a more-realistic repre-
models are ocean heat uptake, meridional over- sentation of atmospheric wind stress at the
turning and ventilation, sea-level variability, and ocean surface and a modified treatment of sub-
global sea-level rise. grid-scale oceanic mixing. The CCSM3.0
model’s improved SST simulation over
5.2.4.1 SEA-SURFACE T EMPERATURE
CCSM2.0 results mainly from changes in rep-
Sea-surface temperature (SST) plays a critical resenting processes associated with the mixed
role in determining climate and the predictabil- layer of upper ocean waters (Danabasoglu et al.
ity of climate changes. Because of interactions 2006).
in atmospheric and ocean circulations at the sur-
Figure 5.9. Maps of Sea Surface Temperature: Model minus Observations

Simulation Errors in
(a) CM2.0
Annual Mean SST.
Units are Kelvin (K). Errors
80°N
are computed as model
minus observations from
Reynolds SST data 40°N –4 –3
(provided by N O AA-CIRES –2
–2
Climate D iagnostics Center, –1 –1
Boulder, Colorado, from
0°
their W eb site,
www.cdc.noaa.gov). (a)
CM2.0 (using model years
101 to 200). (b) CM2.1 40°S
(using model years 101 to 2
1
200). Contour interval is
1 K, except for no shading
80°S
of values between 1 K and
+1 K. [Images from T.L. (b) CM2.1
D elworth et al. 2006:
GFD L’s CM2 global coupled 80°N
climate models. Part 1:
Formulation and simulation
characteristics. J. Climate, 19, 40°N –2 –1
643–684. Reproduced by
–1
permission of the American
0°
40°S
1
2
80°S
150°E 110°W 10°W 90°E
–10 – 8 –6 –4 –3 –2 –1 1 2 3 4 6 8 10
74
In addition to SST mean values, 20th Century 5.2.4.2 M ERID IO N AL O VERTURN IN G

trends of SST changes also are significant for C IRCULATIO N AN D VEN TILATIO N
model evaluation, since ocean SST contributes The planetary-scale circulation transporting
the dominant signal to the observed global sur- heat and freshwater throughout global oceans is
face temperature trend. An intermodel compar- referred to as global thermohaline circulation.
ison of 50-year tropical SST trends is shown in The Atlantic portion is called the Atlantic
Fig. 5.10. Trends range from a low of 0.1°C/50 meridional overturning circulation (AMOC).
yrs to a high of about 0.6°C/50 yrs, with the ob- Tropical and warm waters flow northward via
servational trend estimate given as about the Gulf Stream and North Atlantic Current.
0.43°C/50 yrs. The figure also shows some ran- Southward flow occurs when water is subducted
domness within a group of simulations run by in regions around Labrador and Greenland; sur-
the same model. For example, the two different face waters freshen, become denser, and flow
GFDL model versions discussed above were down the slope to deeper depths. Similar
each run for multiple realizations of the 20th processes occur at locations in the Southern
Century. CM 2.0 simulations are noted by Ocean. “Ventilation” is the name given to the
GFDL201, GFDL202, and GFDL203, and CM process by which these dense surface waters are
2.1 simulations are noted by GFDL211, carried into the ocean interior. An important cli-
GFDL212, and GFDL213. mate parameter is the rate at which this process
occurs. The pattern of circulation may weaken,
SST Trend Figure 5.10.Trends

and Standard
1.2 Deviations of
1
Tropical SST
Between 1950 and
0.8 1999.
(°C / 50 Yrs)
O bservations are shown

0.6 by the leftmost bar in
each figure. All others are
0.4
model results. Error bars
0.2 show 95% significance
levels for trends. [Images
0 from Fig. 9 in D. Z hang and
M.J. McPhaden 2006:
D ecadal variability of the
shallow Pacific meridional
cn .
bs
gfd ir o
m ch
cm
rm
ha m
m ls
h
overturning circulation:
cc r i
3
fgo r
om
3
gfd 01
gfd 02
gfd 03
gfd 11
gfd 12
gis 13
se
se
sm
a
pc
m
m
ir o
cs
O
ir o
gis
l2
l2
l2
l2
l2
l2
gis
sa
dc
Relation to tropical sea-

surface temperatures in
observations and climate
change models. Ocean
M odelling, 15, 250–273.
SST Standard Deviation Used with permission
0.32 from Elsevier.]
0.28
0.24
0.2
(°C)
0.16
0.12
0.08
0.04
0
cn .
bs
gfd ir o
m ch
cm
rm
ha m
gis h
m ls
cc r i
om
3
fgo r
3
gfd 01
gfd 11
gfd 02
gfd 03
gfd 12
gis 13
se
se
sm
a
m
pc
ir o
cs
O
ir o
l2
l2
l2
l2
l2
l2
gis
sa
dc
75
affecting the climate in the region surrounding of the Earth. The calculations for the ocean’s
the North Atlantic. Schmittner, Latif, and northward heat transport in the current genera-
Schneider (2005) examined a small ensemble tion of climate models show that the models
set of simulations to quantify uncertainty in reasonably represent the observations (Delworth
model representation of 20th Century AMOC et al. 2006; Collins et al. 2006a; Schmidt et al.
transports. To make their estimate, they evalu- 2006). The current models have significantly
ated global temperature, global salinity, pycno- improved over the last generation in the North-
cline depth, surface temperature, surface ern Hemisphere. Comparisons of simulated val-
salinity in the Atlantic (SST, SSS), and the over- ues to observed values for the North Atlantic are
turning calculations at three Atlantic locations. within the uncertainty of the observations. In
Their results suggest that temperature is simu- the Southern Hemisphere, the comparisons in
lated most successfully on a large scale and that all the models are not as good, with the Indian
the overturning transports at 24°N are close Ocean transport estimates contributing to a sig-
(~18 Sv) to observed measurements (~15.8 Sv). nificant part of the mismatch. In coupled ocean-
However, the maximum mean overturning atmosphere simulations, erroneous ocean heat
transports in these models are too high, between transport is compensated by changes in atmos-
21.2 and 31.7 Sv, when compared to the ob- pheric heat transport that give a more realistic
served value (17.7 Sv). Several other CMIP3 total heat transport (Covey and Thompson
models underestimated maximum transport. 1989).
The authors do not attempt to explain why mod-
els are different from each other and from ob- Heat Content. The global mean mass-weighted
servations. ocean temperature is called the ocean’s heat
content. Its time evolution is centrally important
Another aspect of planetary-scale ocean circu- in determining how realistically the models re-
lation of interest is transport of mass by the produce heat uptake. The seasonal cycle and
Antarctic Circumpolar Current through the longer-term trends of heat content provide use-
Drake Passage. The passage, between the tip of ful model metrics, although the seasonal cycle
South America and the Antarctic Peninsula, pro- does not affect the deep ocean. An evaluation of
vides a constrained passage to measure the flow temporally evolving ocean-heat content in the
between two large ocean basins. Observed mean CMIP3 suite of climate models shows the mod-
transport is around 135 Sv (Cunningham et al. els’ abilities to simulate the zonally integrated
2003). Russell, Stouffer, and Dixon (2006, annual and semiannual cycle in heat content. In
2007) estimate passage flow for a subset of cli- the middle latitudes (Gleckler, Sperber, and
mate models. Simulated mean values show a AchutaRao 2006), the models do a reasonable
wide range. For example, GFDL and GISS-EH job, although a broad spread of values is appar-
models do fairly well in reproducing the ob- ent for tropical and polar regions. This analysis
served average transport with values between showed that the models replicate the annual
113 and 175 Sv. Once again, the interaction be- cycle’s dominant amplitude along with its phas-
tween the atmospheric and ocean component ing in the midlatitudes (Figs. 5.11 a–b and 5.12
models appears to be important in reproducing a–f). At high latitudes, comparisons with obser-
the observed transport. The strength and loca- vations are not as consistent. Although the an-
tion of the zonal wind stress provided by the at- nual cycle and global trend are reproduced,
mosphere correlate with how well the transport model analyses (e.g., Hansen et al. 2005a, b)
reflects observed values. show they do not simulate decadal changes in
estimates made from observations (Levitus et
5.2.4.3 N O RTH W ARD H EAT T RAN SPO RT
al. 2001). Part of the difficulty of comparisons
A common metric used to quantify the realism at high latitudes and long periods is the paucity
in ocean models is the northward transport of of observational data (Gregory et al. 2004).
heat. This integrated quantity (from top to bot-
tom and across latitude bands) gives an estimate
of how heat moves within the ocean and is im-
portant in balancing the overall heat exchange
between the tropics and the extratropical regions
76
a)
20 Figure 5.11a–b.
cnr m_cm3
Observed and
csiro_mk3_0
Simulated Zonally
gfdl_cm2_1
Integrated Ocean
giss_aom H eat Content
giss_model_e_h (0–250 m).
15 giss_model_e_r
O bservations are
iap_fgoals1_0_g
represented by the curve
miroc3_2_hires labeled “W O A04.” All
miroc3_2_medres other curves are model
A mplit ude (108J/ m)
ncar_ccsm3_0 results. (a) annual cycle

ukmo_hadcm3 amplitude (108 J/m2) and
W O A04 (b) semiannual/annual
10 (A2/A1). [From Fig. 1 in P.J.
Gleckler, K.R. Sperber, and
K. AchutaRao 2006:Annual
cycle of global ocean heat
content: O bserved and
simulated. J. Geophysical
Research, 111, C06008.
5
Reproduced by
permission of the
American Geophysical
Union (AGU).]
0
90N 75 60 45 30 15 Eq 15 30 45 60 75 90
Latit ude
b)
1.5
cnr m_cm3
csiro_mk3_0
gfdl_cm2_1
giss_aom
giss_model_e_h
giss_model_e_r
iap_fgoals1_0_g
1.0 miroc3_2_hires
miroc3_2_medres
A mplit ude (108J/ m)
ncar_ccsm3_0
ukmo_hadcm3
W O A04
0.5
0
90N 75 60 45 30 15 Eq 15 30 45 60 75 90S
Latit ude
77
cnr m_cm3
a) Global csiro_mk3_0 d) Tropics (20S-20N )
5 gfdl_cm2_1 2
4 giss_aom
3 giss_model_e_h
giss_model_e_r 1
2
iap_fgoals1_0_g
1
miroc3_2_hires
0 0
miroc3_2_medres
–1 ncar_ccsm3_0
–2 ukmo_hadcm3 –1
–3 W O A01
–4 W O A04
–5 –2
J F M A M J J A S O N D J F M A M J J A S O N D
b) N orther n H emisphere e) Souther n O cean

12 2
10
H eat Content (10**22J)

8
6 1
4
2
0 0
–2
–4
–6 –1
–8
–10
–12 –2
c) Souther n H emisphere f) Arctic O cean

14 2
12
10
8 1
6
4
2
0 0
–2
–4
–6 –1
–8
–10
–12
–14 –2
Month Month
Figure 5.12a–f. Annual Cycle of Observed and Simulated Basin Average Global Ocean H eat Content (0–250 m).
O bservations are represented by the curves labeled “W O A01” and “W O A04.” Units are 1022 J. Arctic O cean is defined as north of 60°N ,
and Southern O cean is south of 60°S. [From Fig. 3 i n P.J. Gleckler, K.R. Sperber, and K. AchutaRao 2006:Annual cycle of global ocean heat
content: O bserved and simulated. J. Geophysical Research, 111, C06008. Reproduced by permission of the American Geophysical Union
(AGU).]
78
5.2.5 Global Mean Sea-Level Rise 2003 (www.flooddamagedata.org). Losses from

the 1988 drought were estimated at $40 billion
Two separate physical processes contribute to and the 2002 drought at $11 billion. Heat waves
sea-level rising: (1) ocean thermal expansion in 1995 resulted in 739 additional deaths in
from an increase in ocean heat uptake (steric Chicago alone (Whitman et al. 1997). A large
component) and (2) addition of freshwater from component of overall climate change impacts
precipitation, continental ice melt, and river probably will arise from changes in the inten-
runoff (eustatic component). Various ocean sity and frequency of extreme events.
models handle freshwater fluxes in different
ways. With the addition of a free surface in the Modeling of extreme events poses special chal-
current generation of ocean models, freshwater lenges since they are, by definition, rare. Al-
flux into oceans can be included directly though the intensity and frequency of extreme
(Griffies et al. 2001). The freshwater contribu- events are modulated by ocean and land surface
tion is computed in quantities estimated by the state and by trends in the mean climate state, in-
climate model’s atmosphere and ice-sheet com- ternal atmospheric variability plays a very large
ponents (e.g., Church, White, and Arblaster role, and the most extreme events arise from
2005; Gregory, Lowe, and Tett 2006). In gen- chance confluence of unlikely conditions. The
eral, state-of-the-art climate models underesti- very rarity of extreme events makes statistical
mate the combined global mean sea-level rise evaluation of model performance less robust
as compared to tide gauge and satellite altime- than for mean climate. For example, in evaluat-
ter estimates, while the rise for each separate ing a model’s ability to simulate heat waves as
component is within the observed values’ un- intense as that in 1995, only a few episodes in
certainty. The reason for this is an open research the entire 20th Century approach or exceed that
question and may relate either to observational intensity (Kunkel et al. 1996). For such rare
sampling or to incorrectly accounting for all eu- events, estimates of the real risk are highly un-
static contributions. The steric component to certain, varying from once every 30 years to
global mean sea-level rise is estimated at 0.40 ± once every 100 years or more. Thus, a model
0.05 mm/yr from observations (Antonov, Levi- that simulates these occurrences at a frequency
tus, and Boyer 2005). Models simulate a similar of once every 30 years may be performing ade-
but somewhat smaller rise (Gregory, Lowe, and quately, but its performance cannot be distin-
Tett 2006; Meehl et al. 2005). Significant dif- guished from that of the model that simulates a
ferences also occur in the magnitudes of frequency of once every 100 years.
decadal variability between observed and simu-
lated sea level. Progress is being made, however, Although it might be expected that a change in
over the previous generation of climate models. mean climate conditions will apply equally to
When atmospheric effects from volcanic erup- changes in extremes, this is not necessarily the
tions are included, for example, current-gener- case. Using as an example the 50-state record-
ation ocean models capture the volcanoes’ low temperatures, the decade with the largest
observed impact (a decrease in the global mean number of records is the 1930s, yet winters dur-
sea level). Figure 5.13 from Church, White, and ing that decade averaged third warmest since
Arblaster (2005) gives an example of a few 1890; in fact, no significant correlation is
models and their detrended estimate of the his- shown between the number of records and U.S.
toric global mean sea level. It shows the influ- wintertime temperature (Vavrus et al. 2006).
ence of including additional atmospheric Thus, the severest cold air outbreaks in the past
forcing agents in changing the ocean’s steric do not necessarily coincide with cold winters.
height. Another examination of model data showed that
future changes in extreme temperatures differ
5.3 EX T REME EVEN T S from changes in mean temperature in many re-
gions (Hegerl et al. 2004). This means that cli-
Flood-producing precipitation, drought, heat mate model output must be analyzed explicitly
waves, and cold waves have severe impacts on for extremes by examining daily (or even finer–
North America. Flooding resulted in average resolution) data, a resource-intensive effort.
annual losses of $3.7 billion between 1983 and
79
22
(a)
8x10
H eat content Levit us Model (PCM)
Ishii
22
4x10
H eat content (J)
22
−4x10
Pinat ubo
A gung El Chichon
22
−8x10
1960 1965 1970 1975 1980 1985 1990 1995 2000
(b)
10
GMS L Levit us Model (PCM)
Ishii
5
GMSL ( mm)
−5
Pinat ubo
A gung El Chichon
−10
1960 1965 1970 1975 1980 1985 1990 1995 2000
(c)
10
GMS L MIRO C3.2(hires) GISS−ER
MIRO C3.2( medres) PCM
5
GMSL ( mm)
−5
Pinat ubo
A gung El Chichon
−10
1960 1965 1970 1975 1980 1985 1990 1995 2000
Figure 5.13. Observed and Modeled Global Ocean H eat Content (GOH C) and Global Mean Sea Level (GMSL)
for 1960 to 2000.
The response to volcanic forcing, as indicated by differences between pairs of PCM simulations for GO H C (a) and GMSL (b) is shown for
the ensemble mean (bold line) and the three ensemble members (light lines). O bservational estimates of GO H C and GMSL are shown by
the black and blue bold lines. For a and b, all results are for the upper 300 m only and have been detrended over the period 1960 to 2000.
For c, the ensemble mean (full-depth) GMSL for GISS-ER, MIRO C3.2(hires), MIRO C3.2(medres), and PCM models (after subtracting a
quadratic) are shown. [From Fig. 2 in J.A. Church, N .J. W hite, and M. Arblaster 2005: Significant decadal-scale impact volcanic eruptions on
sea level and ocean heat content. N ature, 438(7064), 74–77. Used with permission from N ature Publishing Group.]
80
Evaluation of model performance with respect meterization of subgrid-scale processes, partic-

to extremes is hampered by incomplete data on ularly convection (Chapter 2; Emori and Brown
historical frequency and severity of extremes. 2005; Iorio et al. 2004).
Frich et al. (2002) analyzed ten indicators of cli-
5.3.1 D RO UGH TS AN D EX CESSIVE RAIN FALL
mate extremes and presented global results.
LEAD IN G TO FLO O D S
However, many areas were missing due to lack
of suitable station data, particularly in the trop- Recent analysis indicates a globally averaged
ics. Using some of these indices for compar- trend toward greater areal coverage of drought
isons between models and observations has since 1972 (Dai et al. 2004). A simulation by
become common. Another challenge for model the HadCM3 model reproduces this dry trend
evaluation is the spatially averaged nature of (Burke, Brown, and Christidis 2006) only if an-
model data, representing an entire grid cell, thropogenic forcing is included. A control sim-
while station data represent point observations. ulation indicates that the observed drying trend
For some comparisons, averaging station data is outside the range of natural variability. The
over areas representing a grid cell is necessary. model, however, does not always correctly sim-
ulate the regional distributions of areas of in-
Several approaches are used to evaluate model creasing wetness and dryness. The relationship
performance for simulation of extremes. One between droughts and variability was covered
approach examines whether a model reproduces above in Section 5.2.2.3 Monsoons.
the magnitude of extremes. For example, a daily
rainfall amount of 100 mm or more is expected Several different measures of excessive rainfall
to occur about once every year in Miami, every have been used in analyses of model simula-
6 years in New York City, every 13 years in tions. A common one is the annual maximum
Chicago, and every 200 years in Phoenix. A 5-day precipitation amount, one of the Frich et
useful metric would be the extent to which a al. (2002) indices. This has been analyzed in
model is able to reproduce absolute magnitudes several recent studies (Kiktev et al. 2003;
and spatial variations of such extremes. A sec- Hegerl et al. 2004; Tebaldi et al. 2006). Other
ond approach examines whether a model repro- analyses have examined thresholds of daily pre-
duces observed trends in extremes. Perhaps the cipitation, either absolute (e.g., 50 mm/day in
most prominent observed global trend is an in- Dai 2006) or percentile (e.g., 4th-largest precip-
crease in the frequency of heavy precipitation, itation event equivalent to 99th percentile of 365
particularly during the last 20 to 30 years of the daily values as in Emori et al. 2005). Recent
20th Century. This trend is significant at the 95% studies of model simulations produced for
confidence level for the period 1979 to 2003 CMIP3 provide information on the performance
and at the 99% confidence level for the period of the latest model generation.
1951 to 2003 (Trenberth et al. 2007). Another
notable observed trend is an increase in the Models generally tend to underestimate very
length of the frost-free season. heavy precipitation. This is shown in a compar-
ison between satellite (TRMM) estimates of
In some key respects, model simulation of tem- daily precipitation and model-simulated values
perature extremes probably is less challenging within the 50°S–50°N latitude belt (Dai 2006).
than simulating precipitation extremes, in large TRMM observations derive 7% of total precip-
part due to the scales of these phenomena. The itation from very heavy rainfall of 50 mm or
typical heat wave or cold wave covers a rela- more per day, in contrast to only 0 to 2% for the
tively large region, on the order of several hun- models. For the frequency of very heavy pre-
dred miles or more or a number of grid cells in cipitation of 50 mm or more per day, TRMM
a modern climate model. By contrast, heavy data show a frequency of 0.35% (about once
precipitation can be much more localized, often every 300 days), whereas it is 0.02 to 0.11%
extending over regions of much less than 150 (once every 900 to 5000 days) for the models. A
km, or less than the size of a grid cell. Thus, the global analysis of model simulations showed
modern climate model can simulate directly the that models produced too little precipitation in
major processes causing temperature extremes events exceeding 10 mm/day (Sun et al. 2006).
while heavy precipitation is sensitive to para- Examining how many days it takes to accumu-
81
late two-thirds of annual precipitation, models such as North America, is of interest. These
generally show too many days compared to ob- models have spatial resolution sufficient to re-
servations over North America, although a few solve major mountain chains, and some thus
models are close to reality. In contrast to the display considerable skill in areas where topog-
general finding of a tendency toward underesti- raphy plays a major role in spatial patterns. For
mation, a study (Hegerl et al. 2004) of two mod- example, they are able to reproduce rather well
els indicates generally good agreement with the spatial distribution of the magnitude or ex-
observed annual maximum 5-day precipitation tent of precipitation in the 95th percentile (Leung
amounts over North America for HadCM3 and and Qian 2003), frequency of days with more
even somewhat of an overestimation for than 50 mm and 100 mm (Kim and Lee 2003),
CGCM2. frequency of days over 25 mm (Bell, Sloan, and
Snyder 2004), and annual maximum daily pre-
This model tendency to produce rainfall events cipitation amount (Bell, Sloan, and Snyder
less intense than observed appears to be due in 2004) over the western United States. Kunkel et
part to global models’ low spatial resolution. al. (2002) found that an RCM’s simulation of
Experiments with individual models show that extreme-event magnitude over the United States
increasing resolution improves the simulation varied spatially and depended on event duration.
of heavy events. For example, the fourth-largest There was a tendency for overestimation in
precipitation event in a model simulation with a western United States and good agreement or
resolution of about 300 km averaged 40 mm underestimation in central and eastern United
over the conterminous United States, compared States.
to an observed value of about 80 mm. When the
resolution was increased to 75 km and 50 km, Most studies of observed precipitation extremes
the fourth-largest event was still smaller than suggest that they have increased in frequency
observed but by a much smaller amount (Iorio and intensity during the latter half of the 20th
et al. 2004). A second important factor is the pa- Century. A study by Tebaldi et al. (2006) indi-
rameterization of convection. Thunderstorms cates that models generally simulate a trend to-
are responsible for many intense events, but ward a world characterized by intensified
their scale is smaller than the size of model precipitation, with a greater frequency of heavy-
grids and thus must be indirectly represented in precipitation and high-quantile events, although
models (Chapter 2). One experiment showed with substantial geographical variability. This is
that changes to this representation improve in agreement with observations. Wang and Lau
model performance and, when combined with (2006) find that CGCMs simulate an increasing
high resolution of about 1.1° latitude, can pro- trend in heavy rain over the tropical ocean.
duce quite-accurate simulations of the fourth-
largest precipitation event on a globally 5.3.2 H eat and Cold W aves
averaged basis (Emori et al. 2005). Another ex-
periment found that the use of a cloud-resolv- Analyses of simulations for IPCC AR4 by seven
ing model imbedded in a global model climate models indicate that they reproduce the
eliminated underestimation of heavy events primary features of cold air outbreaks (CAOs),
(Iorio et al. 2004). A cloud-resolving model with respect to location and magnitude (Vavrus
eliminates the need for convection parameteri- et al. 2006). In the analyses, a CAO is an
zation but is very expensive to run. These sets of episode of at least 2 days duration during which
experiments indicate that the problem of heavy- the daily mean winter (December-January-
event underestimation may be reduced signifi- February) surface temperature at a gridpoint is
cantly in future as increases in computer power two standard deviations below the gridpoint’s
allow simulations at higher spatial resolution winter mean temperature. Maximum frequen-
and perhaps eventually the use of cloud-resolv- cies of about four CAO days per winter are sim-
ing models. ulated over western North America and Europe,
while minimal occurrences of less than one day
Improved model performance at higher spatial per winter exist over the Arctic, northern Africa,
resolutions provides motivation for use of re- and parts of the North Pacific. GCMs generally
gional climate models when only a limited area, are accurate in their simulation of primary fea-
82
tures, with high pattern correlation to observa- Niño events, and thus frost-day trends reflect a
tions and maximum number of days meeting more general response to forcings that occurred
CAO criteria around 4 per winter. One favored during the latter part of the 20th Century. An
region for CAOs is in western North America, analysis of short-duration heat waves simulated
extending from southern Alaska into the upper by PCM (Meehl and Tebaldi 2004) indicates
Midwest. Here, models simulate a frequency of good agreement with observed heat waves for
about 4 CAO days per year, in general agree- North America. In that study, heat waves were
ment with the observed values of 3 to 4 days. defined by daily minimum temperature. The
Models underestimate frequency in the south- most intense events occurred in the southeastern
eastern United States (mean simulated values United States for both model simulation and ob-
range from 0.5 to 2 days vs 2 to 2.5 days in observations. The overall spatial pattern of heat-
servations). This regional bias occurs in all wave intensity in the model matched closely
models and reflects the inability of GCMs to with the observed pattern. In a four-member en-
penetrate Arctic air masses far enough south- semble of simulations from HadCM3 (Chris-
eastward over North America. tidis et al. 2005), the model showed a rather
uniform pattern of increases in the warmest
CMIP3 model simulations show a positive trend night for 1950 to 1999. Observations also show
for growing season, heat waves, and warm a global mean increase, but with considerable
nights and a negative trend for frost days and regional variations. In North America, observed
daily temperature range (maximum minus min- trends in the warmest night vary from negative
imum) (Tebaldi et al. 2006). The simulations in- in the south-central sections to strongly positive
dicate that this is in general agreement with in Alaska and western Canada, compared to a
observations, except that there is no observed rather uniform pattern in the model. However,
trend in heat waves. The modeled spatial pat- this discrepancy might be expected, since the
terns generally have larger positive trends in observations probably reflect a strong imprint
western North America than in eastern sections. of internal climate variability that is reduced by
For the United States, this is in qualitative ensemble averaging of the model simulations.
agreement with observations showing that de-
creases in frost-free season and frost days are Analysis of the magnitude of temperature ex-
largest in the western United States (Kunkel et tremes for California in a regional climate
al. 2004; Easterling 2002). model simulation (Bell, Sloan, and Snyder
2004) shows mixed results. The hottest maxi-
Analysis of individual models provides a more mum in the model is 4°C less than observations,
detailed picture of model performance. In a sim- while the coldest minimum is 2.3°C warmer.
ulation from PCM (Meehl, Tebaldi, and Nychka The number of days >32°C is 44 in the model
2004), the largest trends for decreasing frost compared to an observed value of 71. This
days occur in the western and southwestern could result from the lower diurnal temperature
United States (values greater than –2 days per range in the model (15.4°C observed vs 9.7°C
decade). Trends near zero in the upper Midwest simulated). While these results are better than
and northeastern United States show good the driving GCM, RCM results are still some-
agreement with observations. The biggest dis- what deficient, perhaps reflecting the study re-
crepancy between model and observations is gion’s very complex topography.
over parts of the southeastern United States,
where the model shows trends for decreasing Models display some capability to simulate ex-
frost days and observations show slight in- treme temperature and precipitation events, but
creases. This is thought to be a partial conse- there are differences from observed character-
quence of two large El Niño events in istics. Models typically produce global in-
observations during this time period (1982– creases in extreme precipitation and severe
1983 and 1997–1998) when anomalously cool drought and decreases in extreme minimum
and wet conditions occurred over the southeast- temperatures and frost days, in general agree-
ern United States and contributed to slight in- ment with observations. Models have a general,
creases of frost days. The model’s ensemble though not universal, tendency to underestimate
mean averages out effects from individual El the magnitude of heavy precipitation events.
83
Regional-trend features are not always captured.

Since the causes of observed regional-trend
variations are not known in general and such
trends could be due in part to the climate sys-
tem’s stochastic variability, assessing the sig-
nificance of these discrepancies is difficult.
84
6CH APTER
Future Model
D evelopment
Climate models are evolving toward greater comprehensiveness, incorporating such aspects of the
chemical and biological environment as active vegetation on land and oceanic biogeochemistry that
affect and are affected by the physical climate. Climate models are simultaneously evolving to-
ward finer spatial resolution.
Improvements in climate simulations as resolution increases can be both incremental and funda-
mental. Incremental improvements are expected in treatment of the atmosphere due to better
simulation of atmospheric fronts, interactions among extratropical storms and sharp topographic
features, and, especially, tropical storms. In the ocean, finer resolution incrementally improves the
simulation of narrow boundar y currents and the circulation in relatively small basins, such as the
Labrador Sea, that play key roles in oceanic circulation.
More fundamental changes also happen in both the atmosphere and the ocean as resolution im-
proves. In the ocean a key transition occurs at grid scales of tens of kilometers, at which point
mesoscale eddies (see Chapter 2) begin to be explicitly resolved. In the atmosphere, a funda-
mental transition takes place when the grid scale drops to a few kilometers, where direct simu-
lation of dominant deep convective circulations begins to be feasible and the model’s dependence
on uncertain subgrid-scale parameterization of deep moist convection diminishes.
In the following, we discuss these more funda- physical parameters so as to better estimate the
mental oceanic and atmospheric transitions and associated uncertainties [quantifying uncer-
then describe some examples of increased com- tainty in model predictions (called QUMP);
prehensiveness in climate modeling (see also Murphy et al. 2004; climateprediction.net]. Oth-
Chapter 2 for glacial modeling, another impor- ers include the movement toward initializing cli-
tant future development). mate models with estimates of observed
climatic states, particularly the observed
The climate modeling enterprise is evolving oceanic state, so as to optimize the realism of
along additional paths (apart from evolution of decadal forecasts, which marks an evolution to-
the models themselves) that are not discussed ward the merging of seasonal-interannual and
here. One path is the creation of large ensem- decadal forecasting (Troccoli and Palmer 2007).
bles of model simulations by varying uncertain
85
The U.S. Climate Change Science Program Chapter 6 - Future Model D evelopment
6.1 H IGH -RESOLU T ION MODELS to 1/6 of a degree. A much more turbulent flow
is simulated by the model with abundant vortex
6.1.1 Mesoscale Eddy-Resolving generation. This model is beginning to resolve
Ocean Models the spectrum of mesoscale eddies that populate
the Southern Ocean and many other oceanic re-
The distinction between laminar and turbulent gions. As discussed in Chapter 2, the effects on
flow in the ocean is fundamental. Simulations ocean circulation of mesoscale eddy-induced
of the more realistic turbulent regime promise mixing are parameterized in current ocean mod-
to substantially raise the level of realism in els, which can be thought of as essentially lam-
oceanic climate simulations. For example, Fig. inar.
6.1 shows two simulations of the Southern
Ocean by an ocean model developed at the Geo- While progress has been made in recent years,
physical Fluid Dynamics Laboratory (GFDL) explicit simulation of these eddies undoubtedly
(Hallberg and Gnanadesikan 2006). The field is more reliable than mixing parameterizations.
shown is an instantaneous snapshot of the sur- In the Southern Ocean, eddies are thought to
face current speed. Resolution of the model on control the circumpolar current’s response to
the left is about 1° latitude. The result is a rela- wind changes (Hallberg and Gnanadesikam
tively laminar (nonturbulent) flow with a gen- 2006) and the way carbon dioxide is taken up
tly meandering circumpolar current. The figure by the Southern Ocean.
on the right is obtained by reducing the grid size
Figure 6.1. Surface-Current Speed in Two Simulations of the Southern Ocean in Low-
and H igh-Resolution Ocean Models.
[From Fig. 6 in R. H allberg and A. Gnanadesikam 2006:The role of eddies in determining the structure and
response of the wind-driven Southern H emisphere overturning: Results from the modeling eddies in the
Southern O cean (MESO ) project. J. Physical O ceanography, 36, 2232–2252. Reproduced by permission of
the American Meteorological Society (AMS).]
86
Global mesoscale eddy-resolving ocean models Like AGCMs, CRMs must employ empirical
are beginning to be examined in various mod- parameterizations to calculate the impact of
eling centers in the United States and around the subgrid scale processes, but CRMs explicitly
world, even though exploiting such models will represent a larger portion of the size spectrum
require substantial increases in computational of meteorological systems, so the parameteri-
resources. Challenges that may arise when these zations’ impact on large-scale circulation and
models are integrated for long time periods in- climate may be less severe. Most important, cu-
clude maintaining realistically small amounts of mulus parameterizations for deep tropical con-
mixing across constant-density surfaces in the vection are not needed in CRMs. CRMs can
more turbulent flows to avoid distortion of accommodate more realistic microphysical
much slower thermohaline circulations. processes, including those by which aerosols
nucleate cloud drops, allowing more convinc-
As noted in Chapter 5, models provide esti- ing treatment of aerosol and cloud interactions
mates of the climate system’s centennial-scale involved in indirect aerosol radiative forcing.
variability that underlies attribution studies of
climatic trends. Seeing if eddy-resolving However, shallow nonprecipitating convection
OGCMs increase the variability level on long (which produces fair-weather cumulus clouds)
time scales in climate models will be of great is dominated by flows on scales less than 1 km
interest. and will probably still require subgrid-scale pa-
rameterization in foreseeable global CRMs.
6.1.2 Cloud-Resolved Atmospheric Cloud feedbacks in regions of shallow convec-
Models tion are an important source of disparity in cli-
mate sensitivity in CMIP3 models (Bony et al.
As atmospheric models attain higher resolution 2006). Furthermore, most cloud microphysical
and more detailed representation of physical processes take place on CRM subgrid scales
processes, short-range weather prediction and and so must be parameterized. Thus, uncertainty
longer-range climate prediction become more in cloud feedbacks will not disappear when
synergistic (Phillips et al. 2004). This is partic- global CRMs begin to play a role in climate as-
ularly evident in “cloud-resolving models” sessments, but modelers hope that uncertainty
(CRMs) with spatial resolutions of less than a will be reduced substantially.
few kilometers. CRMs can explicitly simulate
atmospheric systems that exist on scales much Global models with CRM resolution have been
smaller than the grid resolution of conventional attempted to date only at the Japanese Earth
atmospheric general circulation models Simulator, but, with continued increase in com-
(AGCMs) (Randall et al. 2003; Khairoutdinov, puter power, global CRMs are expected to be-
Randall, and DeMott 2005). These systems in- come centrally important in climate (as well as
clude mesoscale organizations in squall lines, weather) research. Nevertheless, as noted above,
deep updrafts and downdrafts, and cirrus anvils. major uncertainties in cloud microphysics will
CRMs also allow calculation of cloud proper- remain, especially in the prediction of ice-parti-
ties and amounts based on more realistic small- cle concentrations, fall speed of cloud particles,
scale structure in the flow field. The desired hydrometeorological spectra evolution, and en-
result is not only better simulations of regional trainment rates into convective plumes (Cotton
climates, especially in the tropics, but also more 2003). At CRM resolutions, more sophisticated
reliable estimates of cloud feedbacks and cli- algorithms of radiative-transfer calculation than
mate sensitivity. those in current GCMs may be required because
the plane parallel assumption for convergence of
CRMs are variations of models designed for radiant energy may not be valid. Validation of
mesoscale storm and cumulus convection sim- CRMs probably will continue to take place in re-
ulations. At CRM grid scales, hydrostatic bal- gional models and short-range forecasts, fol-
ance is no longer universally valid. CRMs are lowed by their incorporation into global models.
therefore formulated with nonhydrostatic equa-
tions in which vertical accelerations are calcu- Several observational programs such as the
lated explicitly (Tripoli 1992). DOE Atmospheric Radiation Measurement
87
(ARM) Program have collected data essential to Feedbacks between the physical climate system
evaluate CRMs (M.H. Zhang et al. 2001; Tao et and the carbon cycle are represented plausibly
al. 2004). Results from such programs will facil- but with substantial differences in various
itate improvement of CRM subgrid-scale AOGCM carbon-cycle models. Cox et al.
physics. Extensive parameter-sensitivity tests (2000) obtained a very large positive feedback,
with global models will still be needed to reduce with global warming reducing the fraction of
uncertainties in microphysics and the treatment anthropogenic carbon absorbed by the bios-
of shallow convection for climate sensitivity and phere, thus boosting the model’s simulated at-
regional climate-change simulation. mospheric CO2. Friedlingstein et al. (2001)
obtained much weaker feedback. Thompson et
6.2 BIOGEOCH EMIST RY AN D al. (2004) demonstrated that making different
CLIMAT E MODELS assumptions about the land biosphere within a
single model gave markedly different feedback
6.2.1 Carbon Cycle values. Using the same model, Govindasamy et
al. (2005) noted a positive correlation between
The physical climate system and biogeochemi- the magnitude of carbon-cycle feedback and the
cal processes are tightly coupled. Changes in sensitivity of the physical climate system.
climate affect the exchange of atmospheric CO2
between land surface and ocean, and changes in A recent study examined carbon-cycle feed-
CO2 fluxes affect Earth’s radiative forcing and backs in 11 coupled AOGCM carbon-cycle
thus the physical climate system. Some recently models using the same forcing (Friedlingstein
developed atmosphere-ocean general circula- et al. 2006). The models unanimously agreed
tion models (AOGCMs) include the carbon that global warming will reduce the fraction of
cycle and have confirmed the potential for anthropogenic carbon absorbed by the bios-
strong feedback between it and global climate phere—a positive feedback—but the magnitude
(Cox et al. 2000; Friedlingstein et al. 2001; of this feedback varied widely among models
Govindasamy et al. 2005). The next generation of (Fig. 6.3). When models included an interactive
AOGCMs may include the carbon cycle as well as carbon cycle, predictions of the additional
interactive atmospheric aerosols and chemistry. global warming due to carbon-cycle feedback
Models that include the carbon cycle are able to ranged between 0.1 and 1.5°C. Eight models at-
predict time-evolving atmospheric CO2 concen- tributed most of the feedback to the land bios-
trations using, as input, anthropogenic emissions phere, while three attributed it to the ocean.
rather than assumed concentrations.
These results demonstrate the large sensitivity
Simulation of the global carbon cycle must ac- of climate model output to assumptions about
count for the processes shown in Fig. 6.2. As the carbon-cycle processes. Future carbon-cycle
figure shows, the present-day global carbon cycle models, coupled to physical climate models and
is not in equilibrium because of fossil-fuel burn- constrained by new global remote-sensing
ing and other anthropogenic carbon emissions. datasets and in situ measurements, may allow
These carbon sources must, of course, be in- more definitive projection of CO2 concentra-
cluded in models of climate change. Such a cal- tions in the atmosphere for given emission sce-
culation is not easy because human-induced narios. CCSP SAP 2.2 contains more
changes to the carbon cycle are small compared to information on the carbon cycle and climate
large natural fluxes, as shown in the figure. In ad- change (CCSP 2007).
dition, although the globally and annually aver-
aged carbon reservoirs and fluxes shown in the 6.2.2 Other Biogeochemical Issues
figure are consistent with estimates from a variety
of sources, substantial uncertainties are attached Methane (CH4) is a potent greenhouse gas
to the numbers (e.g., often a factor >2 uncertainty whose atmospheric concentration is controlled
for fluxes; see Prentice et al. 2001). Additional by its emission rate and the atmosphere’s oxida-
uncertainty applies to regional, seasonal, and in- tive capacity (especially hydroxyl radical con-
terannual variations in the carbon cycle. centration). Methane concentrations are now
much higher than in preindustrial times but have
88
Figure 6.2. Global

Global Carbon Cycle as Seen by an AOGCM Carbon Cycle from
the Point of View of
At mosphere
Existing Physical
Climate System
760 Models (Coupled
123
photosynthesis
AOGCMs).
O cean The four boxes represent
Land sur face 92 atmosphere, land surface,
60 91 sur face dissolved
GPP
plant 60 ocean-at m inor ganic ocean, and sea ice— major
} N PP resp. microbial
respiration
s
exchange
920
components of AO GCMs.
Earth system models will
2 r iver
vegetation land-cover photosynthesis 50 39 remineralization evolve from AO GCMs by
1
and soils change
marine biota incorporating relevant
2300 3 biogeochemical cycles into
92 101 the four-box framework
deep-sur face
6 11 ocean (with sea ice not acting as a
fossil fuels fossil- fuel bur ning exchange carbon reser voir). N umbers
and cement production
3500 shown are average values for
deep ocean the 1990s. Small ( ≤1
Sea ice 37000 PgC/year) fluxes such as
Circled numbers = those involving methane are
storage in gigatonnes 0.2 not shown, except for burial
(or petagrams) carbon ocean sediments of 0.2 PgC/year in ocean-
150 bottom sediments, assuming
O ther numbers =
a 50-50 split between plant
fluxes in gigatonnes and microbial respiration.
carbon/year
Figure 6.3.Time Series

1000 of Atmospheric CO 2
from 11 Different
At mospheric CO 2 (ppm)
900 AOGCM Carbon-

Cycle Models.
800 [From Fig. 1(a) of P.
Friedlingstein et al. 2006:
700 Climate-carbon cycle
feedback analysis: Results
600 from the C4MIP model
intercomparison. J. Clim ate ,
500 19, 3337–3353. Reproduced
by permission of the
400 American Meteorological
Society (AMS).]
300
1850 1900 1950 2000 2050 2100
Year
not increased in the past decade, for reasons that carbon by plants. To address this process, dy-
continue to be debated. Whether or not this trend namic vegetation models (in which plant growth
carries into the future has substantial implications is calculated rather than specified a priori) are
for radiative forcing. To resolve this question, under development at modeling centers in the
AOGCMs would need to include atmospheric United States and elsewhere. This inclusion of a
chemistry models incorporating a number of dif- wider range of processes poses challenges [e.g.,
ferent trace gases and reaction rates. it amplifies errors in rainfall prediction (Bonan
and Levis 2006)]. In addition, ecosystems fer-
Another emerging issue is the interactive evo- tilized with CO2 are limited by the availability
lution of climate with the storage of water and of nutrients such as nitrogen and phosphorous
89
that are important to the carbon cycle (Field, working_groups/Biogeo/reports/060328_BGC

Jackson, and Mooney 1995; Schimel 1998; WGrpt.pdf); GFDL’s Earth system model
Nadelhoffer et al. 1999; Shaw et al. 2002; Hun- (gfdl.noaa.gov/~jpd/ esmdt.html); Doney et al.
gate et al. 2003). Future climate-carbon models 2004]. One issue receiving particular attention
probably will need to include these nutrients. in recent years is that ocean productivity may
The few models that do so now show less plant be increased through iron fertilization via dust
growth in response to increasing atmospheric particles, potentially reducing atmospheric CO2
CO2 (Cramer et al. 2001; Oren et al. 2001; (Martin 1991). This effect is being assessed by
Nowak, Ellsworth, and Smith 2004). Incorpo- both observational programs (e.g., Bishop,
ration of other known limiting factors such as Davis, and Sherman 2002) and climate-carbon
acclimation of soil microbiology to higher tem- models (Jickells et al. 2005).
peratures (Kirschbaum 2000; Tjoelker, Oleksyn,
and Reich 2001) will be important in develop- An important challenge to these efforts is the
ing comprehensive Earth system models. complexity of ocean ecosystems. Adding to this
Aerosol modeling also will be a central element complexity are organisms that fix nitrogen and
in future models (this subject will be covered by denitrify, calcify, or silicify; accounting for each
CCSP SAP 2.3, whose estimated publication adds parameterizations and variables to the sys-
date is June 2008). tem (Hood et al. 2006). Biological models need
to be sufficiently complex to capture the ob-
Often, climate-carbon simulations include nat- served variability on various time scales, since
ural ecosystems but do not include the effects this variability provides essential tests for the
of human land-cover and land-management models. As in many aspects of climate model-
changes (e.g., deforestation and reforestation). ing, however, complexity that outgrows the abil-
Land-cover change often is accounted for sim- ity to constrain models with available data
ply by prescribing estimates for the historical should be avoided (Hood et al. 2006).
period (e.g., Houghton 2003) and for future sce-
narios from the IPCC Special Report on Emis- Modeling groups have undertaken systematic
sions Scenarios (IPCC 2000). These estimates comparison of different models in the Ocean
do not include practices such as crop irrigation Carbon Cycle Model Intercomparison Project
and fertilization. Many models with “dynamic (OCMIP) under the auspices of the Interna-
vegetation” do not actually simulate crops; they tional Geosphere-Biosphere Programme.
only allow natural vegetation to grow. Defor- OCMIP’s most recent phase involved 13
estation, land cultivation, and related human ac- groups—including several from the United
tivities probably will be included in at least States—implementing a common biological
some future AOGCMs, enabling more complete model in their different OGCMs (Najjar et al.
assessment of total anthropogenic effects on the 2007). The common biological model includes
global climate and environment (Ramankutty et five prognostic variables: inorganic phosphate
al. 2002; Root and Schneider 1993). (PO42–), dissolved organic phosphorus (DOP),
dissolved oxygen (O2), dissolved inorganic car-
6.2.3 Ocean Biogeochemistry bon (CO2 + HCO3– + CO32–), and total alkalin-
ity (the system’s acid- and base-buffering
Climate change impacts on the marine environ- capacity). Model intercomparison revealed sig-
ment—including changes in the ocean’s biota nificant differences in simulated biogeochemi-
and carbon content due to modified ocean tem- cal fluxes and reservoirs. A biogeochemistry
perature, salinity, and circulation patterns— model’s realism was found to be tied closely to
must be accounted for, along with terrestrial the dynamics of the simulation’s ocean circula-
biogeochemistry, in a complete Earth system tion. Just as for land vegetation modeling, a se-
model. Implementation of ocean biogeochem- rious challenge to climate models is presented
istry processes into AOGCMs is under way by the quality of the physical climate simulation
to improve simulation of the ocean carbon required for realistic biogeochemical modeling.
cycle under various scenarios [e.g., “CCSM
Biogeochemistry Working Group Meeting
Report,” March 2006 (www.ccsm.ucar.edu/
90
7CH APTER
Example Applications
of Climate Model Results
I n this chapter we present several cases where climate model simulation results were used for
studies involving actual and potential end-user applications. W ith the increased availability of cli-
mate model simulation output through the CMIP3 multimodel archive, impacts and applications
users are rapidly applying the model results for their needs. Just as quickly, the breadth and diversity
of applications will continue to grow in the future as climate statistics are no longer considered
stationar y.The examples discussed in this chapter are meant for illustration and do not constitute
a complete accounting of all published instances of applications from model results.The influence
of climate, and therefore climate change, on different natural and societal systems is quite varied.
Some impacts of climate change result primarily from changes in mean conditions. O ther impacts
are sensitive to climate variability— the sequence, frequency, and intensity of specific weather
events. N ote that the climate simulations described below are not offering predictions of 21st
Centur y climate but simply projections of possible climate scenarios. Prediction requires know-
ing in advance how climatic forcings, including those produced by humans, would change in the fu-
ture. SAP 3.2 examines climate projections by CMIP3 models in greater detail.
7.1 APPLYIN G MODEL RESU LT S those presented below because of simulation bi-
TO IMPACT S ases and the coarse spatial resolution of typical
global simulations. Although the use of climate
As shown in previous chapters, climate models projections for impacts is beyond the scope of
give approximate renditions of real climate. this report, aspects of the methodology for using
Consequently, applications of climate model re- the projections are based on the models’ abili-
sults to impact studies require consideration of ties to simulate observed climate. Employing
several limitations that characterize model out- coarse-resolution global model output for re-
put. In principle, using the direct output of cli- gional and local impact studies requires two ad-
mate models is desirable because these results ditional steps—downscaling, as discussed in
represent a physically consistent picture of fu- Chapter 3, and bias removal, or the adjustment
ture climate, including changes in climate vari- of future projections for known systematic
ability and the occurrence of such various model errors, described in Chapters 2 and 5.
weather phenomena as extreme events. In prac-
tice, this is rarely done for applications like
91
The U.S. Climate Change Science Program Chapter 7 - Example Applications of Climate Model Results
7.1.1 Downscaling 7.2 CALIFORN IA CLIMAT E

CH AN GE ASSESSMEN T S
Downscaling is required because of the limita-
tions of coarse spatial resolution in the global One of the most comprehensive uses of climate
models. In mountainous terrain, a set of model model simulation output for applications is
values for a single grid box will represent con- overseen by the California Climate Change
ditions at the mean elevation level of that grid Center. The center was established by a state
box. In reality, however, conditions at moun- agency, the California Energy Commission
taintop and valley locations will be much dif- (CEC), through its Public Interest Energy Re-
ferent. Such processes as local snowpack search program (CEC 2006). The center wanted
accumulation and melting cannot be studied ac- to determine possible impacts of climate change
curately with direct model output. Resolution on California and utilized the CMIP3 model
also limits the accuracy of representation of simulation database as its starting point for cli-
small-scale processes. A prominent example is mate change projections.
precipitation. The occurrence of heavy down-
pours is an important climate feature for certain To generate future California scenarios, re-
impacts, but these events are often localized on searchers selected three climate models from
a scale smaller than a grid box. In many actual the CMIP3 multimodel archive: the National
situations, an area the size of a grid box may ex- Center for Atmospheric Research–U.S. Depart-
perience flooding rains at some points while ment of Energy PCM, the NOAA GFDL
others receive no rain at all. As a result, grid- CM2.1, and the Hadley Centre HadCM3 (Hay-
box precipitation tends to be more frequent, and hoe et al. 2004; Cayan et al. 2006). The models
the largest values typically are smaller than were chosen in large part because of their abil-
those observed at the local scale. Chapter 3 cov- ity to simulate both large-scale global climate
ered both dynamical downscaling with nested features and California’s multiple climatic re-
regional models and statistical downscaling gions when simulations of the 20th Century were
methods that include diverse techniques such as compared with high-resolution observations. Of
weather generators, transfer functions, and particular importance was the correct simula-
weather typing. tion of the state’s precipitation climatology, with
a pronounced wet season from November to
7.1.2 Bias Removal March, during which nearly all annual precipi-
tation falls. Further, these three models offered
A simple approach developed for bias removal a range of sensitivities, with transient climate
during the early days of climate change assess- responses of 1.3 K for PCM, 1.5 K for CM2.1,
ments and still widely used today is sometimes and 2.0 K for HadCM3. Following model se-
dubbed the “delta” method. Climate model out- lection, projections from three scenarios with
put is used to determine future change in cli- low, medium, and high future greenhouse gas
mate with respect to the model’s present-day emissions were chosen to span the range of pos-
climate, typically a difference for temperature sible future California climate states in the 21st
and a percentage change for precipitation. Then, Century. The California scenarios employed a
these changes are applied to observed historical statistical downscaling technique that, used ob-
climate data for input to an impacts model. The servationally, derived probability density func-
delta method assumes that future model biases tions for surface temperature and precipitation
for both mean and variability will be the same to produce corrected model-simulated distribu-
as those in present-day simulations. One highly tion functions (Cayan et al. 2006). Corrections
questionable consequence of this assumption is were then applied to future scenario simulation
that the future frequency and magnitude of ex- results. Once the scenarios were generated, they
treme weather events are the same relative to the were used to quantify possible climate change
mean climate of the future as they are in pres- impacts on public health, water resources, agri-
ent-day climate. Other bias-removal methods culture, forests, and coastal regions (CEC
have been developed, but none are nearly so 2006).
widespread, or they are versions of the delta
method.
92
7.3 DRYLAN D CROP YIELDS applications of the delta method produce daily
climate unchanged in many respects from pres-
The effects of weather and climate on crops are ent-day observed data. The number of precipi-
complex. Despite the fact that many details of tation days and the time between them remains
weather interactions with plant physiology are the same. Also, relative changes in intensity are
poorly understood, numerous realistic crop- the same for light and heavy days. Likewise, the
growth simulation models have been developed. length of extended periods of extreme heat and
Current-generation crop models typically step cold and the intensity of such extremes with re-
through the growth process with daily fre- spect to the new climate mean do not change.
quency and use a number of meteorological
variables as input, typically maximum and min- In a recent study, Zhang (2005) used statistical
imum temperature, precipitation, solar radia- downscaling to estimate Oklahoma wheat yields
tion, and potential evapotranspiration. A key for a future simulation from HadCM3. In this
characteristic of these models is that they have study, mean monthly changes of the means and
been developed for application at a single loca- variances of temperature and precipitation be-
tion and have been validated based on point tween the HadCM3 control and future simula-
data, including meteorological inputs. Thus, tions were used to adjust the parameters of a
their use in assessing climate change impacts on weather generator model. Weather generator pa-
crop yields confronts a mismatch between the rameters include mean precipitation, precipita-
spatially averaged climate model grid-box data tion variance, the probability of a wet day
and the point data expected by crop models. following a wet day, the probability of a dry day
Also, biases in climate model data can have un- following a wet day, mean temperature, and
known effects on crop model results because the temperature variance. The observed data were
dependence of crop yields on meteorological used to determine a relationship between the
variables is highly nonlinear. The typical appli- wet-wet and wet-dry day probabilities and total
cation study circumvents these difficulties by monthly precipitation. This relationship was
avoiding the direct use of climate model output. used to assign future values of those probabili-
ties based on the GCM-simulated precipitation
The delta method continues to be a common ap- changes. With the new set of parameters, the
proach in contemporary crop studies. In the U.S. weather generator simulated multiple years of
National Assessment of the Consequences of daily weather variables for input to the yield
Climate Variability and Change, monthly model. This approach is logical and consistent
changes (model future – model control) were and produces different variability characteris-
applied to observed data, and a weather gener- tics depending on whether future climate is wet-
ator was used to produce daily weather data for ter or drier than the present, unlike the simple
input to impacts models. For example, Winkler delta method applied to daily climate data.
et al. (2002) found a longer growing season and However, these changes are assumed to be sim-
greater seasonal heat accumulation in fruit- ilar to what occurs in the present-day climate
growing regions of the Great Lakes but uncer- between wet and dry periods. Thus, more subtle
tainty about future susceptibility to freezes. climate model–simulated changes that might af-
Olesen et al. (2007) investigated the potential fect yields (e.g., a change to longer wet and dry
impacts of climate change on several European spells without a change in total precipitation)
crops. Crop models were driven by direct output are not transmitted.
of regional climate models and also baseline
(present-day) observed daily climate data ad- 7.4 SMALL W AT ERSH ED
justed by GCM changes using the delta method. FLOODIN G
Thomson et al. (2005) adjusted current daily cli-
mate data with monthly change values derived This application faces many of the same issues
from GCM projections (Smith et al. 2005) and as applying model output to estimate changes
then used them as input to models to study fu- in dryland crop yields. For example, models
ture yields of dryland crops in the United States. used for simulating runoff in small watersheds
National yield changes were found to be up to ± have been validated using point station data. In
25%, depending on the climate scenario. These addition, runoff is a highly nonlinear function
93
of precipitation, and flooding occurrence is par- 7.5 U RBAN H EAT W AVES

ticularly sensitive to the exact frequency and
amount of precipitation for the most extreme This estimation of changes in heat-wave fre-
events. As noted in the “Extreme Events” sec- quency and intensity can be accomplished using
tion of Chapter 5, climate models often under- only near-surface temperature. Because heat
estimate the magnitude of extremes. Again, the waves are large-scale phenomena and near-sur-
delta method is frequently applied to estimate face temperature is rather highly correlated over
the changes in flooding that may result from the scales of GCM grid-boxes, downscaling is
global climate change. Recently, Cameron not usually required for their analysis. Biases,
(2006) determined percentage changes in pre- while remaining an issue, can be accounted for
cipitation from climate model simulations and by using percentile-based definitions of heat
applied them to a stochastic rainfall model to waves. Meehl and Tebaldi (2004) used output
produce precipitation time series for input to a from the PCM for 2080 to 2099 to calculate per-
hydrologic model. Flood magnitudes were esti- centile-based measures of extreme heat; they
mated for return periods of 10 to 200 years and found that heat waves will increase in intensity,
for several climate changes scenarios. In most frequency, and duration. If mortality estimates
cases, flood flows increased, but one scenario are desired, then biases are an issue because ex-
produced a decrease. isting models (Kalkstein and Greene 1997) used
location-specific absolute magnitudes of tem-
Dibike and Coulibaly (2005) applied two statis- perature to estimate mortality.
tical downscaling techniques to an analysis of
flow on a small watershed in northern Quebec. 7.6 W AT ER RESOU RCES IN T H E
One technique used the model of Wilby, Daw- W EST ERN U N IT ED STAT ES
son, and Barrow (2002) to identify a set of
large-scale variables (i.e., pressure, flow, tem- The possibility that climate change may ad-
perature, and humidity) related to surface tem- versely affect limited water resources in the
perature and precipitation in the watershed. The mostly arid and semiarid western United States
resulting statistical relationships were applied poses a threat to the prosperity of that region. A
to the output of a Canadian GCM climate group of university and government scientists,
change simulation to generate future surface under the auspices of the U.S Department of En-
temperature and precipitation time series. The ergy–sponsored Accelerated Climate Prediction
second technique used a weather generator re- Initiative Pilot Project, conducted a coordinated
quiring various statistical parameters, estimated set of studies that represented an end-to-end as-
by comparing surface temperature and precipi- sessment of this issue (Barnett et al. 2004). This
tation data between GCM control and future project is noteworthy because of close coordi-
scenario simulations. The fundamental differ- nation between production of GCM simulations
ence between these two statistical downscaling and the needs of impacts modeling. It also is a
techniques is that the Wilby, Dawson, and Bar- good example of more-sophisticated downscal-
row (2002) model uses a more complete set of ing approaches.
atmospheric data from the GCM output data
while the weather generator uses only surface A suite of carefully selected PCM climate sim-
temperature and precipitation. The resulting ulations was executed (Dai et al. 2004; Pierce
time series from both methods provided input 2004) and then used to drive a regional climate
for a hydrologic model. In both cases, peak model to provide higher-resolution data (Leung
flows are higher in the spring and lower in the et al. 2004), both for direct assessment of effects
early summer in future warmer climates, re- on water resources and for use in impacts mod-
flecting changes in snowmelt timing. A major els. A careful statistical downscaling approach
difference is that the Wilby, Dawson, and Bar- (Wood et al. 2004) also was used to produce an
row (2002) model produces a trend of increas- alternate dataset for input to impacts models.
ing daily precipitation not seen in the weather Using the observationally based 1/8° latitude-
generator data, resulting in larger spring in- by-longitude resolution gridded dataset devel-
creases in peak flow. oped by Maurer et al. (2002), an empirical
mapping function was developed to relate quan-
94
tiles of the simulated monthly temperature and

precipitation frequency distributions from con-
trol runs to the observed climatological monthly
distributions at the GCM grid scale. This em-
pirical mapping was then applied to simulated
future monthly temperature and precipitation
data and spatially disaggregated to the 1/8° res-
olution grid through a procedure that added
small-scale structure. Daily time series of future
climate on the 1/8° grid subsequently were pro-
duced by randomly sampling from historical
data and adding in the changes resulting from
the empirical mapping and disaggregation.
The daily time series were used in a set of stud-

ies to assess water resource impacts (Stewart,
Cayan, and Dettinger 2004; Payne et al. 2004;
VanRheenen et al. 2004; Christensen et al.
2004). The studies, which assumed the IPCC
business-as-usual emissions scenario for the cli-
mate change GCM simulation, indicate that
warmer temperatures will melt the snowpack
about a month earlier throughout western North
America by the end of the 21st Century. The
shift in snowmelt will decrease flows and in-
crease competition for water during the summer
in the Columbia River Basin (Payne et al. 2004).
In the Sacramento River and San Joaquin River
basins, the average April 1 snowpack is pro-
jected to decrease by half. In the Colorado River
basin, a decrease in total precipitation would
mean that total system demand would exceed
river inflows.
95
96
REFEREN C ES
AchutaRao, K.M., et al., 2006: Variability of ocean heat uptake: Anderson, T.L., et al., 2003: Climate forcing by aerosols—a hazy
Reconciling observations and models. J. Geophysical Re- picture. Science, 300, 1103–1104.
search, 111, doi:10.1029/2005JC003136.
Andersson, L., et al., 2006: Impact of climate change and devel-
AchutaRao, K.M., and K.R. Sperber, 2006: ENSO simulation in opment scenarios on flow patterns in the Okavango River. J.
coupled ocean-atmosphere models: Are the current models Hydrology, 331, 43–57.
better? Climate Dynamics, doi:10.1007/s00382-006-0119-7.
Annamalai, H., K. Hamilton, and K.R. Sperber, 2007: South
AchutaRao, K., and K.R. Sperber, 2002: Simulation of the El Asian summer monsoon and its relationship with ENSO in
Niño Southern Oscillation: Results from the Coupled Model the IPCC AR4 simulations. J. Climate, 20(6), 1071–1092.
Intercomparison Project. Climate Dynamics, 19, 191–209.
Annan, J.D., and J.C. Hargreaves, 2006: Using multiple observa-
Ackerman, A.S., et al., 2004: The impact of humidity above strat- tionally based constraints to estimate climate sensitivity. Geo-
iform clouds on indirect aerosol climate forcing. Nature, 432, physical Research Letters, 33(6), doi:10.1029/2005GL025259.
1014–1017.
Annan, J.D., et al., 2005: Efficiently constraining climate sensi-
Ackerman, T. P., and G.M. Stokes, 2003: The Atmospheric Ra- tivity with paleoclimate simulations. SOLA, 1, 181–184.
diation Measurement Program. Physics Today, 56(1), 38–44.
Antic, S., et al., 2006: Testing the downscaling ability of a one-way
Adams, R.M., B.A. McCarl, and O. Mearns, 2003: The effects of nested regional climate model in regions of complex topog-
spatial scale climate scenarios on economic assessments: An raphy. Climate Dynamics, 26, 305–325.
example from U.S. agriculture. Climate Change, 60, 131–148.
Antic, S., et al., 2004: Testing the downscaling ability of a one-way
Albrecht, B.A., 1989: Aerosols, cloud microphysics, and frac- nested regional climate model in regions of complex topog-
tional cloudiness. Science, 245, 1227–1230. raphy. Climate Dynamics, 23(5), 473–493.
Alexander, M., et al., 2006: Extratropical atmosphere-ocean vari- Antonov, J.I., S. Levitus, and T.P. Boyer, 2005: Thermosteric sea level
ability in CCSM3. J. Climate, 19, 2496–2525. rise, 1955–2003. Geophysical Research Letters, 32, L12602.
Alley, R.B., et al., 2005: Ice-sheet and sea-level changes. Science, Arakawa, A., and W.H. Schubert, 1974: Interaction of a cumulus
310, 456–460. cloud ensemble with the large-scale environment. Part I. J.
Atmospheric Sciences, 31, 674–701.
Ammann, C.M., et al., 2003: A monthly and latitudinally varying
volcanic forcing dataset in simulations of 20th Century cli- Arblaster, J.M., and G.A. Meehl, 2006: Contribution of various
mate. Geophysical Research Letters, 30(12), 1657. external forcings to trends in the Southern Annular Mode. J.
Climate, 19, 2896–2905.
Anderson, B.T., et al., 2001: Model dynamics of summertime
low-level jets over northwestern Mexico. J. Geophysical Re- Arendt, A.A., et al., 2002: Rapid wastage of Alaska glaciers and
search, 106(D4), 3401–3413. their contribution to rising sea level. Science, 297, 382–386.
Anderson, C.J., R.W. Arritt, and J.S. Kain, 2007: An alternative Arora, V.K., and G.J. Boer, 2003: A representation of variable root dis-
mass flux profile in the Kain-Fritsch convective parameteri- tribution in dynamic vegetation models. Earth Interactions, 7, 1–19.
zation and its effects in seasonal precipitation. J. Hydromete-
orology, 8(5), 1128–1140, doi:10.1175/JHM624.1. Arzel, O., T. Fichefet, and H. Goosse, 2006: Sea ice evolution over
the 20th and 21st centuries as simulated by current AOGCMs.
Anderson, C.J., et al., 2003: Hydrologic processes in regional cli- Ocean Modelling, 12, 401–415.
mate model simulations of the central United States flood of
June–July 1993. J. Hydrometeorology, 4, 584–598. Aurela, M., T. Laurila, and J.P. Tuovinen, 2004: The timing of
snow melt controls the annual CO2 balance in a subarctic fen.
Geophysical Research Letters, 31, L16119.
97
The U.S. Climate Change Science Program References
Avissar, R., and R.A. Pielke, 1989: A parameterization of heterogeneous Bitz, C.M., and W.H. Lipscomb, 1999: An energy-conserving thermo-
land-surface for atmospheric numerical models and its impact on re- dynamic model of sea ice. J. Geophysical Research, 104, 15669–
gional meteorology. Monthly Weather Review, 117, 2113–2136. 12677.
Baldocchi, D., and P. Harley, 1995: Scaling carbon dioxide and water Bjerknes, J., 1969: Atmospheric teleconnections from the equatorial Pa-
vapour exchange from leaf to canopy in a deciduous forest. I. Leaf cific. Monthly Weather Review, 97, 163–172.
model parameterization. Plant, Cell, Environment, 18, 1146–1156.
Bleck, R., 2002: An oceanic general circulation model framed in hybrid
Bard, E., B. Hamelin, and R.G. Fairbanks, 1990: U-Th ages obtained by isopycnic-Cartesian coordinates. Ocean Modelling, 4, 55–88.
mass spectrometry in corals from Barbados: Sea level during the past
130,000 years. Nature, 346, 456–458. Bleck, R., et al., 1992: Salinity-driven thermocline transients in a wind-
and thermohaline-forced isopycnic coordinate model of the North
Barnett, T.P., et al., 2005: Penetration of human-induced warming into Atlantic. J. Physical Oceanography, 22(12), 1485–1505.
the world’s oceans. Science, 309, 284.
Boer, G., J. Stowasser, and K. Hamilton, 2007: Inferring climate sensi-
Barnett, T., et al., 2004: The effects of climate change on water resources tivity from volcanic events. Climate Dynamics, 28(5), 481–502.
in the West: Introduction and overview. Climatic Change, 62, 1–11.
Boisserie, M., et al., 2006: Evaluation of soil moisture in the Florida
Beckman, A., and R. Doscher, 1997: A method for improved represen- State University climate model: National Center for Atmospheric
tation of dense water spreading over topography in geopotential-co- Research community land model (FSU-CLM) using two reanalyses
ordinate models. J. Physical Oceanography, 27, 581–591. (R2 and ERA40) and in situ observations. J. Geophysical Research,
111(D8), Art. No. D08103.
Bell, J.L., L.C. Sloan, and M.A. Snyder, 2004: Regional changes in ex-
treme climatic events: A future climate scenario. J. Climate, 17(1), Bonan, G. B., and S. Levis, 2006: Evaluating aspects of the community
81–87. land and atmosphere models (CLM3 and CAM3) using a dynamic
global vegetation model. J. Climate, 19, 2290–2301.
Bellouin, N., et al., 2005: Global estimate of aerosol direct radiative forc-
ing from satellite measurements. Nature, 438, 1138–1141. Bonan, G.B., 1995: Sensitivity of a GCM simulation to inclusion of in-
land water surfaces. J. Climate, 8, 2691–2704.
Bengtsson, L., K.I. Hodges, and M. Esch, 2007: Tropical cyclones in a
T159 resolution global climate model: Comparison with observa- Bonan, G.B., D. Pollard, and S.L. Thompson, 1992: Effects of boreal for-
tions and re-analyses. Tellus A, 59(4), 396–416, doi:10.1111/j.1600- est vegetation on global climate. Nature, 359, 716–718.
0870.2007.00236.x.
Bony, S., et al., 2006: How well do we understand and evaluate climate
Beringer, J., et al., 2001: The representation of arctic soils in the land change feedback processes? J. Climate, 19(15), 3445–3482.
surface model: The importance of mosses. J. Climate, 14, 3324–
3335. Bony, S., and J.-L. Dufresne, 2005: Marine boundary layer clouds at the
heart of tropical cloud feedback uncertainties in climate models.
Betts, A.K., and J.H. Ball, 1997: Albedo over the boreal forest. J. Geo- Geophysical Research Letters, 32, L20806.
physical Research–Atmosphere, 102, 28901–28909.
Bony, S., et al., 2004: On dynamic and thermodynamic components of
Biasutti, M., and A. Giannini, 2006: Robust Sahel drying in response to cloud changes. Climate Dynamics, 22, 71–86.
th
late 20 Century forcings. Geophysical Research Letters, 33,
L11706. Bony, S., and K.A. Emanuel, 2001: A parameterization of the cloudiness
associated with cumulus convection: Evaluation using TOGA
Bierbaum, R.M., et al., 2003: Estimating Climate Sensitivity: Report of COARE data. J. Atmospheric Sciences, 58, 3158–3183.
a Workshop. National Academy of Sciences, Washington, D.C.
(books.nap.edu/catalog/10787.html). Boone, A., et al., 2000: The influence of the inclusion of soil freezing on
simulations by a soil-vegetation-atmosphere transfer scheme. J. Ap-
Bishop, J.K.B., R.E. Davis, and J.T. Sherman, 2002: Robotic observa- plied Meteorology, 39, 1544–1569.
tions of dust storm enhancement of carbon biomass in the North Pa-
cific. Science, 298, 817–821. Boyle, J.S., et al., 2008: Climate model forecast experiments for TOGA-
COARE. Monthly Weather Review, in press.
98
Boyle, J.S., 1993: Sensitivity of dynamical quantities to horizontal reso- Bryden, H.L., and S. Imawaki, 2001: Ocean heat transport, 455–474 in
lution for a climate simulation using the ECMWF (cycle 33) model. Siedler, G., J. Church, and J. Gould, eds. Ocean Circulation and Cli-
J. Climate, 6, 796–815. mate: Observing and Modelling the Global Ocean, Academic Press,
San Diego, p.109.
Bougeault, P., et al., 2001: The MAP special observing period. Bulletin
American Meteorological Society, 82(3), 433–462. Burke, E.J., S.J. Brown, and N. Christidis, 2006: Modeling the recent evo-
lution of global drought and projections for the 21st Century with the
Braconnot, P., et al., 2007a: Results of PMIP2 coupled simulations of the Hadley Centre climate model. J. Hydrometeorology, 7, 1113–1125.
mid-Holocene and Last Glacial Maximum. Part 1: Experiments and
large-scale features. Climate Past, 3(2), 261–277. Byerle, L.A., and J. Paegle, 2003: Modulation of the Great Plains low-
level jet and moisture transports by orography and large-scale circu-
Braconnot, P., et al., 2007b: Results of PMIP2 coupled simulations of the lations. J. Geophysical Research, 108(D16), Art. No. 8611.
Mid-Holocene and Last Glacial Maximum. Part 2: Feedbacks with
emphasis on the location of the ITCZ and mid- and high-latitudes Cai, W.J., P. Whetton, and D.J. Karoly, 2003: The response of the Antarc-
heat budget. Climate Past 3(2), 279–296. tic Oscillation to increasing and stabilized atmospheric CO2. J. Cli-
mate, 16, 1525–1538.
Braithwaite, R.J., and S.C.B. Raper, 2002: Glaciers and their contribu-
tion to sea level change. Physics Chemistry Earth, 27, 1445–1454. California Energy Commission (CEC), 2006: Our Changing Climate:
Assessing the Risks to California, A Summary Report from the Cal-
Brankovic, T., and D. Gregory, 2001: Impact of horizontal resolution on ifornia Climate Change Center, 16 pp. (www.climatechange.ca.gov/
seasonal integrations. Climate Dynamics, 18, 123–143. biennial_reports/2006report/index.html).
Breugem, W.-P., W. Hazeleger, and R.J. Haarsma, 2006: Multi-model Callaghan, T.V., et al., 2004: Responses to projected changes in climate
study of tropical Atlantic variability and change. Geophysical Re- and UV-B at the species level. Ambio, 33, 418–435.
search Letters, 33, L23706.
Cameron, D., 2006: An application of the UKCIP02 climate change sce-
Briegleb, B.P., et al., 2002: Description of the Community Climate narios to flood estimation by continuous simulation for a gauged
System Model Version 2 Sea Ice Model, 60 pp. catchment in the northeast of Scotland, U.K. (with uncertainty). J.
(www.ccsm.ucar.edu/models/ccsm2.0/csim/). Hydrology, 328, 212–226.
Brinkop, S., and E. Roeckner, 1995: Sensitivity of a general circulation Camp, C.D., and K.K. Tung, 2007: Surface warming by the solar cycle
model to parameterizations of cloud-turbulence interactions in the as revealed by the composite mean difference projection. Geophys-
atmospheric boundary layer. Tellus, 47A, 197–220. ical Research Letters, 34, L14703, doi:10.1029/2007GL030207.
Brown, T.J., B.L. Hall, and A.L. Westerling, 2004: The impact of Twenty- Cane, M.A., et al., 1997: Twentieth Century sea surface temperature
First Century climate change on wildland fire danger in the western trends. Science, 275, 957–960.
United States: An applications perspective. Climatic Change, 62,
365–388. Capotondi, A., A. Wittenberg, and S. Masina, 2006: Spatial and tempo-
ral structure of ENSO in 20th Century coupled simulations. Ocean
Bryan, F.O., et al., 2006: Changes in ocean ventilation during the 21 st Modelling, 15, 274–298.
Century in the CCSM3. Ocean Modelling, 15, 141–156.
Carril, A.F., C.G. Menéndez, and A. Navarra, 2005: Climate response as-
Bryan, K., 1969a: A numerical method for the study of the circulation of sociated with the Southern Annular Mode in the surroundings of
the world ocean. J. Computational Physics, 4, 347–376. Antarctic Peninsula: A multimodel ensemble analysis. Geophysical
Research Letters, 32, L16713.
Bryan, K., 1969b: Climate and the ocean circulation. III: The ocean
model. Monthly Weather Review, 97, 806–824. Cassano, J.J., et al., 2007: Predicted changes in synoptic forcing of net
precipitation in large Arctic river basins during the 21st Century. J.
Bryan, K., and M.D. Cox, 1967: A numerical investigation of the oceanic Geophysical Research, 112, G04S49, doi:10.1029/2006JG000332.
general circulation. Tellus, 19, 54–80.
Cassano, J.J., P. Uotilla, and A.H. Lynch, 2006: Changes in synoptic
weather patterns in the polar regions in the 20th and 21st centuries.
Part 1: Arctic. International J. Climatology, 26(9), 1181–1199.
99
Cayan, D., et al., 2006: Climate Scenarios from California: A Report Christensen, J.H., et al., 2007: Regional Climate Projections in Climate
from California Climate Change Center, White Paper, 52 pp. Change 2007: The Physical Science Basis. Contribution of Working
(www.energy.ca.gov/2005publications/CEC-500-2005-203/CEC- Group 1 to the Fourth Assessment Report of the Intergovernmental
500-2005-203.SF.PDF). panel on Climate Change, ed. S. Solomon et al. Cambridge Univer-
sity Press, Cambridge, U.K., and New York, N.Y.
Cazenave, A., and R.S. Nerem, 2004: Present-day sea level change: Ob-
servations and causes. Reviews Geophysics, 42(3), RG3001. Christensen, J.H., T. Carter, and F. Giorgi, 2002: PRUDENCE employs
new methods to assess European climate change. Eos, 83, 147.
CCSP, 2007: The First State of the Carbon Cycle Report (SOCCR): The
North American Carbon Budget and Implications for the Global Christensen, N.S., et al., 2004: The effects of climate change on the hy-
Carbon Cycle, ed. A.W. King, et al. Synthesis and Assessment Prod- drology and water resources of the Colorado River basin. Climatic
uct 2.2, U.S. Climate Change Science Program and Subcommittee on Change, 62, 337–363.
Global Change Research, Washington, D.C. (www.climate-
science.gov/Library/sap/sap2-2/final-report/default.htm). Christensen, O.B., 1999: Relaxation of soil variables in a regional cli-
mate model. Tellus, 51A, 674–685.
CCSP, 2006: Temperature Trends in the Lower Atmosphere: Steps for
Understanding and Reconciling Differences, ed. T.R. Karl et al. Syn- Christidis, N., et al., 2005: Detection of changes in temperature extremes
thesis and Assessment Product 1.1, U.S. Climate Change Science during the second half of the 20th Century. Geophysical Research
Program and Subcommittee on Global Change Research, Washing- Letters, 32, L20716.
ton, D.C. (www.climatescience.gov/Library/sap/sap1-1/finalre-
port/default.htm). Christy, J.R., et al., 2007: Tropospheric temperature change since 1979
from tropical radiosonde and satellite measurements. J. Geophysi-
Cess, R.D., et al., 1990: Intercomparison and interpretation of climate cal Research, 112, D06102, doi:10.1029/2005JD006881.
feedback processes in 19 atmospheric general circulation models. J.
Geophysical Research, 95(D10), 16601–16615. Christy, J. R., and R. W. Spencer, 2005: Correcting temperature data
sets. Science, 310, 972.
Cess, R.D., and G.L. Potter, 1988: Exploratory studies of cloud radiative
forcing with a general circulation model. Tellus, 39A, 460–473. Chung, C.E., et al., 2005: Global anthropogenic aerosol direct forcing de-
rived from satellite and ground-based observations. J. Geophysical
Charney, J.G., 1979: Carbon Dioxide and Climate: A Scientific Assess- Research, 110, D24207.
ment. National Academy of Sciences, Washington, D.C., 22 pp.
Church, J.A., and N.J. White, 2006: A 20th Century acceleration in global
Chen, F., and J. Dudhia, 2001: Coupling an advanced land surface–hy- sea-level rise. Geophysical Research Letters, 33, L01602.
drology model with the Penn State–NCAR MM5 modeling system.
Part I: Model implementation and sensitivity. Monthly Weather Re- Church, J.A., N.J. White, and M. Arblaster, 2005: Significant decadal-
view, 129, 569–585. scale impact of volcanic eruptions on sea level and ocean heat con-
tent. Nature, 438(7064), 74–77, doi:10.1038/nature04237.
Cheng, A., and K.-M. Xu, 2006: Simulation of shallow cumuli and their
transition to deep convective clouds by cloud-resolving models with Claussen, M., et al., 2002: Earth system models of intermediate com-
different third-order turbulence closures. Quarterly J. Royal Mete- plexity: Closing the gap in the spectrum of climate system models.
orological Society, 132, 359–382. Climate Dynamics, 18, 579–586.
Cheng, Y., V.M. Canuto, and A.M. Howard, 2002: An improved model for Clement, A.C., and B. Soden, 2005: The sensitivity of the tropical-mean
the turbulent BBL. J. Atmospheric Sciences, 59, 1550–1565. radiation budget. J. Climate, 18, 3189–3203.
Chiacchio, M., J. Francis, and P. Stackhouse, Jr., 2002: Evaluation of Clement, A.C., et al., 1996: An ocean dynamical thermostat. J. Climate,
methods to estimate the surface downwelling longwave flux during 9, 2190–2196.
Arctic winter. J. Applied Meteorology, 41, 306–318.
Clough, S.A., M.J. Iacono, and J.-L. Moncet, 1992: Line-by-line calcu-
Choi, H.I., P. Kumar, and X.-Z. Liang, 2007: Three-dimensional aver- lations of atmospheric fluxes and cooling rates: Application to water
aged soil-moisture transport model with a scalable parameterization vapor. J. Geophysical Research, 97, 15761–15785.
of subgrid topographic variability. Water Resources Research, 43,
WO4414.
100
Collins, M., et al., 2006: Towards quantifying uncertainty in transient Crucifix, M., 2006: Does the last glacial maximum constrain climate
climate change. Climate Dynamics, 27(2–3), 127–147. sensitivity? Geophysical Research Letters, 33, L18701,
doi:10.1029/2006GL027137.
Collins, W.D., et al., 2006(a): The Community Climate System Model
Version 3: CCSM3. J. Climate, 19(11), 2122–2143. Crucifix, M., et al., 2006: Second phase of Paleoclimate Modelling In-
tercomparison Project. EOS, Transactions American Geophysical
Collins, W.D., et al., 2006(b): Radiative forcing by well-mixed green- Union, 86(28), 264.
house gases: Estimates from climate models in the Intergovernmen-
tal Panel on Climate Change (IPCC) Fourth Assessment Report Cubasch, U., et al., 2001: Projections of future climate change, 525–582
(AR4). J. Geophysical Research, 111, D14317. in Climate Change 2001: The Scientific Basis, ed. J.T. Houghton, et
al. Cambridge University Press, Cambridge, U.K.
Collins, W.D., et al., 2004: Description of the NCAR Community At-
mospheric Model (CAM3). NCAR Technical Note, National Center Cunningham, S.A., et al., 2003: Transport and variability of the Antarc-
for Atmospheric Research, NCAR/TN-464+STR, 226 pp. tic Circumpolar Current in Drake Passage. J. Geophysical Research,
108, doi:10.1029/2001JC001147.
Colman, R., 2003: A comparison: Climate feedbacks in general circula-
tion models. Climate Dynamics, 20, 865–873. Curry, J.A., and A.H. Lynch, 2002: Comparing Arctic regional climate
models. Eos, 83, 87.
Cook, K.H., and E.K. Vizy, 2006: Coupled model simulations of the
West African monsoon system: 20th Century simulations and 21st Curry, J.A., et al., 1996: Overview of Arctic cloud and radiation char-
Century predictions. J. Climate, 19, 3681–3703. acteristics. J. Climate, 9, 1731–1764.
Coppola, E., and F. Giorgi, 2005: Climate change in tropical regions Curry, J.A., J. Schramm, and E.E. Ebert, 1995: Sea ice-albedo climate
from high-resolution time-slice AGCM experiments. Quarterly J. feedback mechanism. J. Climate, 8, 240–247.
Royal Meteorological Society, 131, 3123–3145.
Cusack, S., J.M. Edwards, and R. Kershaw, 1999: Estimating subgrid
Cotton, W.R., 2003: Cloud modeling from days of EML to the present— variance of saturation and its parameterization for use in a GCM
Have we made progress? 95–106 in AMS Meteorological Mono- cloud scheme. Quarterly J. Royal Meteorological Society, 125,
graphs—Symposium on Cloud Systems, Hurricanes, and TRMM. 3057–3076.
Covey, C., et al., 2003: An overview of results from the Coupled Model Dai, A., 2006: Precipitation characteristics in eighteen coupled climate
Intercomparison Project, Global Planetary Change, 37(1–2), 103– models. J. Climate, 19, 4605–4630.
133.
Dai, A., et al., 2004: The ACPI climate change simulations. Climatic
Covey, C., et al., 2000: The seasonal cycle in coupled ocean-atmosphere Change, 62, 29–43.
general circulation models. Climate Dynamics, 16, 775–787.
Danabasoglu, G., et al., 2006: Diurnal coupling in the tropical oceans of
Covey, C., L.C. Sloan, and M.I. Hoffert, 1996: Paleoclimate data con- CCSM3. J. Climate, 19, 2347–2365, doi:10.1175/JCLI3739.1.
straints on climate sensitivity: The paleocalibration method. Climatic
Change, 32, 165–184. Davies, H.C., and R.E. Turner, 1977: Updating prediction models by dy-
namical relaxation: An examination of the technique. International
Cox, P.M., et al., 2000: Acceleration of global warming due to carbon- J. Climatology, 103, 225–245.
cycle feedbacks in a coupled model. Nature, 408, 184–187.
Davies, H.C., 1976: Lateral boundary formulation for multilevel predic-
Cramer, W., et al., 2001: Global response of terrestrial ecosystem and tion models. Quarterly J. Royal Meteorological Society, 102, 405–
function to CO2 and climate change: Results from six dynamic 418.
global vegetation models. Global Change Biology, 7, 357–373.
DeCaria, A.J., et al., 2005: Lightning-generated NOx and its impact on
Crane, R.G., and B.C. Hewitson, 1998: Doubled CO2 precipitation tropospheric ozone production: A three-dimensional modeling study
changes for the Susquehanna basin: Down-scaling from the GENE- of a stratosphere–troposphere experiment: Radiation, aerosols, and
SIS general circulation model. International J. Climatology, 18, 65– ozone (STERAO-A) thunderstorm. J. Geophysical Research, 110,
76. D14303, doi:10.1029/2004JD00556.
101
de Elía, R., R. Laprise, and B. Denis, 2002: Forcasting skill limits of Dickinson, R.E., A. Henderson-Sellers, and P.J. Kennedy, 1993: Bios-
nested, limited-area models: A perfect-model approach. Monthly phere-Atmosphere Transfer Scheme (BATS) version 1e as coupled
Weather Review, 130, 2006–2023. to the NCAR Community Climate Model. NCAR Technical Note,
NCAR/TN-387+STR, National Center for Atmospheric Research,
Del Genio, A.D., et al., 2005: Cumulus microphysics and climate sensi- Boulder, Colo., 72 pp. [Available from NCAR; P.O. Box 3000; Boul-
tivity. J. Climate, 18, 2376–2387, doi:10.1175/JCLl3413.1. der, CO 80305]
Del Genio, A.D., A. Wolf, and M.-S. Yao, 2005: Evaluation of regional Diffenbaugh, N.S., M.A. Snyder, and L.C. Sloan, 2004: Could CO2-in-
cloud feedbacks using single-column models. J. Geophysical Re- duced land-cover feedbacks alter near-shore upwelling regimes? Pro-
search, 110, D15S13, doi:10.1029/2004JD005011. ceedings National Academy Sciences, 101, 27–32.
Del Genio, A.D., et al., 1996: A prognostic cloud water parameterization Dimitrijevic, M., and R. Laprise, 2005: Validation of the nesting tech-
for global climate models. J. Climate, 9, 270–304. nique in a regional climate model and sensitivity tests to the resolu-
tion of the lateral boundary conditions during summer. Climate
Del Genio, A.D., and M.-S. Yao, 1993: Efficient cumulus parameteriza- Dynamics, 25, 555–580.
tion for long-term climate studies: The GISS scheme. The Repre-
sentation of Cumulus Convection in Numeric Models. Dimotakis, P.E., 2005: Turbulent mixing. Annual Review Fluid Me-
Meteorological Monograph, No. 46, American Meteorological So- chanics, 37, 329–356.
ciety, 181–184.
Doney, S.C., et al., 2004: Ocean Carbon and Climate Change (OCCC): An
Delworth, T.L., et al., 2006: GFDL’s CM2 global coupled climate mod- Implementation Strategy for U.S. Ocean Carbon Cycle Science. Uni-
els. Part 1: Formulation and simulation characteristics. J. Climate, versity Corporation for Atmospheric Research, Boulder, Colo., 108 pp.
19, 643–674.
Douglass, D.H., and R.S. Knox, 2005: Climate forcing by the volcanic
Denis, B., R. Laprise, and D. Caya, 2003: Sensitivity of a regional climate eruption of Mount Pinatubo. Geophysical Research Letters, 32,
model to the resolution of the lateral boundary conditions. Climate L05710, doi:10.1029/2004GL022119.
Dynamics, 20, 107–126.
Ducharne, A., et al., 2003: Development of a high resolution runoff rout-
Denis, B., et al., 2002: Downscaling ability of one-way, nested regional ing model, calibration, and application to assess runoff from the
climate models: The big-brother experiments. Climate Dynamics, LMD GCM. J. Hydrology, 280, 207–228.
18, 627–646.
Duffy, P.B., et al., 2003: High-resolution simulations of global climate.
Déqué, M., et al., 2005: Global high resolution versus Limited Area Part 1: Present climate. Climate Dynamics, 21, 371–390.
Model climate change projections over Europe: Quantifying confi-
dence level from PRUDENCE results. Climate Dynamics, 25, 653– Dutay, J.-C., et al., 2002: Evaluation of ocean model ventilation with
670. CFC-11: Comparison of 13 global ocean models. Ocean Modelling,
4, 89–120.
Déqué, M., and J.P. Piedelievre, 1995: High-resolution climate simula-
tion over Europe. Climate Dynamics, 11, 321–339. Duynkerke, P.G., and S.R. de Roode, 2001: Surface energy balance and
turbulence characteristics observed at the SHEBA ice camp during
Dethloff, K., et al., 2006: A dynamical link between the Arctic and the FIRE III. J. Geophysical Research, 106, 15313–15322.
global climate system. Geophysical Research Letters, 33, L03703,
doi:10.1029/2005GL025245. Easterling, D.R., 2002: Recent changes in frost days and the frost-free
season in the United States. Bulletin American Meteorological Soci-
Dettinger, M.D., et al., 2004: Simulated hydrologic responses to climate ety, 83, 1327–1332.
variations and change in the Merced, Carson, and American River
basins, Sierra Nevada, California, 1900–2099. Climatic Change, 62, Ebert, E.E., J.L. Schramm, and J.A. Curry, 1995: Disposition of solar
283–317. radiation in sea ice and the upper ocean. J. Geophysical Research,
100, 15965–15975.
Dibike, Y.B., and P. Coulibaly, 2005: Hydrologic impact of climate
change in the Saguenay watershed: Comparison of downscaling Ebert, E.E., and J.A. Curry, 1993: An intermediate one-dimensional ther-
methods and hydrologic models. J. Hydrology, 307, 145–163. modynamic sea ice model for investigating ice-atmosphere interac-
tions. J. Geophysical Research, 98, 10085–10109.
102
Ellingson, R., and Y. Fouquart, 1991: The intercomparison of radiation Fox-Rabinovitz, M.S., L.L. Takacs, and R.C. Govindaraju, 2002: A vari-
codes in climate models: An overview. J. Geophysical Research, able-resolution stretched-grid general circulation model and data as-
96(D5), 8925–8927. similation system with multiple areas of interest: Studying the
anomalous regional climate events of 1998. J. Geophysical Research,
Emanuel, K.A., 1994: Atmospheric Convection. Oxford University 107(D24), Art. No. 4768.
Press, Oxford, U.K., 580 pp.
Fox-Rabinovitz, M.S., et al., 2001: A variable-resolution stretched-grid
Emanuel, K.A., 1991: A scheme for representing cumulus convection general circulation model: Regional climate simulation. Monthly
in large-scale models. J. Atmospheric Sciences, 48, 2313–2335. Weather Review, 129, 453–469.
Emori, S., and S.J. Brown, 2005: Dynamic and thermodynamic changes Fox-Rabinovitz, M.S., and R.S. Lindzen, 1993: Numerical experiments
in mean and extreme precipitation under changed climate. Geo- on consistent horizontal and vertical resolution for atmospheric mod-
physical Research Letters, 32, L17706. els and observing systems. Monthly Weather Review, 121, 264–271.
Emori, S., et al., 2005: Validation, parameterization dependence, and fu- Frame, D.J., et al., 2005: Constraining climate forecasts: The role of prior
ture projection of daily precipitation simulated with a high-resolution assumptions. Geophysical Research Letters, 32, L09702,
atmospheric GCM. Geophysical Research Letters, 32(6), L06708. doi:10.1029/2004GL022241.
Enfield, D.B., A.M. Mestas-Nuñez, and P.J. Trimble, 2001: The Atlantic Frei, C., et al., 2006: Future change of precipitation extremes in Europe:
Multidecadal Oscillation and its relationship to rainfall and river An intercomparison of scenarios from regional climate models. J.
flows in the continental U.S., Geophysical Research Letters, 28, Geophysical Research, 111(D6), Art. No. D06105.
2077–2080.
Frich, P., et al., 2002: Observed coherent changes in climatic extremes
Essery, R., and J. Pomeroy, 2004: Implications of spatial distributions of during the second half of the Twentieth Century. Climate Research,
snow mass and melt rate for snow-cover depletion: Theoretical con- 19, 193–212.
siderations. Annals Glaciology, 38, 261–265.
Fridlind, A.M., et al., 2004: Evidence for the predominance of mid-tro-
Fedorov, A., and S.G. Philander, 2000: Is El Niño changing? Science, pospheric aerosols as subtropical anvil cloud nuclei. Science,
288, 1997–2002. 304(5671), 718–722.
Field, C., R. Jackson, and H. Mooney, 1995: Stomatal responses to in- Friedlingstein, P., et al., 2006: Climate-carbon cycle feedback analysis:
creased CO2: Implications from the plant to the global scale. Plant, Results from the C4MIP model intercomparison. J. Climate, 19,
Cell, Environment, 18, 1214–1225. 3337–3353.
Forest, C.E., P.H. Stone, and A.P. Sokolov, 2006: Estimated PDFs of cli- Friedlingstein, P., et al., 2001: Positive feedback between future climate
mate system properties including natural and anthropogenic forc- change and the carbon cycle. Geophysical Research Letters, 28(8),
ings. Geophysical Research Letters, 33, L01705, 1543–1546.
doi:10.1029/2005GL023977.
Fröhlich, C., 2006: Solar irradiance variability since 1978. Space Science
Forster, P., et al., 2007: Changes in Atmospheric Constituents and in Ra- Reviews, 125, 53–65, doi:10.1007/s11214-006-9046-5.
diative Forcing. In Climate Change 2007: The Physical Science
Basis. Contribution of Working Group 1 to the Fourth Assessment of Fu, C.B., et al., 2005: Regional climate model intercomparison project
the Intergovernmental Panel on Climate Change, ed. S. Solomon et for Asia. Bulletin American Meteorological Society, 86, 257–266.
al. Cambridge University Press, Cambridge, U.K., and New York,
N.Y. Fyfe, J.C., G.J. Boer, and G.M. Flato, 1999: The Arctic and Antarctic os-
cillations and their projected changes under global warming. Geo-
Foukal, P., et al., 2006: Variations in solar luminosity and their effect on physical Research Letters, 11, 1601–1604.
the Earth’s climate. Nature, 443, 161–166.
Ganachaud, A., and C. Wunsch, 2000: Improved estimates of global
Fox-Rabinovitz, M.S., et al., 2006: Variable resolution general circula- ocean circulation, heat transport and mixing from hydrographic data.
tion models: Stretched-grid model intercomparison project Nature, 408, 453–457.
(SGMIP). J. Geophysical Research, 111(D16), Art. No. D16104.
103
Gao, S., L. Ran, and X. Li, 2006: Impacts of ice microphysics on rain- Giorgi, F., et al., 1996: A regional model study of the importance of local
fall and thermodynamic processes in the tropical deep convective versus remote controls of the 1988 drought and the 1993 flood over
regime: A 2D cloud-resolving modeling study. Monthly Weather Re- the central United States. J. Climate, 9, 1150–1162.
view, 134(10), 3015–3024.
Giorgi, F., M.R. Marinucci, and G.T. Bates, 1993: Development of a sec-
GARP, 1975: The Physical Basis of Climate Modeling, Global Atmos- ond-generation regional climate model (RegCM2). Part II: Convec-
pheric Research Program (GARP), Publication Series #16, April tive processes and assimilation of lateral boundary conditions.
1975. Monthly Weather Review, 121, 2814–2832.
Gates, W.L., et al., 1999: An Overview of the Atmospheric Model Inter- Giorgi, F., and L.O. Mearns, 1991: Approaches to the simulation of re-
comparison Project (AMIP). Bulletin American Meteorological So- gional climate change—A review. Reviews Geophysics, 29, 191–
ciety, 80(1), 29–55, doi:10.1175/1520-0477(1999)080<0029: 216.
AOOTRO.2.0.CO;2.
Gleckler, P.J., K.E. Taylor, and C. Doutriaux, 2008: Performance metrics
Gates, W.L., 1992: AMIP: The Atmospheric Model Intercomparison for climate models, J. Geophysical Research, 113, D06104,
Project. Bulletin American Meteorological Society, 73, 1962–1970. doi:10.1029/2007JD008972.
Gent, P.R., et al., 2006: Ocean chlorofluorocarbon and heat uptake dur- Gleckler, P.J., K.R. Sperber, and K. AchutaRao, 2006: Annual cycle of
th
ing the 20 Century in the CCSM3. J. Climate, 19, 2366–2381. global ocean heat content: Observed and simulated. J. Geophysical
Research, 111, C06008.
Gent, P., and J.C. McWilliams, 1990: Isopycnal mixing in ocean circu-
lation models. J. Physical Oceanography, 20, 150–155. Gnanadesikan, A., et al., 2006: GFDL’s CM2 global coupled climate
models. Part II: The baseline ocean simulation. J. Climate, 19, 675–
GFDL Global Atmospheric Model Development Team (GAMDT), 697.
2004: The new GFDL global atmosphere and land model
AM2/LM2: Evaluation with prescribed SST conditions. J. Climate, Gorham, E., 1991: Northern peatlands: Role in the carbon cycle and
17(24), 4641–4673. probable responses to climatic warming. Ecological Applications, 1,
182–195.
Ghan, S. J., et al., 2000: An intercomparison of single column model
simulations of summertime midlatitude continental convection. J. Govindasamy, B., et al., 2005: Increase of carbon cycle feedback with
Geophysical Research, 105, 2091–2124. climate sensitivity: Results from a coupled climate and carbon cycle
model. Tellus, 57B, 153–163.
Gillett, N.P., and D.W.J. Thompson, 2003: Simulation of recent Southern
Hemisphere climate change. Science, 302, 273–275, Grabowski, W.W., 2003: MJO-like coherent structures: Sensitivity sim-
doi:10.1126/science.1087440. ulations using the cloud-resolving convection parameterization
(CRCP). J. Atmospheric Sciences, 60, 847–864.
Giorgi, F., and L.O. Mearns, 2003: Probability of regional climate change
based on the reliability ensemble averaging (REA) method. Geo- Grabowski, W.W., and M.W. Moncrieff, 2001: Large-scale organization
physical Research Letters, 30, Art. No. 1629. of tropical convection in two-dimensional explicit numerical simu-
lations. International J. Climatology, 127, 445–468.
Giorgi, F., et al., 2001: Regional climate change information—Evalua-
tion and projections. Chapter 10, 583–638, in Climate Change 2001: Gregory, J.M., J.A. Lowe, and S.F.B. Tett, 2006: Simulated global-mean
The Scientific Basis, ed. J.T. Houghton, et al. Cambridge University sea level changes over the last half-millennium. J. Climate, 19, 4576–
Press, Cambridge, U.K. 4592.
Giorgi, F., and X. Bi, 2000: A study of internal variability of a regional Gregory, J.M., et al., 2004: Simulated and observed decadal variability
climate model. J. Geophysical Research, 105, 29503–29521. in ocean heat content. Geophysical Research Letters, 31(15),
L15312.
Giorgi, F., and L.O. Mearns, 1999: Introduction to special section: Re-
gional climate modeling revisited. J. Geophysical Research, Gregory, J.M., 1999: Representation of the radiative effect of convec-
104(D6), 6335–6352. tive anvils. Hadley Centre Technical Note 7, Hadley Centre for Cli-
mate Prediction and Research, Met Office, Fitzroy Road, Exeter,
EX1 3BP, U.K.
104
Gregory, J.M., and J.F.B. Mitchell, 1997: The climate response to CO2 Guo, Z., D.H. Bromwich, and J.J. Cassano, 2003: Evaluation of polar
of the Hadley Centre coupled AOGCM with and without flux ad- MM5 simulations of Antarctic atmospheric circulation. Monthly
justment. Geophysical Research Letters, 24, 1943–1946. Weather Review, 131, 384–411.
Gregory, D., and S. Allen, 1991: The Effect of Convective Scale Down- Gutowski, W.J., et al., 2007a: A possible constraint on regional precipi-
drafts upon NWP and Climate Simulations, 122-123 in Ninth Con- tation intensity changes under global warming. J. Hydrometeorol-
ference on Numerical Weather Prediction. American Meteorological ogy, 8, 1382–1396, doi: 10.1175/2007JHM817.1.
Society, Denver, Colo.
Gutowski, W.J., et al., 2007b: Influence of Arctic wetlands on Arctic at-
Gregory, D., and P.R. Rowntree, 1990: A mass flux convection scheme mospheric circulation. J. Climate, 20, 4243–4254, doi:10.1175/
with representation of cloud ensemble characteristics and stability JCL14243.1.
dependent closure. Monthly Weather Review, 118, 1483–1506.
Gutowski, W.J., et al., 2003: Temporal-spatial scales of observed and sim-
Grell, G.A., et al., 2000a: Application of a multiscale, coupled ulated precipitation in central U.S. climate. J. Climate, 16, 3841–3847.
MM5/chemistry model to the complex terrain of the VOTALP val-
ley campaign. Atmospheric Environment, 34, 1435–1453. Gutowski, W.J., et al., 2002: A Coupled Land–Atmosphere Simulation
Program (CLASP). J. Geophysical Research, 107(D16), 4283,
Grell, G.A., et al., 2000b: Nonhydrostatic climate simulations of pre- doi:1029/2001JD000392.
cipitation over complex terrain. J. Geophysical Research–Atmos-
pheres, 105 (D24), 29595–29608. Gutowski, W.J., Z. Ötles, and Y. Chen, 1998: Effect of ocean-surface het-
erogeneity on climate simulation. Monthly Weather Review, 126,
Grell, G.A., H. Dudhia, and D. S. Stanfler, 1994: A description of the 1419–1429.
fifth generation Penn State–NCAR Mesoscale Model (MM5).
NCAR Technical Note. NCAR/TN-3981STR, National Center for Hack, J.J., 1994: Parameterization of moist convection in the National
Atmospheric Research, Boulder, Colo., 122 pp. [Available from Center for Atmospheric Research Community Climate Model
NCAR; P.O. Box 3000; Boulder, CO 80305] (CCM2). J. Geophysical Research, 99, 5551–5568.
Grell, G.A., 1993: Prognostic evaluation of assumptions used by cumu- Hagemann, S., and L. Dümenil, 1998: A parameterization of the lateral
lus parameterizations. Monthly Weather Review, 121, 764–787. water flow for the global scale. Climate Dynamics, 14, 17–31.
Griffies, S.M., et al., 2005: Formulation of an ocean model for global cli- Hall, A., 2004: The role of surface albedo feedback in climate. J. Cli-
mate simulations. Ocean Science, 1, 45–79. mate, 17, 1550–1568.
Griffies, S.M., et al., 2001: Tracer conservation with an explicit free sur- Hallberg, R., and A. Gnanadesikan, 2006: The role of eddies in deter-
face method for z-coordinate ocean models. Monthly Weather Re- mining the structure and response of the wind-driven Southern
view, 129, 1081–1098. Hemisphere overturning: Results from the Modeling Eddies in the
Southern Ocean (MESO) project. J. Physical Oceanography, 36,
Griffies, S.M., 1998: The Gent-McWilliams skew-flux. J. Physical 2232–2252.
Oceanography, 28, 831–841.
Haltiner, G.J., and R.T. Williams, 1980: Numerical Prediction and Dy-
Gritsun, A., and G. Branstator, 2007: Climate response using a three- namic Meteorology, Second Edition, John Wiley & Sons, New York,
dimensional operator based on the fluctuation–dissipation theorem. 477 pp.
J. Atmospheric Sciences, 64(7), 2558–2575
Han, J., and J.O. Roads, 2004: U.S. climate sensitivity simulated with
Gu, L., et al., 1999: Micrometeorology, biophysical exchanges, and NEE the NCEP regional spectral model. Climate Change, 62, 115–154.
decomposition in a two-storey boreal forest—Development and test
of an integrated model. Agricultural Forest Meteorology, 94, 123– Hansen, J., et al., 2007: Climate simulations for 1880–2003 with
148. GISS modelE. Climate Dynamics, 29(7-8), 661–696 (arxiv.org/
abs/physics/0610109).
Guilyardi, E., 2006: El Niño-mean state-seasonal cycle interactions in a
multi–model ensemble. Climate Dynamics, 26, 329–348. Hansen, J., et al., 2006: Global temperature change. Proceedings
National Academy Sciences, 103, 14288–14293, doi:10.1073/
pnas.0606291103.
105
Hansen, J., et al., 2005a: Earth’s energy imbalance: Confirmation and Hegerl, G.C., et al., 2006: Climate sensitivity constrained by temperature
implications. Science, 308, 1431–1435. reconstructions over the past seven centuries. Nature, 440, 1029–1032.
Hansen, J., et al., 2005b: Efficacy climate forcings. J. Geophysical Re- Hegerl, G.C., et al., 2004: Detectability of anthropogenic changes in an-
search, 110, D18104, doi:10.1029/2005JD005776. nual temperature and precipitation extremes. J. Climate, 17(19),
3683–3700.
Hansen, J., et al., 1996: Global surface air temperature in 1995: Return
to pre-Pinatubo level. Geophysical Research Letters, 23, 1665–1668, Held, I.M., and B.J. Soden, 2006: Robust responses of the hydrological
doi:10.1029/96GL01040. cycle to global warming. J. Climate, 19, 5686–5699.
Hansen, J., et al., 1993: How sensitive is the world’s climate? National Held, I.M., et al., 2005: Simulation of Sahel drought in the 20th and 21st
Geographic Research Exploration, 9, 42–158. centuries. Proceedings National Academy Sciences, 102, 17891–
17896.
Hansen, J., et al., 1985: Climate response times: Dependence on climate
sensitivity and ocean mixing. Science, 229, 857–859. Held, I.M., and B.J. Soden, 2000: Water vapor feedback and global
warming. Annual Review of Energy Environment, 25, 441–475.
Hansen, J., et al., 1984: Climate Sensitivity: Analysis of Feedback Mech-
anisms, 130–163 in Climate Processes and Climate Sensitivity. Mau- Held, I., and M.J. Suarez, 1994: A proposal for the intercomparison of the
rice Ewing Series, 5, ed. J.E. Hansen and T. Takahashi. American dynamical cores of atmospheric general circulation models. Bulletin
Geophysical Union, Washington, D.C. American Meteorological Society, 75(10), 1825–1830.
Hansen, J., et al., 1981: Climate impact of increasing atmospheric car- Held, I.M., R.S. Hemler, and V. Ramaswamy, 1993: Radiative-convective
bon dioxide. Science, 213, 957–966. equilibrium with explicit two-dimensional moist convection. J. At-
mospheric Sciences, 50, 3909–3927.
Harries, J.E., et al., 2001: Increases in greenhouse forcing inferred from
the outgoing longwave radiation spectra of the Earth in 1970 and Held, I.M., S.W. Lyons, and S. Nigam, 1989: Transients and the extrat-
1997. Nature, 410, 355–357. ropical response to El Niño. J. Atmospheric Sciences, 46(1), 163–174.
Hartmann, D.L., and K. Larson, 2002: An important constraint on trop- Helfand, H.M., and J.C. Labraga, 1988: Design of a non-singular level
ical cloud–climate feedback. Geophysical Research Letters, 29, 2.5 second order closure model for the prediction of atmospheric
1951, doi:10.1029/2002GL015835. turbulence. J. Atmospheric Sciences, 45, 113–132.
Hartmann, D.L., and M.L. Michelsen, 2002: No evidence for iris. Bul- Hellstrom, C., et al., 2001: Comparison climate change scenarios for
letin American Meteorological Society, 83(2), 249–254. Sweden based on statistical and dynamical downscaling of monthly
precipitation. Climate Research, 19, 45–55.
Hay, L.E., et al., 2006: One-way coupling of an atmospheric and a hy-
drologic model in Colorado. J. Hydrometeorology, 7, 569–589. Henderson-Sellers, A., 2006: Improving land-surface parameterization
schemes using stable water isotopes: Introducing the “iPILPS” ini-
Hayhoe, K., et al., 2004: Emissions pathways, climate change, and im- tiative. Global Planetary Change, 51, 3–24.
pacts on California. Proceedings National Academy Sciences
101(34), 12422–12427. Henderson-Sellers, A., et al., 2003: Predicting land-surface climates—
Better skill or moving targets? Geophysical Research Letters, 30,
Haylock, M.R., et al., 2006: Downscaling heavy precipitation over the 1777.
U.K.: A comparison of dynamical and statistical methods and their
future scenarios. International J. Climatology, 26, 1397–1415. Henderson-Sellers, A., et al., 1995: The Project for Intercomparison of
Land-Surface Parameterization Schemes (PILPS) —Phase 2 and
Hegerl, G., et al. 2007: Understanding and attributing climate change. In Phase 3. Bulletin American Meteorological Society, 76, 489–503.
Climate Change 2007: The Physical Science Basis. Contribution of
Working Group 1 to the Fourth Assessment of the Intergovernmental Hewitson, B.C., and R.G. Crane, 1996: Climate downscaling: Techniques
Panel on Climate Change. Cambridge University Press, Cambridge, and application. Climate Research, 7, 85–95.
U.K., and New York, N.Y.
Hewitt, C.D., and D.J. Griggs, 2004: Ensembles-based predictions: Cli-
mate changes and their impacts. Eos, 85, 566.
106
Heyen, H., H. Fock, and W. Greve, 1998: Detecting relationships be- Hori, M.E., and H. Ueda, 2006: Impact of global warming on the East
tween the interannual variability in ecological time series and cli- Asian winter monsoon as revealed by nine coupled atmosphere–
mate using a multivariate statistical approach—A case study on ocean GCMs. Geophysical Research Letters, 33,
Helgoland Roads zooplankton. Climate Research, 10, 179–191. doi:10.1029/2005GL024961.
Hibler, W.D., 1979: A dynamic thermodynamic sea ice model. J. Phys- Horowitz, L.W., et al., 2003: A global simulation of tropospheric ozone
ical Oceanography, 9, 815–846. and related tracers: Description and evaluation of MOZART, version
2. J. Geophysical Research–Atmospheres, 108(D24), 4784.
Hines, K.M., et al., 1999: Surface energy balance of the NCEP MRF and
NCEP-NCAR reanalysis in Antarctic latitudes during FROST. Houghton, R., 2003: Revised estimates of the annual net flux of carbon
Weather Forecasting, 14, 851–866. to the atmosphere from changes in land use and land management
1850–2000. Tellus, 55B, 378–390.
Hirano, A., R. Welch, and H. Lang, 2003: Mapping from ASTER stereo
image data: DEM validation and accuracy assessment. ISPRS J. Huang, B., P.H. Stone, and C. Hill, 2003: Sensitivities of deep-ocean
Photogrammetry Remote Sensing, 57, 356–370. heat uptake and heat content to surface fluxes and subgridscale pa-
rameters in an ocean general circulation model with idealized geom-
Hoerling, M., et al., 2006: Detection and attribution of 20th Century etry. J. Geophysical Research, 108(C1), 3015, doi:10.1029/
northern and southern African rainfall change. J. Climate, 19, 3989– 2001JC001218.
4008.
Hungate, B.A., et al., 2003: Nitrogen and climate change. Science, 302,
Hoffert, M.I., and C. Covey, 1992: Deriving global climate sensitivity 1512–1513.
from paleoclimate reconstructions. Nature, 360, 573–576.
Hunke, E.C., and Y. Zhang, 1999: A comparison of sea ice dynamics
Holland, M.M., and M.N. Raphael, 2006: Twentieth Century simulation models at high resolution. Monthly Weather Review, 127, 396–408.
of the Southern Hemisphere climate in coupled models. Part II: Sea
ice conditions and variability. Climate Dynamics, 26, 229–245. Hunke, E.C., and J.K. Dukowicz, 1997: An elastic-viscous-plastic model
for sea ice dynamics. J. Physical Oceanography, 27, 1849–1867.
Holland, M.M., and C.M. Bitz, 2003: Polar amplification: Climate
change in coupled models. Climate Dynamics, 21, 221–232. Hurrell, J.W., 1995: Decadal trends in the North Atlantic Oscillation and
relationships to regional temperature and precipitation. Science, 269,
Holtslag, A.A.M., and B.A. Boville, 1993: Local versus nonlocal bound- 676–679.
ary-layer diffusion in a global climate model. J. Climate, 6, 1825–
1842. Iorio, J.P., et al., 2004: Effects of model resolution and subgrid-scale
physics on the simulation of precipitation in the continental United
Hong, S.-Y., and H.-M.H. Juang, 1998: Orography blending in the lateral States. Climate Dynamics, 23, 243–258.
boundary of a regional model. Monthly Weather Review, 126, 1714–
1718. IPCC, 2007a: Couplings Between Changes in the Climate System and
Biogeochemistry. In Climate Change 2007: The Physical Science
Hood, R.R., et al., 2006: Pelagic functional group modeling: Progress, Basis. Contribution of Working Group 1 to the Fourth Assessment
challenges and prospects. Deep-Sea Research II, 53, 459–512. Report of the Intergovernmental Panel on Climate Change, ed. S.
Solomon et al. Cambridge University Press, Cambridge, U.K., and
Hoogenboom, G., J.W. Jones, and K.J. Boote, 1992: Modeling growth, New York (www.ipcc.ch).
development, and yield of grain legumes using SOYGRO, PNUT-
GRO, and BEANGRO—A review. Transactions ASAE, 35, 2043– IPCC, 2007b: Summary for Policymakers. In Climate Change 2007:
2056. The Physical Science Basis. Contribution of Working Group 1 to the
Fourth Assessment Report of the Intergovernmental Panel on Cli-
Hope, P.K., N. Nicholls, and J.L. McGregor, 2004: The rainfall response mate Change, ed. S. Solomon et al. Cambridge University Press,
to permanent inland water in Australia. Australian Meteorological Cambridge, U.K., and New York (www.ipcc.ch).
Magazine, 53, 251–262.
IPCC, 2001: Climate Change 2001: The Scientific Basis: Contribution
Horel, J.D., and J.M. Wallace, 1981: Planetary-scale atmospheric phe- of Working Group 1 to the Third Assessment Report of the Intergov-
nomena associated with the Southern Oscillation. Monthly Weather ernmental Panel on Climate Change, ed. J.T. Houghton et al. Cam-
Review, 109, 813–829. bridge University Press, Cambridge, U.K., and New York.
107
IPCC, 2000: Special Report on Emissions Scenarios, ed. N. Nakicen- Kalkstein, L.S., and J.S. Greene, 1997: An evaluation of climate and
ovic and R. Swart (www.grida.no/climate/ipcc/emission/). mortality relationships in larger U.S. cities and the possible impacts
of a climate change. Environmental Health Perspectives, 105, 84–93.
IPCC, 1990: Climate Change: The IPCC Scientific Assessment, ed. J.T.
Houghton, G.J. Jenkins, and J.J. Ephraums. Cambridge University Kalnay, E., et al., 1996: The NCEP/NCAR 40-year reanalysis project.
Press, Cambridge, U.K. Bulletin American Meteorological Society, 77(3), 437–471.
Irannejad, P., A. Henderson-Sellers, and S. Sharmeen, 2003: Importance Kanamitsu, M., et al., 2002: NCEP-DEOAMIP-II reanalysis (R-2). Bul-
of land-surface parameterization for latent heat simulation. Geo- letin American Meteorological Society, 83, 1631–1643.
physical Research Letters, 30, 1904.
Kattenberg, A., et al., 1996: Climate Models Projections of Future Cli-
Jakob, C., and G. Tselioudis, 2003: Objective identification of cloud mate. Chapter 6, Climate Models–Projections of Future Climate, 285–
regimes in the tropical western Pacific. Geophysical Research Let- 358, in Climate Change 1995–The Science. Climate Change, ed. J.T.
ters, 30, 2082. Houghton, et al. Cambridge University Press, Cambridge, U.K.
Janowiak, J.E., 1988: An investigation of interannual rainfall variability Kattsov, V., and E. Källén, 2005: Future Changes of Climate: Modelling
in Africa. J. Climate, 1, 240–255. and Scenarios for the Arctic Region in Arctic Climate Impact As-
sessment (ACIA). Cambridge University Press, Cambridge, U.K.,
Jickells, T.D., et al., 2005: Global iron connections between desert dust, 1042 pp.
ocean biogeochemistry, and climate. Science, 308, 67–71.
Kattsov, V.M., et al., 2000: Atmospheric Climate Models: Simulation of
Jones, C.A., and J.R. Kiniry, 1986: CERES-Maize: A Simulation Model the Arctic Ocean Fresh Water Budget Components, 209–247 in The
of Maize Growth and Development. Texas A&M University Press, Freshwater Budget of the Arctic Ocean, ed. E.L. Lewis. Kluwer Ac-
College Station, Tex. ademic Publishers, Dordrecht, The Netherlands.
Jones, P.D., et al., 2006: Global and hemispheric temperature anom- Khain, A., and A. Pokrovsky, 2004: Simulation of effects of atmospheric
alies—land and marine instrumental records. Carbon Dioxide In- aerosols on deep turbulent convective clouds using a spectral mi-
formation Analysis Center, Oak Ridge National Laboratory, U.S. crophysics mixed-phase cumulus cloud model. Part II: Sensitivity
DOE (cdiac.esd.ornl.gov/trends/temp/jonescru/jones.html). study. J. Atmospheric Sciences, 61, 2963–2982.
Jones, P.D., et al., 1999: Surface air temperature and its changes over the Khairoutdinov, M., D. Randall, and C. DeMott: 2005: Simulations of the
past 150 years. Reviews Geophysics, 37, 173–199. Atmospheric General Circulation Using a Cloud-Resolving Model
as a Superparameterization of Physical Processes, J. Atmospheric
Jones, P.D., T. Jónsson, and D. Wheeler, 1997: Extension to the North Sciences, 62(7), 2136–2154, doi:10.1175/JAS3453.1.
Atlantic Oscillation using early instrumental pressure observations
from Gibraltar and south-west Iceland. International J. Climatology, Kidson, J.W., and C.S. Thompson, 1998: Comparison of statistical and
17, 1433–1450. model-based downscaling techniques for estimating local climate
variations. J. Climate, 11, 735–753.
Jones, R.G., J.M. Murphy, and M. Noguer, 1995: Simulation of climate
change over Europe using a nested regional climate model. Part I: Kiehl, J.T., 2007: Twentieth Century climate model response and climate
Assessment of control climate, including sensitivity to location of sensitivity. Geophysical Research Letters, 34, L22710,
lateral boundaries. International J. Climatology, 121, 1413–1449. doi:10.1029/2007GL031383.
Joseph, R., and S. Nigam, 2006: ENSO evolution and teleconnections in Kiehl, J.T., et al., 2006: The climate sensitivity of the Community Climate
IPCC’s 20th Century climate simulations: Realistic representation? System Model: CCSM3. J. Climate, 19, 2584–2596.
J. Climate 19(17), 4360–4377.
Kiehl, J.T., et al., 1996: Description of the NCAR Community Climate
Kain, J.S., and J.M. Fritsch, 1993: Convective Parameterization in Model (CCM3). NCAR Technical Note. NCAR/TN-420+STR, Na-
Mesoscale Models: The Kain-Fritsch Scheme. In The Representa- tional Center for Atmospheric Research, Boulder, Colo., 152 pp.
tion of Cumulus Convection in Numerical Models, Meteorological [Available from NCAR; P.O. Box 3000; Boulder, CO 80305]
Monographs. American Meteorological Society 46, 165–170.
Kiktev, D., et al., 2003: Comparison of modeled and observed trends in
indices of daily climate extremes. J. Climate, 16, 3560–3571.
108
Kim, J., et al., 2005: The effects of the Gulf of California SSTs on warm- Koren, V., et al., 1999: A parameterization of snowpack and frozen
season rainfall in the southwestern United States and northwestern ground intended for NCEP weather and climate models. J. Geo-
Mexico: A regional model study. J. Climate, 18, 4970–4992. physical Research, 104(D16), 19569–19585.
Kim, J., and J.E. Lee, 2003: A multiyear regional climate hindcast for Kraus, E.B., and J.S. Turner, 1967: A one-dimensional model of the sea-
the western United States using the mesoscale atmospheric simula- sonal thermocline. II: The general theory and its consequences. Tel-
tion model. J. Hydrometeorology, 4(5), 878–890. lus, 19, 98–105.
Kimoto, M., 2005: Simulated change of the East Asian circulation under Kravtsov, S.V., and C. Spannagle, 2007: Multi-decadal climate variabil-
global warming scenario. Geophysical Research Letters, 32, L16701. ity in observed and modeled surface temperatures. J. Climate, sub-
mitted.
King, J.C., and J. Turner, 1997: Antarctic Meteorology and Climatology.
Cambridge University Press, Cambridge, U.K., 256 pp. Krinner, G., 2003: Impact of lakes and wetlands on boreal climate. J.
Geophysical Research, 108(D16), 4520.
Kirschbaum, M.U.F., 2000: Will changes in soil organic carbon act as a
positive or negative feedback on global warming? Biogeochemistry, Krinner, E., 1991: Northern peatlands: Role in the carbon cycle and
48, 21–51. probable responses to climatic warming. Ecological Applications, 1,
182–195.
Kitoh, A., and T. Uchiyama, 2006: Changes in onset and withdrawal of
the East Asian summer rainy season by multi-model global warming Kripalani, R.H., J.H. Oh, and H.S. Chaudhari, 2007: Response of the
experiments. J. Meteorological Society Japan, 84, 247–258. East Asian summer monsoon to doubled atmospheric CO2: Coupled
climate models simulations and projections under IPCC AR4. The-
Kleidon, A., 2004: Global datasets of rooting zone depth inferred from oretical Applied Climatology, 87, 1–28.
inverse methods. J. Climate, 17, 2714–2722.
Kripalani, R.H., et al., 2007: South Asian summer monsoon precipita-
Klein, S.A., and C. Jakob, 1999: Validation and sensitivities of frontal tion variability: Coupled climate simulations and projections under
clouds simulated by the ECMWF model. Monthly Weather Review, IPCC AR4. Theoretical Applied Climatology, doi:10.1007/s00704-
127, 2514–2531. 006-0282-0.
Klemp, J.B., and R. Wilhelmson, 1978: The simulation of three-dimen- Krueger, S.K., 1988: Numerical simulation of tropical cumulus clouds
sional convective storm dynamics. J. Atmospheric Sciences, 35, and their interaction with the subcloud layer. J. Atmospheric Sci-
1070–1096. ences, 45, 2221–2250.
Knowles, N., and D.R. Cayan, 2004: Elevational dependence of projected Kuang, Z., and D.L. Hartmann, 2007: Testing the fixed anvil temperature
hydrologic changes in the San Francisco estuary and watershed. Cli- (FAT) hypothesis in a cloud-resolving model. J. Climate, 20, 2051–
matic Change, 62, 319–336. 2057.
Knutson, T.R., et al., 2007: Simulation of recent multidecadal increase Kunkel, K.E., et al., 2006: Can CGCMs simulate the Twentieth Century
of Atlantic hurricane activity using an 18-km regional model. Bul- “warming hole” in the central United States? J. Climate, 19, 4137–
letin American Meteorological Society, 88, 1549–1565. 4153.
Knutson, T.R., et al., 2006: Assessment of Twentieth Century regional Kunkel, K.E., et al., 2004: Temporal variations in frost-free season in
surface temperature trends using the GFDL CM2 coupled models. J. the United States: 1895–2000. Geophysical Research Letters, 31,
Climate, 19(9), 1624–1651. L03201.
Knutson, T.R., and S. Manabe, 1998: Model assessment of decadal vari- Kunkel, K.E., et al., 2002: Observations and regional climate model sim-
ability and trends in the tropical Pacific Ocean. J. Climate, 11, 2273– ulations of heavy precipitation events and seasonal anomalies: A
2296. comparison. J. Hydrometeorology, 3, 322–334.
Knutti, R., et al., 2006: Constraining climate sensitivity from the sea- Kunkel, K.E., et al., 1996: The July 1995 heat wave in the Midwest: A
sonal cycle in surface temperature. J. Climate, 19(17), 4224–4233. climatic perspective and critical weather factors. Bulletin American
Meteorological Society, 77, 1507–1518.
109
Kuo, H.L., 1974: Further studies of the parameterization of the influence Leung, L.R., Y. Qian, and X. Bian, 2003: Hydroclimate of the western
of cumulus convection on large-scale flow. J. Atmospheric Sciences, United States based on observations and regional climate simulation
31, 1232–1240. of 1981–2000. Part I: Seasonal statistics. J. Climate, 16(12), 1892–
1911.
Laprise, R., 2003: Resolved scales and nonlinear interactions in limited-
area models. J. Atmospheric Sciences, 60, 768–779. Leung, L.R., and M.S. Wigmosta, 1999: Potential climate change im-
pacts on mountain watersheds in the Pacific Northwest. J. American
Large, W., J.C. McWilliams, and S.C. Doney, 1994: Oceanic vertical Water Resources Association, 35, 1463–1471.
mixing: A review and a model with a nonlocal boundary mixing pa-
rameterization. Reviews Geophysics, 32, 363–403. Levitus, S., et al., 2001: Anthropogenic warming of Earth’s climate sys-
tem. Science, 292, 267–270.
Latif, M., et al., 2001: ENSIP: The El Niño simulation intercomparison
project. Climate Dynamics, 18, 255–272. Li, K.Y., et al., 2006: Root-water-uptake based upon a new water stress
reduction and an asymptotic root distribution function. Earth Inter-
th
Lau, K.-M., et al., 2006: A multi-model study of the 20 Century simu- actions, 10, Art. No. 14.
lations of Sahel drought from the 1970s to 1990s. J. Geophysical Re-
search, 111(D0711). Li, W., R. Fu, and R.E. Dickinson, 2006: Rainfall and its seasonality over
the Amazon in the 21st Century as assessed by the coupled models
Lawrence, D.M., and A.G. Slater, 2005: A projection of severe near-sur- for the IPCC AR4. J. Geophysical Research, 11, D02111.
face permafrost degradation during the 21st Century. Geophysical
Research Letters, 32, L24401. Li, X., and T. Koike, 2003: Frozen soil parameterization in SiB2 and its
validation with GAME-Tibet observations. Cold Regions Science
Le Treut, H., Z.X. Li, and M. Forichon, 1994: Sensitivity of the LMD Technology, 36, 165–182.
general circulation model to greenhouse forcing associated with two
different cloud water parameterizations. J. Climate, 7, 1827–1841. Li, X., et al., 2001: A comparison of two vertical-mixing schemes in a Pa-
cific Ocean general circulation model. J. Climate, 14, 1377–1398.
Le Treut, H., and Z.X. Li, 1991: Sensitivity of an atmospheric general
circulation model to prescribed SST changes: Feedback effects as- Li, Z.X., 1999: Ensemble atmospheric GCM simulation climate inter-
sociated with the simulation of cloud optical properties. Climate Dy- annual variability from 1979 to 1994. J. Climate, 12, 986–1001.
namics, 5, 175–187.
Liang, X.-Z., et al., 2006: Regional climate model downscaling of the
Leaman, K., R. Molinari, and P. Vertes, 1987: Structure and variability U.S. summer climate and future change. J. Geophysical Research,
of the Florida current at 27N: April 1982–July 1984. J. Physical 111, D10108.
Oceanography, 17, 565–583.
Liang, X.-Z., et al., 2004: Regional climate model simulation of summer
Lean, J., J. Beer, and R. Bradley, 1995: Reconstruction of solar irradiance precipitation diurnal cycle over the United States. Geophysical Re-
since 1610: Implications for climate change. Geophysical Research search Letters, 31, L24208.
Letters, 22, 3195–3198.
Liang, X., Z. Xie, and M. Huang, 2003: A new parameterization for sur-
Lee, H.-C., A. Rosati, and M.J. Spelman, 2006: Barotropic tidal mixing face and groundwater interactions and its impact on water budgets
effects in a coupled climate model: Ocean conditions in the northern with the variable infiltration capacity (VIC) land surface model. J.
Atlantic. Ocean Modelling, 11, 464–477. Geophysical Research, 108, 8613.
Leith, C., 1975: Climate response and fluctuation dissipation. J. Atmos- Liang, X.-Z., K.E. Kunkel, and A.N. Samel, 2001: Development of a re-
pheric Sciences, 32, 2022–2026. gional climate model for U.S. midwest applications. Part 1: Sensi-
tivity to buffer zone treatment. J. Climate, 14, 4363–4378.
Leung, L.R., et al., 2004: Mid-century ensemble regional climate change
scenarios for the western United States. Climate Change, 62, 75– Libes, S.M., 1992: An Introduction to Marine Biogeochemistry. New
113. York, Wiley, 734 pp.
Leung, L.R., and Y. Qian, 2003: The sensitivity of precipitation and Lin, J.L., et al., 2006: Tropical intraseasonal variability in 14 IPCC AR4
snowpack simulations to model resolution via nesting in regions of climate models. Part I: Convective signals. J. Climate, 19(12), 2665–
complex terrain. J. Hydrometeorology, 4, 1025–1043. 2690.
110
Lin, J.L., et al., 2004: Stratiform precipitation, vertical heating profiles, Lorenz, P., and D. Jacob, 2005: Influence of regional scale information
and the Madden–Julian Oscillation. J. Atmospheric Sciences, 61, on the global circulation: A two-way nesting climate simulation.
296–309. Geophysical Research Letters, 32, Art. No. L18706.
Lin, S.J., and R.B. Rood, 1996: Multidimensional flux-form semi-la- Lucarini, V., et al., 2006: Intercomparison of the Northern Hemisphere
grangian transport schemes. Monthly Weather Review, 124, 2046– winter mid-latitude atmospheric variability of the IPCC models. Cli-
2070. mate Dynamics, 28(7–8), 829–848.
Lin, W.Y., and M.H. Zhang, 2004: Evaluation of clouds and their radia- Lumpkin, R., and K. Speer, 2003: Large-scale vertical and horizontal
tive effects simulated by the NCAR Community Atmospheric Model circulation in the North Atlantic Ocean. J. Physical Oceanography,
CAM2 against satellite observations. J. Climate, 17, 3302–3318. 33, 1902–1920.
Lindzen, R.S., M.-D. Chou, and A. Y. Hou, 2001: Does Earth have an Luo, L.F., et al., 2003: Effects of frozen soil on soil temperature, spring
adaptive infrared iris? Bulletin American Meteorological Society, 82, infiltration, and runoff: Results from the PILPS 2(d) experiment at
417–432. Valdai, Russia. J. Hydrometeorology, 4, 334–351.
Lindzen, R.S., and C. Giannitsis, 1998: On the climatic implications of Lynch, A.H., J.A. Maslanik, and W.L. Wu, 2001: Mechanisms in the de-
volcanic cooling. J. Geophysical Research, 103(D6), 5929–5941. velopment of anomalous sea ice extent in the western Arctic: A case
study. J. Geophysical Research, 106(D22), 28097–28105.
Lindzen, R.S., 1994: Climate dynamics and global change. Annual Re-
view Fluid Mechanics, 26, 353–378, doi:10.1146/annurev.fl. Lynch, A.H., et al., 1995: Development of a regional climate model of
26.010194.002033. the western Arctic. J. Climate, 8, 1555–1570.
Lindzen, R.S., and M.S. Fox-Rabinovitz, 1989: Consistent vertical and Lynn, B.H., et al., 2005: Spectral (bin) microphysics coupled with a
horizontal resolution. Monthly Weather Review, 117, 2575–2583. mesoscale model (MM5). Part II: Simulation of a CaPE rain event
with a squall line. Monthly Weather Review, 133, 59–71.
Liston, G.E., 2004: Representing subgrid snow cover heterogeneities in
regional and global models. J. Climate, 17, 1381–1397. Maak, K., and H. von Storch, 1997: Statistical downscaling of monthly
mean temperature to the beginning of flowering of Galanthus nivalis
Liu, Z., 1998: On the role of ocean in the transient response of tropical L. in Northern Germany. International J. Biometeorology,
climatology to global warming. J. Climate, 11, 864–875. 41(1), 5–12.
Lock, A.P., et al., 2000: A new boundary layer mixing scheme. Part I: Malevsky–Malevich, S.P., et al., 1999: The evaluation of climate change
Scheme description and single-column model tests. Monthly Weather influence on the permafrost season soil thawing regime. Contempo-
Review, 128, 3187–3199. rary Investigation Main Geophysical Observatory, 1, 33–50 (in
Russian).
Lock, A., 1998: The parameterization of entrainment in cloudy bound-
ary layers. Quarterly J. Royal Meteorological Society, 124, 2729– Maltrud, M.E., et al., 1998: Global eddy-resolving ocean simulations
2753. driven by 1985–1995 atmospheric winds. J. Geophysical Research,
103(C13), 30, 825–30, 853.
Lockwood, M., and C. Fröhlich, 2007: Recent oppositely directed trends
in solar climate forcings and the global mean surface air temperature. Manabe, S., et al., 1991: Transient responses of a coupled ocean-at-
Proceedings Royal Society A, 463, 2447–2460, mosphere model to gradual changes of atmospheric CO2. Part I: An-
doi:10.1098/rspa.2007.1880. nual mean response. J. Climate, 4, 785–818.
Lofgren, B.M., 2004: A model for simulation of the climate and hydrol- Manabe, S., and A. Broccoli, 1985: A comparison of climate model sen-
ogy of the Great Lakes basin. J. Geophysical Research, 109(D18), sitivity with data from the last glacial maximum. J. Atmospheric Sci-
Art. No. D18108. ences, 42, 2643–2651.
Lohmann, U., and E. Roeckner, 1996: Design and performance of a new Manabe, S., and R.T. Wetherald, 1975: The effects of doubling the CO2
cloud microphysics scheme developed for the ECHAM4 general cir- concentration on the climate of a general circulation model. J. At-
culation model. Climate Dynamics, 12, 557–572. mospheric Sciences, 32, 3–15.
111
Manabe, S., 1969: Climate and the ocean circulation. 1: The atmospheric McGregor, J.L., 1997: Regional climate modelling. Meteorology At-
circulation and hydrology of the Earth’s surface. Monthly Weather mospheric Physics, 63, 105–117.
Review, 97, 739–774.
McPhaden, M.J., et al., 1998: The Tropical Ocean Global Atmosphere
Manabe, S., and R.T. Wetherald, 1967: Thermal equilibrium of the at- (TOGA) observing system: A decade of progress. J. Geophysical
mosphere with a given distribution of relative humidity. J. Atmos- Research, 103, 14169–14240.
pheric Sciences, 24(3), 241–259.
Mearns, L.O., 2003: Issues in the impacts of climate variability and
Manabe, S., J. Smagorinsky, and R.F. Strickler, 1965: Simulated clima- change on agriculture—Applications to the southeastern United
tology of a general circulation model with a hydrological cycle. States. Climate Change, 60, 1–6.
Monthly Weather Review, 93, 769–798.
Mearns, L.O., et al., 2003: Climate scenarios for the southeastern U.S.
Martin, G.M., et al., 2000: A new boundary layer mixing scheme. Part based on GCM and regional model simulations. Climate Change, 60,
II: Tests in climate and mesoscale models. Monthly Weather Review, 7–35.
128, 3200–3217.
Mearns, L.O., et al., 1999: Comparison of climate change scenarios gen-
Martin, J.H., et al., 1994: Testing the iron hypothesis in ecosystems of the erated from regional climate model experiments and statistical down-
equatorial Pacific Ocean. Nature, 371, 123–129. scaling. J. Geophysical Research, 104, 6603–6621.
Martin, J.H., 1991: Iron as a limiting factor in oceanic productivity. In Mechoso, C.R., et al., 1995: The seasonal cycle over the tropical Pacific
Primary Productivity and Biogeochemical Cycles in the Sea, ed. P.G. in coupled ocean-atmosphere general circulation models. Monthly
Falkowski and A.D. Woodhead, 123–137. Plenum Press, New York. Weather Review, 123, 2825–2838.
Maurer, E. P., et al., 2002: A long-term hydrologically-based data set of Meehl, G.A., et al., 2006: Climate change projections for the Twenty-
land surface fluxes and states for the conterminous United States. J. First Century and climate change commitment in the CCSM3. J. Cli-
Climate, 15, 3237–3251. mate, 19, 2597–2616.
Maxwell, R.M., and N.L. Miller, 2005: Development of a coupled land Meehl, G.A., et al., 2005: How much more will global warming and sea
surface and groundwater model. J. Hydrometeorology, 6, 233–247. level rise? Science, 307, 1769–1772.
May, W., and E. Roeckner, 2001: A time-slice experiment with the Meehl, G.A., and C. Tebaldi, 2004: More intense, more frequent, and
ECHAM4 AGCM at high resolution: The impact of horizontal res- longer lasting heat waves in the 21st Century. Science, 305, 994–997.
olution on annual mean climate change. Climate Dynamics 17(5–
6), 407–420. Meehl, G.A., C. Tebaldi, and D. Nychka, 2004: Changes in frost days in
simulations of Twenty-First Century climate. Climate Dynamics, 23,
McCreary, J., and P. Lu, 1994: Interaction between the subtropical and 495–511.
equatorial ocean circulation—The subtropical cell. J. Physical
Oceanography, 24, 466–497. Mellor, G.L., and T. Yamada, 1982: Development of a turbulence clo-
sure model for geophysical fluid problems. Reviews Geophysics
McCulloch, M.T., and T. Ezat, 2000: The coral record of the last inter- Space Physics, 20, 851–875.
glacial sea levels and sea surface temperatures. Chemical Geology,
169, 107–129. Mellor, G.L., and T. Yamada, 1974: A hierarchy of turbulent closure
models for planetary boundary layers. J. Atmospheric Sciences, 31,
McFarlane, N.A., 1987: The effect of orographically excited gravity 1791–1806.
wave drag on the general circulation of the lower stratosphere and
troposphere. J. Atmospheric Sciences, 44, 1775–1800. Menon, S., and A. Del Genio, 2007: Evaluating the Impacts of Car-
bonaceous Aerosols on Clouds and Climate, 34–48 in Human-In-
McGregor, J.L., 1999: Regional Modelling at CAR: Recent Develop- duced Climate Change: An Interdisciplinary Assessment, ed. M.E.
ments, 43–48 in Parallel Computing in Meteorology and Oceanog- Schlesinger et al. Cambridge University Press, Cambridge, U.K., and
raphy. BMRC Research Report No. 75, Bureau of Meteorology, New York, N.Y.
Melbourne, Australia.
Merryfield, W.J., 2006: Changes to ENSO under CO2 doubling in a
multi-model ensemble. J. Climate, 19, 4009–4027.
112
Miguez-Macho, G., G.L. Stenchikov, and A. Robock, 2005: Regional Morel, A., and D. Antoine, 1994: Heating rate within the upper ocean in
climate simulations over North America: Interaction of local relation to its bio-optical state. J. Physical Oceanography, 24, 1652–
processes with improved large-scale flow. J. Climate, 18, 1227–1246. 1665.
Miller, D.A., and R.A. White, 1998: A coterminous United States mul- Morrison, H., and J.O. Pinto, 2005: Mesoscale modeling of springtime
tilayer soil characteristics dataset for regional climate and hydrology Arctic mixed-phase stratiform clouds using two-moment bulk mi-
modeling. Earth Interactions, 2, 1–26. crophysics scheme. J. Atmospheric Sciences, 62, 3683–3704.
Miller, R.L., G.A. Schmidt, and D.T. Shindell, 2006: Forced variations Murphy, J.M., et al., 2004: Quantification of modelling uncertainties in
of annular modes in the 20th Century IPCC AR4 simulations. J. Geo- a large ensemble of climate change simulations. Nature, 430(7001),
physical Research, 111, D18101, doi:10.1029/2005JD006323. 768–772.
Min, S.-K., and A. Hense, 2007: A Bayesian assessment of climate Murray, R.J., 1996: Explicit generation of orthogonal grids for ocean
change using multi-model ensembles. Part II: Regional and seasonal models. J. Computational Physics, 126, 251–273.
mean surface temperature. J. Climate, 20(12), 1769–1790.
Nadelhoffer, K.J., et al., 1999: Nitrogen makes a minor contribution to
Min, S.-K., and A. Hense, 2006: A Bayesian assessment of climate carbon sequestration in temperate forests. Nature, 398, 145–148.
change using multi-model ensembles. Part I: Global mean surface
temperature. J. Climate, 19, 3237– 3256. Najjar, R.G., et al., 2007: Impact of circulation on export production,
dissolved organic matter, and dissolved oxygen in the ocean: Results
Minschwaner, K., E.D. Essler, and P. Sawaengphokhai, 2006: Multi- from OCMIP-2. Global Biogeochemistry Cycles, 21, GB3007,
model analysis of the water vapor feedback in the tropical upper tro- doi:10.1029/2006GB002857.
posphere. J. Climate, 19, 5455–5464.
NARCCAP (North American Regional Climate Change Assessment),
Minschwaner, K., and A.E. Dessler, 2004: Water vapor feedback in the cited as 2007: www.narccap.ucar.edu/.
tropical upper troposphere: Model results and observations. J. Cli-
mate, 17(6), 272–1282. Neelin, J.D., et al., 1992: Tropical air-sea interaction in general circula-
tion models. Climate Dynamics, 7, 73–104.
Mirocha, J.D., B. Kosovic, and J.A. Curry, 2005: Vertical heat transfer in
the lower atmosphere over the Arctic Ocean during clear-sky periods. Niu, G.Y., and Z.L. Yang, 2006: Effects of frozen soil on snowmelt runoff
Boundary-Layer Meteorology, 117, 37–71. and soil water storage at a continental scale. J. Hydrometeorology, 7,
937–952.
Mitchell, J.F.B., et al., 1999: Towards the construction of climate change
scenarios. Climatic Change, 41(3–4), 547–581. Nordeng, T.E., 1994: Extended Versions of the Convective Parameteri-
zation Scheme at ECMWF and Their Impact on the Mean and Tran-
Mitchell, J.F.B., C.A. Senior, and W.J. Ingram, 1989: CO2 and climate: sient Activity of the Model in the Tropics. Technical Memorandum
A missing feedback? Nature, 341, 132–134. 206, European Center for Medium Range Weather Forecasting
(ECMWF), Reading, U.K.
Mitchell, T., 2003: Pattern scaling: An examination of the accuracy of the
technique for describing future climates. Climatic Change, 60, 217– Norris, J., and C.P. Weaver, 2001: Improved techniques for evaluating
242. GCM cloudiness applied to the NCAR CCM3. J. Climate, 14, 2540–
2550.
Miura, H., et al., 2005: A climate sensitivity test using a global cloud
resolving model under an aqua planet condition. Geophysical Re- Nowak, R.S., D.S. Ellsworth, and S.D. Smith, 2004: Tansley review:
search Letters, 32, L19717. Functional responses of plants to elevated atmospheric CO2—Do
photosynthetic and productivity data from FACE experiments sup-
Mo, K.C., et al., 2005: Impact of model resolution on the prediction of port early predictions? New Phytologist, 162, 253–280.
summer precipitation over the United States and Mexico. J. Climate,
18, 3910–3927. Oglesby, R.J., and B. Saltzman, 1992: Equilibrium climate statistics of
a general circulation model as a function of atmospheric carbon diox-
Moorthi, S., and M.J. Suarez, 1992: Relaxed Arakawa-Schubert: A pa- ide. Part I: Geographic distributions of primary variables. J. Climate,
rameterization of moist convection for general circulation models. 5(1), 66–92.
113
Ohlmann, J.C., 2003: Ocean radiant heating in climate models. J. Cli- Parkinson, C.L., K.Y. Vinnikov, and D.J. Cavalieri, 2006b: Correction to
mate, 16, 1337–1351. evaluation of the simulation of the annual cycle of Arctic and Antarc-
tic sea ice coverages by 11 major global climate models. J. Geo-
Olesen, J.E., et al., 2007: Uncertainties in projected impacts of climate physical Research, 111, C12009, doi:10.1029/2006JC003949.
change on European agriculture and terrestrial ecosystems based on
scenarios from regional climate models. Climatic Change, 81, 123– Paulson, C.A., and J.J. Simpson, 1977: Irradiance measurements in the
143. upper ocean. J. Applied Oceanography, 7, 952–956.
Oleson, K.W., et al., 2004: Technical Description of the Community Land Pavolonis, M., J.R. Key, and J.J. Cassano, 2004: Study of the Antarctic
Model (CLM). NCAR Technical Note. NCAR/TN-461+STR, Na- surface energy budget using a polar regional atmospheric model
tional Center for Atmospheric Research, Boulder, Colo., 173 pp. forced with satellite-derived cloud properties. Monthly Weather Re-
[Available from NCAR; P.O. Box 3000; Boulder, CO 80305] view, 132, 654–661.
Oouchi, K., et al., 2006: Tropical cyclone climatology in a global-warm- Pawson, S., et al., 2000: The GCM-Reality Intercomparison Project for
ing climate as simulated in a 20- km mesh global atmospheric model: SPARC GRIPS): Scientific issues and initial results. Bulletin Amer-
Frequency and wind intensity analyses. J. Meteorological Society ican Meteorological Society, 81, 781–796.
Japan, 84, 259–276.
Payne, A.J., et al., 2004: Recent dramatic thinning of largest West Antarc-
Oren, R., et al., 2001: Soil fertility limits carbon sequestration by forest tic ice stream triggered by oceans. Geophysical Research Letters,
ecosystems in a CO2–enriched atmosphere. Nature, 411, 469–477. 31, L23401.
Ott, L.E., et al., 2007: The effects of lightning NOx production during the Payne, J.T., et al., 2004: Mitigating the effects of climate change on the
21 July EULINOX storm studied with a 3-D cloud-scale chemical water resources of the Columbia River basin. Climatic Change, 62,
transport mode. J. Geophysical Research, 112, D05307, 233–256.
doi:10.1029/2006JD007365.
Peltier, W.R., 2004: Global glacial isostasy and the surface of the ice-
Our Changing Planet: The FY 1992 U.S. Global Change Research age earth: The ice-5G (VM2) model and GRACE. Annual Review
Program. 1991. A Report by the Committee on Earth and Environ- Earth Planetary Science, 32, 111–149.
mental Sciences. A Supplement to the U.S. President’s Fiscal Year
1992 Budget. Philander, S.G.H., 1990: El Niño, La Niña, and the Southern Oscillation.
Academic Press, San Diego, Calif., 293 pp.
Overgaard, J., D. Rosbjerg, and M.B. Butts, 2006: Land-surface mod-
elling in hydrological perspective–A review. Biogeoscience, 3, 229– Philander, S.G.H., and R.C. Pacanowski, 1981: The oceanic response to
241. cross-equatorial winds (with application to coastal upwelling in low
latitudes). Tellus, 33, 201–210.
Pacanowski, R.C., and S.G.H. Philander, 1981: Parameterization of ver-
tical mixing in numerical models of tropical oceans. J. Physical Philip, S.Y., and G. J. van Oldenborgh, 2006: Shifts in ENSO coupling
Oceanography, 11, 1443–1451. processes under global warming. Geophysical Research Letters, 33,
L11704.
Paegle, J., K.C. Mo, and J. Nogués-Paegle, 1996: Dependence of simu-
lated precipitation on surface evaporation during the 1993 United Phillips, T.J., et al., 2004: Evaluating parameterizations in general circu-
States summer floods. Monthly Weather Review, 124, 345–361. lation models: Climate simulation meets weather prediction. Bul-
letin American Meteorological Society, doi:10.1175/
Pan, Z., et al., 2004: Altered hydrologic feedback in a warming climate BAMS-85-12-1903.
introduces a “warming hole.” Geophysical Research Letters, 31,
L17109, doi:10.1029/2004GL020528. Piani, C., et al., 2005: Constraints on climate change from a multi-thou-
sand member ensemble of simulations. Geophysical Research Let-
Pan, Z., et al., 2001: Evaluation of uncertainties in regional climate ters, 32, L23825.
change simulations. J. Geophysical Research, 106, 17735–17752.
Pierce, D.W., et al., 2006: Three-dimensional tropospheric water vapor in
Parkinson, C.L., K.Y. Vinnikov, and D.J. Cavalieri, 2006a: Evaluation coupled climate models compared with observations from the AIRS
of the simulation of the annual cycle of Arctic and Antarctic. J. Geo- satellite system. Geophysical Research Letters, 33, L21701.
physical Research, 111, C07012.
114
Pierce, D.W., 2004: Future change in biological activity in the north Pa- Raisanen, J., 2002: CO2–induced changes in interannual temperature
cific due to anthropogenic forcing of the physical environment. Cli- and precipitation variability in 19 CMIP experiments. J. Climate, 15,
matic Change, 62, 389–418. 2395–2411.
Pincus, R., H.W. Barker, and J.J. Morcrette, 2003: A fast, flexible, Raisanen, J., and T.N. Palmer, 2001: A probability and decision-model
approximate technique for computing radiative transfer in analysis of a multimodel ensemble of climate change simulations. J.
inhomogeneous cloud fields. J. Geophysical Research, Climate, 14, 3212–3226.
doi:10.1209/2002JD0033222003.
Rajagopalan, B., U. Lall, and M.A. Cane, 1997: Anomalous ENSO oc-
Pinto, J.O., J.A. Curry, and J.M. Intrieri, 2001: Cloud-aerosol interac- currences: An alternate view. J. Climate, 10(9), 2351–2357.
tions during autumn over Beaufort Sea. J. Geophysical Research,
106, 15077–15097. Ramankutty, N., et al., 2002: The global distribution of cultivable lands:
Current patterns and sensitivity to possible climate change. Global
Pitman, A.J., et al., 1999: Uncertainty in the simulation of runoff due to Ecology Biogeography, 11, 377–392.
the parameterization of frozen soil moisture using the GSWP
methodology. J. Geophysical Research, 104, 16879–16888. Ramaswamy, V., et al., 2001: Radiative Forcing Climate Change, 349–
416, in Climate Change 2001: The Scientific Basis, ed. J.T.
Pitsch, H., 2006: Large-eddy simulation of turbulent combustion. An- Houghton, et al. Cambridge University Press, Cambridge, U.K.
nual Review Fluid Dynamics, 38, 453–482.
Randall, D.A., et al., 2007: Climate Models and Their Evaluation, Chap-
Plummer, D.A., et al., 2006: Climate and climate change over North ter 8, in Climate Change 2007. The Fourth Scientific Assessment, ed.
America as simulated by the Canadian Regional Climate Model. J. S. Solomon et al. Intergovernmental Panel on Climate Change
Climate, 19, 3112–3132. (IPCC) (www.ipcc.ch).
Polvani, L.M., R.K. Scott, and S.J. Thomas: 2004: Numerically con- Randall, D.A., et al., 2003: Breaking the cloud parameterization dead-
verged solutions of the global primitive equations for testing the dy- lock. Bulletin American Meteorological Society, 84, 1547–1564,
namical core of atmospheric GCMs. Monthly Weather Review, 132, doi:10.1175/BAMS-84-11-1547.
2539–2552.
Randall, D.A., et al., 2000: Cloud Feedbacks in Frontiers in Science:
Pope, V.D., et al., 2000: The impact of new physical parameterizations in Climate Modeling, ed. J.T. Kiehl and V. Ramanathan. Proceedings
the Hadley Centre Climate Model—HadAM3. Climate Dynamics, of a symposium in honor of Robert D. Cess.
16, 123–146.
Raper, S.C.B., J.M. Gregory, and R.J. Stouffer, 2002: The role of climate
Prentice, I.C., et al., 2001: The Carbon Cycle and Atmospheric Carbon sensitivity and ocean heat uptake on AOGCM transient temperature
Dioxide. Chapter 3 in Climate Change 2001: The Scientific Basis: response. J. Climate, 15, 124–130.
Contribution of Working Group 1 to the Third Assessment Report of
the Intergovernmental Panel on Climate Change, ed. J.T. Houghton et Raphael, M.N., and M.M. Holland, 2006: Twentieth Century simulation
al. Cambridge University Press, Cambridge, U.K., and New York, N.Y. of the Southern Hemisphere climate in coupled models. Part 1:
Large-scale circulation variability. Climate Dynamics, 26, 217–228.
Privé, N.C., and R.A. Plumb, 2007: Monsoon dynamics with interactive
forcing. Part I: Axisymmetric studies. J. Atmospheric Sciences, Rasch, P.J., and J.E. Kristjánsson, 1998: A comparison of the CCM3
64(5), 1417–1430. model climate using diagnosed and predicted condensate parame-
terizations. J. Climate, 11, 1587–1614.
Qian, J.-H., W.-K. Tao, and K.-M. Lau, 2004: Mechanisms for torrential
rain associated with the mei-yu development during SCSMEX 1998. Rasmussen, E.M., and J.M. Wallace, 1983: Meteorological aspects of
Monthly Weather Review, 132, 3–27. the El Niño/Southern Oscillation. Science, 222, 1195–2002.
Qian, J.-H., F. Giorgi, and M.S. Fox–Rabinovitz, 1999: Regional Rawlins, M.A., et al., 2003: Simulating pan-Arctic runoff with a macro-
stretched grid generation and its application to the NCAR RegCM. scale terrestrial water balance model. Hydrology Proceedings, 17,
J. Geophysical Research, 104(D6), 6501–6514. 2521–2539.
Qu, X., and A. Hall, 2006: Assessing snow albedo feedback in simulated
climate change. J. Climate, 19(11), 2617–2630.
115
Rayner, N.A., et al., 2003: Global analyses of sea surface temperature, Rosenzweig, C., et al., 2002: Increased crop damage in the U.S. from
sea ice, and night marine air temperature since the late Nineteenth excess precipitation under climate change. Global Environmental
Century. J. Geophysical Research, 108(D14), 4407. Change, 12, 197–202.
Reichler, T., and J. Kim, 2008: How well do coupled models simulate Rothstein, L.M., et al., 2006: Modeling ocean ecosystems: The PARA-
today's climate? Bulletin American Meteorological Society, 89(3), DIGM program. Oceanography, 19, 22–51.
doi:10.1175/BAMS-89-3-303, (www.met.utah.edu/reichler/publica-
tions/papers/Reichler_07_BAMS_CMIP.pdf). Rotstayn, L.D., and U. Lohmann, 2002: Tropical rainfall trends and the
indirect aerosol effect. J. Climate, 15, 2103–2116.
Rignot, E., and P. Kanagaratnam, 2006: Changes in the velocity structure
of the Greenland ice sheet. Science, 311, 986–990. Rotstayn, L.D., 1997: A physically based scheme for the treatment of
stratiform clouds and precipitation in large-scale models. Part I: De-
Ringer, M.A., and R.P. Allan, 2004: Evaluating climate model simula- scription and evaluation of microphysical processes. Quarterly J.
tions of tropical clouds. Tellus, 56A, 308–327. Royal Meteorological Society, 123, 1227–1282.
Rinke, A., et al., 2006: Evaluation of an ensemble of Arctic regional cli- Rowell, D.P., et al., 1992: Modelling the influence of global sea surface
mate models: Spatiotemporal fields during the SHEBA year. Cli- temperatures on the variability and predictability of seasonal Sahel
mate Dynamics, doi:10.1007/s00382-005-0095-3. rainfall. Geophysical Research Letters, 19, 905–908.
Roads, J., S.-C. Chen, and M. Kanamitsu, 2003: U.S. regional climate Ruiz-Barradas, A., and S. Nigam, 2006: IPCC’s Twentieth Century cli-
simulations and seasonal forecasts. J. Geophysical Research, mate simulations: Varied representations of North American hydro-
108(D16), Art. No. 8606. climate variability. J. Climate, 19, 4041–4058.
Roads, J.O., et al., 1999: Surface water characteristics in the NCEP Rummukainen, M., et al., 2004: The Swedish Regional Climate Mod-
Global Spectral Model and reanalysis. J. Geophysical Research, elling Programme, SWECLIM: A review. Ambio, 33, 176–182.
4(D16), 19307–19327.
Ruosteenoja, K., H. Tuomenvirta, and K. Jylha, 2007: GCM-based re-
Roberts, M.J., and R. Wood, 1997: Topographic sensitivity studies with a gional temperature and precipitation change estimates for Europe
Bryan-Cox-type ocean model. J. Physical Oceanography, 27, 823– under four SRES scenarios applying a super-ensemble pattern-scal-
836. ing method. Climatic Change, 81, Supplement 1,
doi:10.1007/s10584-006-9222-3.
Roe, G.H., and M.B. Baker, 2007: Why is climate sensitivity so unpre-
dictable? Science, 318(5850), 629–632, doi:10.1126/science.1144735. Russell, G.L., et al., 2000: Comparison of model and observed regional
temperature changes during the past 40 years. J. Geophysical Re-
Roeckner, E., et al., 2006: Sensitivity of simulated climate to horizontal search, 105, 14891–14898.
and vertical resolution in the ECHAM5 atmosphere model. J. Cli-
mate, 19, 3771–3791. Russell, G.L., J.R. Miller, and D. Rind, 1995. A coupled atmosphere-
ocean model for transient climate change studies. Atmosphere-
Roeckner, E., et al., 1996: The atmospheric general circulation model Ocean, 33(4), 683–730.
ECHAM-4: Model description and simulation of present-day cli-
mate. Report 128, Max-Planck-Institut für Meteorologie, Hamburg, Russell, J.L., R. Stouffer, and K.W. Dixon, 2007: Corrigendum. J. Cli-
Germany. mate, 20, 4287.
Roeckner, E., et al., 1987: Cloud optical depth feedbacks and climate Russell, J.L., R.J. Stouffer, and K.W. Dixon, 2006: Intercomparison of the
modelling. Nature, 329, 138–140. Southern Ocean circulations in IPCC coupled model control simu-
lations. J. Climate, 19, 4560–4575.
Root, T.L., and S.H. Schneider, 1993: Can large-scale climatic models be
linked with multiscale ecological studies? Conservation Biology, Ryan, B.F., et al., 2000: Simulations of a cold front by cloud-resolving,
7(2), 256–270. limited-area, and large-scale models, and a model evaluation using
in situ and satellite observations. Monthly Weather Review, 128,
Ropelewski, C.F., and M.S. Halpert, 1987: Global and regional scale pre- 3218–3235.
cipitation patterns associated with the El Niño Southern Oscillation.
116
Saji, N.H., S.-P. Xie, and T. Yamagata, 2005: Tropical Indian Ocean vari- Schneider, S.H., and C. Mass, 1975: Volcanic dust, sunspots, and tem-
ability in the IPCC Twentieth Century climate simulations. J. Cli- perature trends. Science, 190, 741–746.
mate, 19(17), 4397.
Schopf, P., et al., 2003: Coupling Process and Model Studies of Ocean
Saji, N.H., et al., 1999: A dipole mode in the tropical Indian Ocean. Na- Mixing to Improve Climate Models—A Pilot Climate Process Mod-
ture, 401, 360–363. eling and Science Team (a U.S. CLIVAR paper,
www.usclivar.org/CPT/Ocean_mixing_whitepaper.pdf).
Santer, B.D., et al., 2007: Identification of human-induced changes in at-
mospheric moisture content. Proceedings National Academy Sci- Schramm, J.L., et al., 1997: Modeling the thermodynamics of a sea ice
ences, 104, 15248–15253. thickness distribution. Part 1: Sensitivity to ice thickness resolution.
J. Geophysical Research, 102, 23079–23091.
Santer, B.D., et al., 2005: Amplification of surface temperature trends
and variability in the tropical atmosphere. Science, 309, 1551–1556. Schwartz, S.E., 2007: Heat capacity, time constant, and sensitivity of
Earth's climate system. J. Geophysical Research, 112, D24S05,
Santer, B.D., et al., 2001: Accounting for the effects of volcanoes and doi:10.1029/2007JD008746, 2007
ENSO in comparisons of modeled and observed temperature trends.
J. Geophysical Research, 106, 28033–28059. Schweitzer, L., 2006: Environmental justice and hazmat transport: A
spatial analysis in southern California. Transportation Research. Part
Santer, B.D., et al., 1990: Developing Climate Scenarios from Equilib- D–Transportation Environment, 11, 408–421.
rium GCM Results, Report No. 47, Max Planck Institute for Mete-
orology, Hamburg. Segal, M., et al., 1997: Small lake daytime breezes: Some observational
and conceptual evaluations. Bulletin American Meteorological Soci-
Saraf, A.K., et al., 2005: Digital Elevation Model (DEM) generation ety, 78, 1135–1147.
from NOAA-AVHRR night-time data and its comparison with
USGS-DEM. International J. Remote Sensing, 26, 3879–3887. Segal, M., and R.W. Arritt, 1992: Nonclassical mesoscale circulations
caused by surface sensible heat-flux gradients. Bulletin American
Sardeshmukh, P.D., and B.J. Hoskins, 1988: The generation of global Meteorological Society, 73, 1593–1604.
rotational flow by steady idealized tropical divergence. J. Atmos-
pheric Sciences, 45, 1228–1251. Sellers, P.J., et al., 1996: A revised land surface parameterization (SiB2)
for atmospheric GCMs. Part 1: Model formulation. J. Climate, 9,
Sato, M., et al., 1993: Stratospheric aerosol optical depth, 1850–1990. J. 676–705.
Geophysical Research, 98, 22987–22994.
Sellers, P.J., et al., 1986: A simple biosphere model (SiB) for use within
Sausen, R., S. Schubert, and L. Dumenil, 1994: A model of the river general-circulation models. J. Atmospheric Sciences, 43, 503–531.
runoff for use in coupled atmosphere-ocean models. J. Hydrology,
155, 337–352. Semtner, A.J., 1976: A model for the thermodynamic growth of sea ice
in numerical investigations of climate. J. Physical Oceanography, 6,
Schimel, D.S., 1998: The carbon equation. Nature, 393, 208–209. 27–37.
Schmidt, G.A., et al., 2006: Present day atmospheric simulations using Seneviratne, S.I., et al., 2006: Land-atmosphere coupling and climate
GISS ModelE: Comparison to in-situ, satellite, and reanalysis data. change in Europe. Nature, 443, 205–209.
J. Climate, 19, 153–192, doi:10.1175/JCLI3612.1.
Senior, C.A., and J.F.B. Mitchell, 1996: Cloud Feedbacks in the Unified
Schmittner, A., M. Latif, and B. Schneider, 2005: Model projections of UKMO GCM, in Climate Sensitivity to Radiative Perturbations,
the North Atlantic thermohaline circulation for the 21st Century as- Physical Mechanism and Their Validation, ed. H. Le Treut. Springer,
sessed by observations. Geophysical Research Letters, 32, 331 pp.
doi:10.1029/2005GL024368.
Senior, C.A., and J.F.B. Mitchell, 1993: Carbon dioxide and climate: The
Schneider, S.H., and S.L. Thompson, 1981: Atmospheric CO2 and cli- impact of cloud parameterization. J. Climate, 6, 5–21.
mate: Importance of the transient response. J. Geophysical Research,
86, 3135–3147. Seth, A., and F. Giorgi, 1998: The effects of domain choice on summer
precipitation simulation and sensitivity in a regional climate model.
J. Climate, 11, 2698–2712.
117
Shaw, M.R., et al., 2002: Grassland responses to global environmental Soden, B.J., and I.M. Held, 2006: An assessment: Climate feedbacks in
changes suppressed by elevated CO2. Science, 298, 1987–1990. coupled ocean–atmosphere models. J. Climate, 19, 3354–3360.
Shepherd, A., and D. Wingham, 2007: Recent sea-level contributions of Soden, B.J., et al., 2005: The radiative signature of upper tropospheric mois-
the Antarctica and Greenland ice sheets. Science, 315, 1529–1532. tening. Science, 310(5749), 841–844, doi:10.1126/science.1115602.
Shindell, D.T., et al., 2006: Solar and anthropogenic forcing of tropical Soden, B.J., A.J. Broccoli, and R.S. Hemler, 2004: On the use of cloud
hydrology. Geophysical Research Letters, 33, L24706, forcing to estimate cloud feedback. J. Climate, 17, 3661–3665.
doi:10.1029/2006GL027468.
Soden, B.J., et al., 2002: Global cooling after the eruption of Mount Pinatubo:
Shindell, D.T., et al., 1999: Simulation of recent northern winter climate A test of climate feedback by water vapor. Science, 296, 727–730.
trends by greenhouse-gas forcing. Nature, 399, 452–455.
Soden, B.J., 2000: The diurnal cycle of convection, clouds, and water
Shukla, J., et al., 2006: Climate model fidelity and projections of cli- vapor in the tropical upper troposphere. Geophysical Research Let-
mate change. Geophysical Research Letters, 33, L07702, ters, 27, 2173–2176.
doi:10.1029/2005GL025579.
Soden, B.J., and I.M. Held, 2000: An assessment of climate feedbacks in
Siddall, M., et al., 2003: Sea-level fluctuations during the last glacial coupled ocean–atmosphere models. J. Climate, 19, 3354–3360.
cycle. Nature, 423, 853–858.
Sokolov, A.P., and P.H. Stone, 1998: A flexible climate model for use in
Skamarock, W.C., et al., 2005: A Description of the Advanced Research integrated assessments. Climate Dynamics, 14, 291–303.
WRF Version 2. NCAR Technical Note. NCAR/TN–468+STR, Na-
tional Center for Atmospheric Research, Boulder, Colo., 88 pp. Solman, S.A., M.N. Nunez, and P.R. Rowntree, 2003: On the evaluation
[Available from NCAR; P.O. Box 3000; Boulder, CO 80305] of the representation of mid-latitude transients in the Southern Hemi-
sphere by HadAM2B GCM and the impact of horizontal resolution.
Slater, A.G., et al., 2001: The representation of snow in land surface Atmosfera, 16, 245–272.
schemes: Results from PILPS 2(d). J. Hydrometeorology, 2, 7–25.
Somerville, R.C.J., and L.A. Remer, 1984: Cloud optical thickness feed-
Small, C., V. Gornitz, and J.E. Cohen, 2000. Coastal hazards and the backs in the CO2 climate problem. J. Geophysical Research, 89,
global distribution of population. Environmental Geoscience, 7, 3– 9668–9672.
12.
Spencer, R.W., et al., 2007: Cloud and radiation budget changes associ-
Smethie, W.M., Jr., and R.A. Fine, 2001: Rates of North Atlantic deep ated with tropical intraseasonal oscillations. Geophysical Research
water formation calculated from chlorofluorocarbon inventories. Letters, 34, L15707, doi:10.1029/2007GL029698.
Deep Sea Research, Part 1: Oceanographic Research Papers, 48,
189– 215. Stainforth, D.A., et al., 2005: Uncertainty in predictions of the climate
response to rising levels of greenhouse gases. Nature, 443, 403–406.
SMIC, 1971: Inadvertent Climate Modification: Report of the Study of
Man’s Impact on Climate, Massachusetts Institute of Technology Stephens, G.L., 2005: Cloud feedbacks in the climate system: A critical
Press, Cambridge, Mass., 308 pp. review. J. Climate, 18, 237–273.
Smith, R.D., and P.R. Gent, 2004: Reference Manual for the Parallel Stewart, I.T., D.R. Cayan, and M.D. Dettinger, 2004: Changes in
Ocean Program (POP), Ocean Component of the Community Cli- snowmelt runoff timing in western North America under a “business
mate System Model (CCSM2.0 and 3.0). Technical Report LA-UR- as usual” climate change scenario. Climatic Change, 62, 217–232.
02-2484, Los Alamos National Laboratory, Los Alamos, New
Mexico (www.ccsm.ucar.edu/models/ccsm3.0/pop). Stott, P.A., and C.E. Forest, 2007: Ensemble climate predictions using
climate models and observational constraints. Philosophical Trans-
Smith, R.N.B., 1990: A scheme for predicting layer clouds and their actions Royal Society A: Mathematical, Physical, and Engineering
water content in a general circulation model. Quarterly J. Royal Me- Sciences, 365(1857), 2029–2052, doi: 10.1098/rsta.2007.2075.
teorological Society, 116, 435–460.
Stott, P.A., et al., 2006: Observational constraints on past attributable
Smith, S.J., et al., 2005: Climate change impacts for the conterminous warming and predictions of future global warming. J. Climate,
USA, Part 1: Scenarios and context. Climatic Change, 69, 7–25. 19(13), 3055–3069, doi:10.1175/JCLI3802.1.
118
Stott, P.A., G.S. Jones, and J.F.B. Mitchell, 2003: Do models underesti- Takle, E.S., et al., 1999: Project to Intercompare Regional Climate Sim-
mate the solar contribution to recent climate change? J. Climate, 16, ulations (PIRCS): Description and initial results. J. Geophysical Re-
4079–4093, doi:10.1175/1520-0442(2003)016!4079: search, 104(D16), 19443–19461.
DMUTSCO2.0.CO;2.
Talley, L.D., J.L. Raid, and P.E. Robbins, 2003: Data-based meridional
Stouffer, R.J., et al., 2006: GFDL’s CM2 global coupled climate models. overturning stream functions for the global ocean. J. Climate, 16,
Part IV: Idealized climate response. J. Climate, 19, 723–740. 3213–3226.
Stouffer, R.J., J. Russell, and M.J. Spelman, 2006: Importance of oceanic Tao, W.-K., 2007: Cloud-resolving modeling. J. Meteorological Society
heat uptake in transient climate change. Geophysical Research Let- Japan. 125th Anniversary Special Issue, 85B, 305-330.
ters, 33(17), L17704, doi:10.1029/2006GL027242.
Tao, W.-K., et al., 2004: Atmospheric energy budget and large-scale pre-
Stouffer, R.J., G. Hegerl, and S. Tett, 2000: A comparison of surface air cipitation efficiency of convective systems during TOGA COARE,
temperature variability in three 1000-yr coupled ocean-atmosphere GATE, SCSMEX and ARM: Cloud-resolving model simulations. J.
model integrations. J. Climate, 13, 513–537. Atmospheric Sciences, 61, 2405–2423.
Strack, J.E., R.A. Pielke, and J. Adegoke, 2003: Sensitivity of model- Tao, W.-K., 2003: Goddard Cumulus Ensemble (GCE) model: Applica-
generated daytime surface heat fluxes over snow to land-cover tion for understanding precipitation processes. Cloud systems, hur-
changes. J. Hydrometeorology, 4, 24–42. ricanes, and the Tropical Rainfall Measuring Mission (TRMM): A
Tribute to Dr. Joanne Simpson, Meteorological Monograph. Bulletin
Stratton, R.A., 1999: A high resolution AMIP integration using the American Meteorological Society, 51, 107–138.
Hadley Centre model HadAM2b. Climate Dynamics, 15, 9–28.
Tao, W.-K., et al., 1999: On equilibrium states simulated by cloud-re-
Sturm, M., et al., 2005: Changing snow and shrub conditions affect solving models. J. Atmospheric Sciences, 56, 3128–3139.
albedo with global implications. J. Geophysical Research–Biogeo-
sciences, 110, Art. No. G01004. Tebaldi, C., et al., 2006: Going to the extremes: An intercomparison of
model-simulated historical and future changes in extreme events.
Sturm, M., et al., 2001: Snow-shrub interactions in Arctic tundra: A hy- Climate Change 79(3–4), 185–211.
pothesis with climatic implications. J. Climate, 14, 336–344.
Tebaldi, C., et al., 2005: Quantifying uncertainty in projections of re-
Sud, Y.C., and G.K. Walker, 1999: Microphysics of clouds with the Re- gional climate change: A Bayesian approach to the analysis of mul-
laxed Arakawa–Schubert Scheme (McRAS). Part II: Implementa- timodel ensembles. J. Climate, 18, 1524–1540.
tion and performance in GEOS II GCM. J. Atmospheric Sciences,
56(18), 3221–3240. Tenhunen, J.D., et al., 1999: Ecosystem Studies, Land Use, and Resource
Management, 1–19 in Integrating Hydrology, Ecosystem Dynamics,
Sui, C.-H., X. Li, and K.-M. Lau, 1998: Radiative-convective processes and Biochemistry in Complex Landscapes, ed. J.D. Tenhunen and P.
in simulated diurnal variations of tropical oceanic convection. J. At- Kabat. Wiley, Chichester.
mospheric Sciences, 55, 2345–2359.
Thompson, D.W.J., and S. Solomon, 2002: Interpretation of recent
Sun, S., and J. Hansen, 2003: Climate simulations for 1951–2050 with a Southern Hemisphere climate change. Science, 296, 895–899,
coupled atmosphere-ocean model. J. Climate, 16, 2807–2826, doi:10.1126/ science.1069270.
doi:10.1175/1520–0442.
Thompson, D.W.J., and J.M. Wallace, 2000: Annual modes in the ex-
Sun, S., and R. Bleck, 2001: Atlantic thermohaline circulation and its re- tratopical circulation. Part I: Month-to-month variability. J. Climate,
sponse to increasing CO2 in a coupled atmosphere-ocean model. 13, 1000–1016.
Geophysical Research Letters, 28, 4223–4226.
Thompson, D.W.J., and J.M. Wallace, 1998: The Arctic Oscillation sig-
Sun, Y., et al., 2006: How often does it rain? J. Climate, 19, 916–934. nature in the wintertime geopotential height and temperature fields.
Geophysical Research Letters, 25, 1297–1300.
Svensmark, H., 2007: Cosmoclimatology: A new theory emerges. As-
tronomy Geophysics, 48, 118124, doi:10.1111/j.1468-4004.2007. Thompson, S., et al., 2004: Quantifying the effects of CO2-fertilized
48118.x. vegetation on future global climate and carbon dynamics. Geophys-
ical Research Letters, 31(23), L23211.
119
Thomson, A.M., et al., 2005: Climate change impacts for the contermi- Tripoli, G.J., and W.R. Cotton, 1989: Numerical study of an observed
nous USA, Part 3: Dryland production of grain and forage crops. orogenic mesoscale convective system. Part 2: Analysis of govern-
Climatic Change, 69, 43–65. ing dynamics. Monthly Weather Review, 117, 305–328.
Thorne, P.W., et al., 2007: Tropical vertical temperature trends: A real Troccoli, A., and T.N. Palmer, 2007: Ensemble decadal predictions from
discrepancy? Geophysical Research Letters, 34, L16702, analysed initial conditions. Philosophical Transactions Royal Soci-
doi:10.1029/2007GL029875. ety A, 365 (No. 1857).
Tiedtke, M., 1993: Representation of clouds in large-scale models. Tselioudis, G., and C. Jakob, 2002: Evaluation of midlatitude cloud prop-
Monthly Weather Review, 121, 3040–3061. erties in a weather and a climate model: Dependence on dynamic
regime and spatial resolution. J. Geophysical Research, 107, 4781.
Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus pa-
rameterization in large scale models. Monthly Weather Review, 117, Tselioudis, G., Y.-C. Zhang, and W.R. Rossow, 2000: Cloud and radia-
1779–1800. tion variations associated with northern midlatitude low and high sea
level pressure regimes. J. Climate, 13, 312–327, doi:10.1175/1520-
Tjernström, M., et al., 2005: Modelling the Arctic boundary layer: An 0442(2000).
evaluation of six ARCMIP regional-scale models with data from the
SHEBA project. Boundary-Layer Meteorology, 117, 337–381. Tsushima, Y., A. Abe-Ouchi, and S. Manabe, 2005: Radiative damping
of annual variation in global mean surface temperature: Compari-
Tjernström, M., M. Zagar, and G. Svensson, 2004: Model simulations son between observed and simulated feedback. Climate Dynamics,
of the Arctic atmospheric boundary layer from the SHEBA year. 24, 591–597.
Ambio, 33, 221–227.
Twomey, S., 1977: The influence of pollution on the short wave albedo
Tjoelker, M.G., J. Oleksyn, and P.B. Reich, 2001: Modelling respiration of clouds. J. Atmospheric Sciences, 34, 1149–1152.
of vegetation: Evidence for a general temperature-dependent Q10.
Global Change Biology, 7(2), 223–230. Ueda, H., et al., 2006: Impact of anthropogenic forcing on the Asian sum-
mer monsoon as simulated by eight GCMs. Geophysical Research
Tompkins, A., 2002: A prognostic parameterization for the subgrid-scale Letters, 33, doi:10.1029/2005GL025336.
variability of water vapor and clouds in large-scale models and its use
to diagnose cloud cover. J. Atmospheric Sciences, 59, 1917–1942. Uotila, P., et al., 2007: Changes in Antarctic net precipitation in the 21st
Century based on IPCC model scenarios. J. Geophysical Research,
Trenberth, K.E., J. Fasullo, and L. Smith, 2005: Trends and variability 112, D10107, doi:10.1029/2006JD007482.
in column-integrated atmospheric water vapor. Climate Dynamics,
24 (7–8), 741–758, doi:10.1007/s00382-005-0017-4. Uppala, S.M., et al., 2005: The ERA-40 reanalysis. Quarterly J. Royal
Meteorological Society, 131(612), 2961–3012, doi:10.1256/
Trenberth, K.E., et al., 1998: Progress during TOGA in understanding qj.04.176. Index is at www.ecmwf.int/research/era/ ERA-40_Atlas/
and modeling global teleconnections associated with tropical sea docs/index.html.
surface temperatures. J. Geophysical Research, 103 (special TOGA
issue), 14291–14324. van Oldenborgh, G.J., S.Y. Philip, and M. Collins, 2005: El Niño in a
changing climate: A multi-model study. Ocean Science, 1, 81–95.
Trenberth, K.E., and T.J. Hoar, 1997: El Niño and climate change. Geo-
physical Research Letters, 24, 3057–3060. van Ulden, A.P., and G.J. van Oldenborgh, 2006: Large-scale atmos-
pheric circulation biases and changes in global climate model sim-
Trenberth, K.E., and J. Hurrell, 1994: Decadal atmosphere-ocean vari- ulations and their importance for climate change in Central Europe.
ations in the Pacific. Climate Dynamics, 9, 303–319. Atmospheric Chemistry Physics, 6(4), 863–881
Trier, S. B., et al., 1996: Structure and evolution of the 22 February 1993 VanRheenen, N.T., et al., 2004: Potential implications of PCM climate
TOGA COARE squall line: Numerical simulations. J. Atmospheric change scenarios for Sacramento–San Joaquin River basin hydrology
Sciences, 53, 2861–2886. and water resources. Climatic Change, 62, 257–281.
Tripoli, G.J., 1992: A nonhydrostatic mesoscale model designed to sim- Vavrus, S., et al., 2006: The behavior of extreme cold air outbreaks under
ulate scale interaction. Monthly Weather Review, 120, 1342–1359. greenhouse warming. International J. Climatology, 26, 1133–1147.
120
Velicogna, I., and J. Wahr, 2006: Acceleration of Greenland ice mass loss Webb, M.J., et al., 2001: Combining ERBE and ISCCP data to assess
in spring 2004. Nature, 443(7109), 329–331. clouds in the Hadley Centre, ECMWF and LMD atmospheric cli-
mate models. Climate Dynamics, 17, 905–922.
Vidale, P.L., et al., 2003: Predictability and uncertainty in a regional cli-
mate model. J. Geophysical Research, 108(D18), 4586. Webster, P.J., et al., 1998: Monsoons: Processes, predictability, and the
prospects for prediction. J. Geophysical Research, 103(C7), 14451–
Vinnikov, K.Y., D.J. Cavalieri, and C.L. Parkinson, 2006: A model as- 14510, doi:10.1029/97JC02719.
sessment of satellite observed trends in polar sea ice extents. Geo-
physical Research Letters, 33, L05704. Wei, H., et al., 2002: Calibration and validation of a regional climate
model for pan-Arctic hydrologic simulation. J. Climate, 15, 3222–
Vitart, F., and J.L. Anderson, 2001: Sensitivity of Atlantic tropical storm 3236.
frequency to ENSO and interdecadal variability of SSTs in an en-
semble of AGCM integrations. J. Climate, 14(4), 533–545. Wentz, F.J., et al., 2007: How much more rain will global warming bring?
Science, 317, 233–235.
Völker, C., D.W.R. Wallace, and D.A. Wolf-Gladrow, 2002: On the role
of heat fluxes in the uptake of anthropogenic carbon in the North At- Wetherald, R.T., and S. Manabe, 1988: Cloud feedback processes in
lantic. Global Biogeochemical Cycles, 16(4), 1138, doi:10.1029/ general circulation models. J. Atmospheric Sciences, 45, 1397–1415.
2002GB001897.
Whitman, S., et al., 1997: Mortality in Chicago attributed to the July
Von Storch, H., E. Zorita, and U. Cubasch, 1993: Downscaling of global 1995 heat wave. American J. Public Health, 87, 1515–1518.
climate change estimates to regional scales: An application to Iber-
ian rainfall in wintertime. J. Climate, 6, 1161–1171. Widman, M., C.S. Bretherton, and E.P. Salathe, Jr., 2003: Statistical pre-
cipitation downscaling over the Northwestern United States using
Wang, B., 1995: Interdecadal changes in El Niño onset in the last four numerically simulated precipitation as a predictor. J. Climate, 16,
decades. J. Climate, 8, 267–284. 799–816.
Wang, C., 2005: A modeling study of the response of tropical deep con- Wigley, T.M.L., et al., 2005: Effect of climate sensitivity on the response
vection to the increase of cloud condensation nuclei concentration: 1. to volcanic forcing. J. Geophysical Research, 110, D09107.1–
Dynamics and microphysics. J. Geophysical Research, 110, D21211. D09107.8, doi:10.1029/2004JD005557.
Wang, H., and K.-M. Lau, 2006: Atmospheric hydrological cycle in the Wigley, T.M.L., and M.E. Schlesinger, 1985: Analytical solution for the effect
tropics in Twentieth Century coupled climate simulations. J. Climate, of increasing CO2 on global mean temperature. Nature, 315, 649–652.
26, 655–678.
Wilby, R.L., et al., 2004: Guidelines for Use of Climate Scenarios De-
Wang, M., et al., 2007: Intrinsic versus forced variation in coupled cli- veloped from Statistical Downscaling Methods. IPCC Data Distri-
mate model simulations over the Arctic during the Twentieth Century. bution Centre, University of East Anglia, U.K., 27 pp.
J. Climate, 20(6), 1093–1107. (ipcc-ddc.cru.uea.ac.uk/guidelines).
Wang, W., and M.E. Schlesinger, 1999: The dependence on convection Wilby, R.L., R. Dawson, and E.M. Barrow, 2002: SDSM: A decision
parameterization of the tropical intraseasonal oscillation simulated support tool for the assessment of regional climate change assess-
by the UIUC 11-layer atmospheric GCM. J. Climate, 12(5), 1423– ments. Environmental Modelling Software, 17, 145–157.
1457.
Wilby, R.L., et al., 2000: Hydrological responses to dynamically and sta-
Wang, Y., et al., 2004: Regional climate modeling: Progress, challenges, tistically downscaled general circulation model output. Geophysical
and prospects. J. Meteorological Society Japan, 82, 1599–1628. Research Letters, 27, 1199–1202.
Warrach, K., H.T. Mengelkamp, and E. Raschke, 2001: Treatment of Wilby, R.L., et al., 1998: Statistical downscaling of general circulation
frozen soil and snow cover in the land surface model SEWAB. model output: A comparison of methods. Water Resources Research,
Theoretical Applied Climatology, 69, 23–37. 34, 2995–3008.
Webb, M.J., et al., 2006: On the contribution of local feedback mecha- Wilby, R.L., and T.M.L. Wigley, 1997: Downscaling general circulation
nisms to the range of climate sensitivity in two GCM ensembles. model output: A review of methods and limitations. Progress Phys-
Climate Dynamics, 27(1), 17–38, doi:10.1007/s00382-006-0111-2. ical Geography, 21, 530–548.
121
Wilks, D.S., and R.L. Wilby, 1999: The weather generation game: A re- Wu, X., and M.W. Moncrieff, 2001: Long-term behavior of cloud sys-
view of stochastic weather models. Progress Physical Geography, tems in TOGA COARE and their interactions with radiative and sur-
23, 329–357. face processes. Part III: Effects on the energy budget and SST. J.
Atmospheric Sciences, 58, 1155–1168.
Williams, K.D., et al., 2006: Evaluation of a component of the cloud re-
sponse to climate change in an intercomparison of climate models. Wyant, M.C., et al., 2006: A comparison of tropical cloud properties and
Climate Dynamics, 145, 145–165. responses in GCMs using mid-tropospheric vertical velocity. Cli-
mate Dynamics, 27, 261–279.
Williams, K.D., M.A. Ringer, and C.A. Senior, 2003: Evaluating the
cloud response to climate change and current climate variability. Cli- Wyant, M.C., M. Khairoutdinov, and C.S. Bretherton, 2006: Climate
mate Dynamics, 20, 705–721. sensitivity and cloud response of a GCM with a superparameteriza-
tion. Geophysical Research Letters, 33, L06714.
Willson, R.C., and A.V. Mordvinov, 2003: Secular total solar irradiance
trend during solar cycles 21–23. Geophysical Research Letters, Wyrtki, K., 1975: El Niño—The dynamic response of the equatorial Pa-
30(5), 1199–1202, doi:10.1029/2002GL016038, 2003. cific Ocean to atmospheric forcing. J. Physical Oceanography, 5,
572–584.
Wilson, D.R., and S.P. Ballard, 1999: A microphysics-based precipita-
tion scheme for the U.K. Meteorological Office numerical weather Xie, P., and P.A. Arkin, 1997: Global precipitation: A 17-year monthly
prediction model. International J. Climatology, 125, 1607–1636. analysis based on gauge observations, satellite estimates, and nu-
merical model outputs. Bulletin American Meteorological Society
Wilson, M.F., et al., 1987: Sensitivity of the Biosphere Atmosphere 78, 2539–2558.
Transfer Scheme (BATS) to the inclusion of variable soil character-
istics. J. Climate Applied Meteorology, 26, 341–362. Xie, S., et al., 2005: Simulations of midlatitude frontal clouds by single-
column and cloud-resolving models during the Atmospheric Radia-
Wilson, T.B., et al., 2003: Evaluation of the importance of Lagrangian tion Measurement March 2000 cloud intensive operational period. J.
canopy turbulence formulations in a soil-plant-atmosphere model. Geophysical Research, 110, D15S03.
Agricultural Forest Meteorology, 115, 51–69.
Xie, S., et al., 2004: Impact of a revised convective triggering mecha-
Winkler, J.A., et al., 2002: Possible impacts of projected temperature nism on Community Atmosphere Model, Version 2, simulations: Re-
change on commercial fruit production in the Great Lakes Region. sults from short-range weather forecasts. J. Geophysical Research,
J. Great Lakes Research, 28, 608–625. 109, D14102, doi:10.1029/2004JD004692.
Winton, M., 2000: A reformulated three-layer sea ice model. J. Atmos- Xu, K.-M., et al., 2005: Modeling springtime shallow frontal clouds with
pheric Oceanic Technology, 17, 525–531. cloud-resolving and single-column models. J. Geophysical Research,
110, D15S04, doi:10.1029/2004JD005153.
Wittenberg, A.T., et al., 2006: GFDL’s CM2 global coupled climate mod-
els. Part III: Tropical Pacific climate and ENSO. J. Climate, 19, Xu, K.-M., and D.A. Randall, 1998: Influence of large-scale advective
698–722. cooling and moistening effects on the quasi-equilibrium behavior of
explicitly simulated cumulus ensembles. J. Atmospheric Sciences,
Wood, A.W., et al., 2004: Hydrological implications of dynamical and 55, 896–909.
statistical approaches to downscaling climate model outputs. Cli-
mate Change, 62, 189–216. Xue, Y., et al., 2001: The impact of land surface processes on simula-
tions of the U.S. hydrological cycle: A case study of the 1993 flood
Woodward, F.I., and M.R. Lomas, 2004: Vegetation dynamics—simu- using the SSiB land surface model in the NCEP ETA regional model.
lating responses to climatic changes. Biological Reviews 79, 643– Monthly Weather Review, 129(12), 2833–2860.
670.
Yamaguchi, K., A. Noda, and A. Kitoh, 2005: The changes in permafrost
Wu, X., et al., 2007: Coupling of convective momentum transport with induced by greenhouse warming: A numerical study applying mul-
convective heating in global climate simulations. J. Atmospheric Sci- tiple-layer ground model. J. Meteorological Society Japan, 83, 799–
ences, 64(4), 1334–1349. 815.
Yang, Z.W., and R. Arritt, 2002: Tests of a perturbed physics ensemble

approach for regional climate modeling. J. Climate, 15, 2881–2896.
122
Yao, M.-S., and A.D. Del Genio, 2002: Effects of cloud parameteriza- Zhang, M.H., et al., 2001: Objective analysis of the ARM IOP data:
tion on the simulation climate changes in the GISS GCM. Part II: Sea Method and sensitivity. Monthly Weather Review, 129, 295–311.
surface temperature and cloud feedbacks. J. Climate, 15, 2491–2503.
Zhang, M.H., et al., 1994: Diagnostic study of climate feedback
Yeh, P.J.-F., and E.A.B. Eltahir, 2005: Representation of water table dy- processes in atmospheric general circulation models. J. Geophysical
namics in a land surface scheme. Part 1: Model development. J. Cli- Research, 99, 5525–5537.
mate, 18, 1861–1880.
Zhang, R., and T.L. Delworth, 2006: Impact of Atlantic multidecadal os-
Yokohata, T, et al., 2005: Climate response to volcanic forcing: Valida- cillations on India/Sahel rainfall and Atlantic hurricanes. Geophys-
tion of climate sensitivity of a coupled atmosphere-ocean general ical Research Letters, 33, L17712, doi:10.1029/2006GL026267.
circulation model. Geophysical Research Letters, 32(21), L21710.1–
L21710.4. Zhang, X., and J.E. Walsh, 2006: Toward a seasonally ice-covered Arc-
tic Ocean: Scenarios from the IPCC AR4 model simulations. J. Cli-
York, J.P., et al., 2002: Putting aquifers into atmospheric simulation mod- mate, 19, 1730–1747.
els: An example from the Mill Creek Watershed, northeastern
Kansas. Advances Water Resources, 25, 221–238. Zhang, X.-C., 2005: Spatial downscaling of global climate model output
for site-specific assessment of crop production and soil erosion.
Yu, H., et al., 2006: A review of measurement-based assessment of Agricultural Forest Meteorology, 135, 215–229.
aerosol direct radiative effect and forcing. Atmospheric Chemistry
Physics, 6, 613–666. Zhang, Y.C., A.N. Huang, and X.S. Zhu, 2006: Parameterization of the
thermal impacts of sub-grid orography on numerical modeling of
Yu, X., and M.J. McPhaden, 1999: Seasonal variability in the equatorial the surface energy budget over East Asia. Theoretical Applied Cli-
Pacific. J. Physical Oceanography, 29, 925–947. matology, 86, 201–214.
Zebiak, S.E., and M.A. Cane, 1987: A model El Niño-Southern Oscil- Zhu, J., and X.-Z. Liang, 2007: Regional climate model simulations of
lation. Monthly Weather Review, 115, 2262–2278. U.S. precipitation and surface air temperature during 1982–2002: In-
terannual variation. J. Climate, 20(2), 218–232.
Zhang, C., et al., 2006: Simulations of the Madden-Julian Oscillation in
four pairs of coupled and uncoupled global models. Climate Dy- Zorita, E., and H. von Storch, 1999: The analog method as a simple sta-
namics 27(6), 573–592. tistical downscaling technique: Comparison with more complicated
methods. J. Climate, 12, 2474–2489.
Zhang, D., and M.J. McPhaden, 2006: Decadal variability of the shallow
Pacific meridional overturning circulation: Relation to tropical sea
surface temperatures in observations and climate change models.
Ocean Modelling, 15, 250–273.
Zhang, G.J., and N.A. McFarlane, 1995: Sensitivity climate simulations

to the parameterization of cumulus convection in the Canadian Cli-
mate Centre general circulation model. Atmosphere–Ocean, 33,
407–446.
Zhang, J., and D. Rothrock, 2000: Modeling Arctic sea ice with an effi-
cient plastic solution, J. Geophysical Research, 105, 3325–3338.
Zhang, M.H., et al., 2005: Comparing clouds and their seasonal varia-
tions in 10 atmospheric general circulation models with satellite
measurements. J. Geophysical Research, 110, D15S02, doi:10.1029/
2004JD005021.
Zhang, M., 2004: Cloud-climate feedback: How much do we know? In

Observation, Theory, and Modeling of Atmospheric Variability.
World Scientific Series on Meteorology of East Asia, Vol. 3, ed. Zhu
et al. World Scientific Publishing Co., Singapore, 632 pp.
123
The U.S. Climate Change Science Program
124
U.S. Climate Change Science Program
1717 Pennsylvania Avenue, N W • Suite 250 • W ashington, D.C . 20006 USA
1-202-223-6262 (voice) • 1-202-223-3065 (fax)
http://www.climatescience.gov

Climate Models An Assessment of Strengths and Limitations

Uploaded by

Copyright:

Available Formats

Climate Models An Assessment of Strengths and Limitations

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Climate Models An Assessment of Strengths and Limitations

Uploaded by

Copyright:

Available Formats

Climate Models

U.S. Climate Change Science Program

Director, U.S. Climate Change Science Program...........................................................William J. Brennan

Director, Climate Change Science Program Office .......................................................Peter A. Schultz

Lead Agency Principal Representative to CCSP;

Product Lead; Department of Energy,

Synthesis and Assessment Product Advisory Group Chair;

Synthesis and Assessment Product Coordinator,

OTH ER AGEN CY REPRESEN TATIVES

National Aeronautics and Space Administration ...........................................................Donald Anderson

Synthesis and Assessment Product 3.1

ED ITO RIAL TEAM

Recommended Citation for the entire report

TABLE O F CO N TEN TS Executive Summary............................................................................................1

Scientists extensively use mathematical models of Earth’s climate, executed

still needed for full scientific evaluation of the

cycle in solar irradiance; paleoclimatic infor-

Atmospheric Atmospheric Atmospheric

1.2 CLIMAT E MODEL processes and their interactions. These mathe-

and their use in “downscaling” global model re-

Figure 2.2. Schematic

Figure 2.3. Schematic

2.5 COMPON EN T COU PLIN G 2.5.1 N OAA GFDL Model-

Tuning the Global Mean Energy Balance

Figure 2.4. A Snapshot

Figure 2.5. Model Metrics for 23 Different Climate Fields.

Time-dependent climate-forcing simulations Time-evolving climate forcing also is used as

3.1 T YPES OF DOW N SCALIN G SIMU LAT ION S

The response of climate to a perturbation such as a change in carbon dioxide concentration, or

more detailed treatments of clouds and other Table 4.1

exploration of model input parameters in one

mation, particularly from the peak of the last Ice

4.1.2 Observational Constraints on Some studies have argued that observations of

Positive water-vapor feedback, resulting from 4.3.1 Greenhouse Gases

one AOGCM simulation using one idealized

Although it introduces its own uncertainties, the

5.1 MEAN SU RFACE ing” of the models’ energy balance as described

CM2.0 minus observed CM2.1 minus observed

–7 –5 –3 –1.75 –1.25 –0.75 –0.25 0.25 0.75 1.25 1.75 3 5 7

Figure 5.1a–e. Observed and GFDL Model-Simulated Precipitation (mm/day).

CCSM3 (yrs 400–499) DJF

Min = 236.83 Max = 305.62

Min = 220.88 Max = 305.82

Min = –16.12 Max = 20.98

Figure 5.2d. CCSM3 Annual Mean Simulated-Minus-Observed Sea Surface

Figure 5.3a. Simulation Global Warming relative to 1900 for gfdl_cm2_1

Volc” is the average of all

Global Warming relative to 1900 for giss_model_e_r

Global Warming relative to 1900 for ncar_ccsm3_0 Figure 5.3c. Simulation

Global Warming Relative to 1900 for American Models Figure 5.3d.

ensemble means for CCSM3

Surface Temperature Change Based on Local Linear Trends (°C)

–2 –1 –.6 –.1 .1 .3 .6 1 1.5 3

Figure 5.4. N ear-Surface Temperature Changes as Observed (top panels) and as

D JF 300 hPa v' v' ( m2s–2) Figure 5.5a.Top:

50°E 150°E 110°W 10°W

50°E 150°E 110°W 10°W

5.2.2.2 T RO PIC AL STO RMS in simulating the effects of El Niño on Atlantic

Figure 5.6. Cumulative 50