Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 19
CHAPTER ONE
Introduction
C
limate information is being used by a vast array of organizations within the
public and private sectors, with decisions based on climate information being
made every day. Users of climate information include national security planners,
infrastructure decision makers, public policy makers, insurance companies, water man-
agers, agricultural managers, and more. Each of these communities has different needs
for climate data from numerical simulations, with different time horizons and differ-
ent tolerances for uncertainty. Many user groups want very highly spatially resolved
information about the likely range of climate variability and extreme events such as
droughts, floods, or heat waves, while others are looking for data on long-term trends.
Some concrete examples of current users of climate information are farmers, city plan-
ners, water managers, and insurance companies, and details about their use of climate
information are described in Box 1.1.
BOX 1.1 EXAMPLES OF CLIMATE DATA USERS
Climate data are needed by many individuals and companies. Below are several representa-
tive examples of individuals and organizations who use climate data, why they need them, how
they are used, and what the payoff is.
Farmers
Farmers have always been close to weather and climate, as their economic success depends
on the right timing of planting, irrigation, and harvesting and the right choice of crops for the
local climate. In their day-to-day decision making about irrigation, farmers depend heavily on
short-term weather forecasts, which give them information not only about temperature and
precipitation but also about soil moisture levels that are crucial for many crops. One concrete
example is corn farming—a $15.1 billion business in the United Statesa—which is very sensitive
to drought and low soil moisture. Decisions made on the time scales of weeks to seasons rely on
short-term and seasonal forecasts of the soil moisture, which have become invaluable tools to
help farmers decide on irrigation needs during drought conditions; it is estimated that by 2015
improved weather forecasts will allow the agriculture sector to save $61 million on irrigation
water costs (Centrec Consulting Group, 2007). On time scales of seasons to years, forecasts of El
Niño/La Niña conditions help farmers to decide when to plant and harvest their crop, with an
estimated economic benefit on the order of $500 million to $950 million per year from the sea-
sonal El Niño/La Niña forecast for the U.S. agricultural sector (Chen et al., 2002). On even longer
time scales, the changing climate is shifting growing seasons and regions. Farmers are directly
19
OCR for page 20
A N AT I O N A L S T R AT E G Y F O R A D VA N C I N G C L I M AT E M O D E L I N G
BOX 1-1 CONTINUED
impacted because many of them have specialized in growing specific crops, which in turn are
often highly specialized for the climatic conditions they tolerate (see Figure 1). Longer-term re-
gional climate projections of precipitation, temperature, and soil moisture will allow farmers to
decide which crops to focus on in the future and to prepare for investments in new technologies
needed to successfully grow new crops.
Mayors of Large Cities
One of the main concerns about climate change is associated with the projected increase in
the frequency, duration, and intensity of heat waves. According to the National Weather Service
(NWS),“heat is the number one weather-related killer in the U.S.,” b claiming more lives each year
than floods, lightning, tornadoes, and hurricanes combined. Heat waves also increase the peak
demand for electricity, with the potential for blackouts and the high economic cost associated
with them. (Estimates for the August 2003 blackout that affected numerous cities in the United
States and Canada ranged from $4 billion to $10 billion [U.S.-Canada Power System Outage Task
Force, 2004]). Using a heat index that considers absolute temperature and humidity to assess how
hot it really feels, the NWS forecasts extreme heat events several days in advance. This allows city
officials to prepare for heat waves by warning the public, instituting energy-saving programs, and
designating community cooling centers, reducing some of the negative impacts of heat waves
FIGURE 1 U.S. Department of Agriculture plant hardiness zone maps are used extensively by gardeners and
growers to determine which plants are most likely to thrive at a location. Maps are based on the average an-
nual minimum winter temperature, divided into 10°F zones. The map on the left is based on data from 1974-
1986, and the map on the right is based on data from 1976-2005. The more recent map (right) is generally
one half-zone warmer than the previous map (left). SOURCES: http://arborday.org/media/map_change.cfm;
http://planthardiness.ars.usda.gov/PHZMWeb/AboutWhatsNew.aspx (both accessed October 11, 2012).
20
OCR for page 21
Introduction
BOX 1-1 CONTINUED
and saving lives. In the longer term, climate projection data allow mayors and other planners
to develop adaptation strategies (NPCC, 2010) to help plan for some of the negative impacts of
these changes. These adaptation strategies include programs to increase the energy efficiency of
buildings, investments in power grid infrastructure, and zoning changes to mandate the planting
of street trees in heat-stressed neighborhoods. Improved climate data (Figure 2) can help cities
make more informed decisions on long-term infrastructure investments that will help to protect
the health and economic interests of their constituents.
FIGURE 2 Heat waves are projected to occur more frequently in the future. Map shows the projected fre-
quency of extreme heat for later in the century (2080-2099 average). Extreme heat refers to a day so hot
that it occurred only once every 20 years in the past, and the projections show that extreme heat will occur
every 1-3 years in much of the United States by the end of the century.
21
OCR for page 22
A N AT I O N A L S T R AT E G Y F O R A D VA N C I N G C L I M AT E M O D E L I N G
BOX 1-1 CONTINUED
Hydropower System Managers
The Federal Columbia River Power System generates more than 76,000 gigawatt-hours
(GWh) of electricity per year, accounting for about 30 percent of the electricity used by the more
than 15 million people in the Pacific Northwest and having an estimated worth of approximately
$4 billion per year (BPA, 2010). To continue generating power at this level, river managers like
those for the Columbia River power system need to make both short- and long-term decisions
regarding how much water to store (compared to natural flow), which requires climate data
to predict and adapt to future changes in river flow. The climate data most needed by river
power management are temperature, precipitation, and wind, with information preferably at
high spatial resolutions of 1-10 km and with daily or higher frequency. Current climate data are
only available at much lower resolutions, but even these data have been useful in projecting
seasonal changes, such as increased winter runoff but less spring/summer runoff. Managers
also use longer-term projections of climate change to make decisions on modifying existing
infrastructure and/or acquiring additional infrastructure (for example, Figure 1.2). Managers such
as those who monitor the Columbia River desire more reliable and higher-resolution climate
data to help with planning and ultimately their ability to continue to supply power to millions
of Americans (Figure 3).
FIGURE 3 Managers of hydropower systems such as those of the Federal Columbia River Power System re-
quire climate information for both short-term operational decisions and long-term infrastructure planning.
SOURCE: Steven Pavlov, http://commons.wikimedia.org/wiki/File: Grand_Coulee_Dam_in_the_evening.jpg
(accesed June 8, 2012).
22
OCR for page 23
Introduction
BOX 1-1 CONTINUED
Insurance Companies
Insurance companies provide insurance to people and businesses against the impacts of
natural disasters. Insurance rates for weather- and climate-related disasters such as floods, high
winds, droughts, etc., are based on the expected occurrence of those events. To realistically assess
the probabilities of weather- and climate-related natural disasters, insurance companies have
been using climate data on past weather events for many years to develop specific risk models
for different regions and operations (e.g., transportation, farming, and construction). Weather-
and climate-related losses have increased rapidly in recent years (Figure 4), with record-breaking
FIGURE 4 Annual occurrence of natural disasters in the United States, broken down by origin as of 2010,
shows that the past may no longer be a reliable guide to the future. Record-breaking insured losses from
weather- and climate-related disasters of over $50 billion were recorded in 2011. SOURCE: Munich RE;
http://www.munichre.com/app_pages/www/@res/pdf/media_relations/ press_dossiers/hurricane/2011-
half-year-natural-catastrophe-review-usa_en.pdf (accessed September 14, 2012).
23
OCR for page 24
A N AT I O N A L S T R AT E G Y F O R A D VA N C I N G C L I M AT E M O D E L I N G
BOX 1-1 CONTINUED
insured losses of more than $50 billion in 2011.c More and more large insurance and reinsurance
companies are recognizing that climate change poses an enormous challenge to their busi-
ness. Accurately reflecting changed risks and actively and profitably managing climate change
impacts, rather than withdrawing from high-risk markets, is a major challenge for the insurance
industry. To address it, new kinds of climate data are required, focusing on projections rather than
historical observations. High-quality regional climate projections of variables such as sea level,
temperature, precipitation, wind, and extreme events will be crucial for the insurance industry
to rise to this challenge, so insurers can continue to provide disaster coverage for people and
businesses in the United States and use their past experience with risk mitigation (e.g., fire and
earthquake building codes) to help prevent losses of lives and property (Mills and Lecomte, 2006).
National Security Sector
National security planners and decision makers use climate information and forecasts over
a broad range of time scales. The February 2010 Quadrennial Defense Review notes that climate
change will play a significant role in the future security environment for the United States (Gates,
2010). Concurrently, the U.S. Department of Defense and its military services are developing poli-
cies and plans to understand and manage the effects of climate change on military operating
environments, missions, and facilities (NRC, 2011c). It has been estimated that $100 billion of
naval facilities are at risk from sea-level rise of 3 feet or more (NRC, 2011c) (Figure 5). The national
security risks associated with a changing climate have also recently been assessed in a report by
the Center for American Progress (Werz and Conley, 2012). The Navy would like to use climate
model outputs for information related to increasing Arctic maritime activity, water and resource
scarcity, and the impact of sea-level rise on installations (NRC, 2011c). In order to use climate
model projections to inform its decisions, the Navy would need high-spatial-resolution regional
climate models on decadal time scales, uncertainty quantification of the models, and probability
distribution functions in the model output. The Navy is a “good example of a stakeholder that
has very specific needs in applications related to its infrastructure and operations, disease, civil
instability, migration, water resources, and energy” (NRC, 2011c).
The Building Community
The built environment (buildings, communications, energy, industrial facilities, transporta-
tion, waste, water, and associated natural features) shelters and supports most human activities
and constitutes a large portion of the nation’s wealth (Figure 6). It has important roles in reduction
of greenhouse gas emissions and in measures to help society adapt economically, environmen-
tally, and socially to climate change. The building community includes professionals—including
architects, engineers, geologists, landscape architects, and planners—as well as owners, investors,
facilities managers, contractors, manufacturers of building materials, health and safety regulators,
and stakeholders served or affected by the built environment (nearly everyone).
The building community uses climate information, particularly on extremes, to ensure that
buildings are safe, functional, and resilient. Historically, the extreme environments used in assess-
24
OCR for page 25
Introduction
BOX 1-1 CONTINUED
FIGURE 5 The amphibious assault ship USS Kearsarge (LHD 3) pulls away from its berth at Naval Station
Norfolk. An estimated $100 billion worth of naval facilities are at risk from sea-level rise of 3 feet or more.
SOURCE: http://www.navy.mil/view_single.asp?id=125450 (accessed June 6, 2012).
ment and design of the built environment have not been based on climate or weather models.
Rather, extreme environments have been defined by statistics of historical records, albeit to
within observation and sampling errors. With climate and weather changing, historical records
no longer are adequate predictors of future extremes. However, advanced modeling capabilities
potentially can provide useful predictions of extreme environments.
Often decisions about buildings and other infrastructure are made for very long time
scales—decades and beyond. When looking at building decisions related to material choices,
siting, and building design, there are any number of questions related to climate, including:
How heavy are future rains and/or snowfalls likely to be? What range of temperatures is likely?
What will average precipitation rates mean for the water table? Will it flood? Adaptation of the
built environment to climate change is particularly important because it has significant resource
implications. The U.S. Department of Commerce estimates total construction spending in the
United States to be more than $820,000 million annually.d
25
OCR for page 26
A N AT I O N A L S T R AT E G Y F O R A D VA N C I N G C L I M AT E M O D E L I N G
BOX 1-1 CONTINUED
FIGURE 6 Construction of the Sovereign, Atlanta, Georgia. The building community uses climate informa-
tion to make decisions about building materials, siting, and building design. These types of infrastructure
decisions can have implications for decades. As the climate changes, information from climate models is
being used as a guide to future climate conditions. SOURCE: Conor Carey, http://commons.wikimedia.org/
wiki/File:Sovereign-Atlanta.jpg (accessed June 6, 2012).
a www.epa.gov/oecaagct/ag101/cropmajor.html (accessed October 11, 2012).
b http://www.nws.noaa.gov/os/heat/index.shtml(accessed November 30, 2012).
c www.noaa.gov/extreme2011/ (accessed October 11, 2012).
d www.census.gov/construction/c30/c30index.html (accessed October 11, 2012).
26
OCR for page 27
Introduction
Over the next several decades climate change and its myriad consequences will be
further unfolding and likely accelerating (NRC, 2011a). Probable impacts from climate
change, including sea-level rise, a seasonally ice-free Arctic, large-scale ecosystem
changes, regional droughts, and intense flooding events, will increase demand for
climate information. The value of this climate information is large. One of the more
prominent places to see this is through the impacts of extreme climate and weather
events; extreme climate and weather events are one of the leading causes of eco-
nomic and human losses, with total losses between 1980 and 2009 exceeding $700
billion (NCDC, 2010) and damages from more than 14 weather- and climate-related
disasters totaling more than $50 billion in 2011 alone.1 Climate change is affecting the
occurrence of and impacts from extreme events, such that the past is not necessarily
a reliable guide for the future, which further underscores the value of climate informa-
tion in the future.
An example of the value of climate information on shorter time scales comes from the
flooding throughout the Upper Midwest in the spring and summer of 2011. Extensive
rainfall in the spring and summer of 2011 led to flooding of the Mississippi and Mis-
souri rivers. Prior to that spring, climate predictions showed increased risk of flood-
ing throughout much of the Upper Midwest as a result of above-average snowpack
melting and precipitation levels (Figure 1.1), allowing government authorities to plan
ahead. According to the National Oceanic and Atmospheric Administration (NOAA),
these climate predictions allowed the government to coordinate “with local, state and
federal agencies before and during the flooding, so that emergency officials could
make important decisions to best protect life and limit property damage.”2 Such deci-
sions included evacuations and destruction of levees in some locations to allow excess
waters to flow into floodways.
In looking at longer time scales, climate models can provide information on projected
rainfall runoff for the coming decades (Figure 1.2). Some areas of the United States,
such as the Southwest, are projected to see decreases in average rainfall, while some
areas, like the Northeast, will see increases. Such changes will have major implications
for future water supplies, crop yields, and wildfire risks, among other effects. This type
of projected information allows counties and states to plan ahead for these condi-
tions, including decisions regarding infrastructure. However, the relationship between
regional drought and predictable patterns of climate variability is complicated, so
users of climate information must understand and deal with considerable predictive
uncertainty.
1 http://www.noaa.gov/extreme2011/ (accessed October 11, 2012).
2 http://www.noaa.gov/extreme2011/mississippi_flood.html (accessed October 11, 2012).
27
OCR for page 28
A N AT I O N A L S T R AT E G Y F O R A D VA N C I N G C L I M AT E M O D E L I N G
FIGURE 1.1 The spring flood risk outlook from NOAA’s National Weather Service for 2011. Extensive flood-
ing of Mississippi and Missouri rivers occurred in 2011. SOURCE: http://www.noaa.gov/extreme2011/mis-
sissippi_flood.html (accessed October 11, 2012).
WHAT IS A CLIMATE MODEL?
Information about the future of the climate system comes from computer models
that simulate the climate system. Climate models are mathematical representations
of physical, chemical, and biological processes in Earth’s climate system (Figure 1.3).
Computer models are a part of everyday life—there are models that forecast weather,
simulate how to fly an airplane, predict tides, and aid in drug discovery. Models are
used to study processes that are inherently complex, require large amounts of infor-
mation, or are impractical to study directly. They are essential tools for understanding
the world and allow climate scientists to make projections about the future.
28
OCR for page 29
Introduction
FIGURE 1.2 Longer-time-scale climate projections can assist in long-term planning. The figure shows pro-
jected changes in annual average runoff for 2041-2060 relative to a 1901-1970 baseline by water resource
region, based on analyses using emissions that fall between the lower and higher emissions scenarios.
Lower average runoff is expected in the Southwest and greater runoff is projected for the Northeast. Col-
ors indicate percentage changes in runoff, with hatched areas indicating greater confidence due to strong
agreement among model projections. SOURCE: USGCRP, 2009.
The many different kinds of climate models are all derived from fundamental physi-
cal laws such as Newton’s laws of motion and the chemistry and thermodynamics
of gases, liquids, solids, and electromagnetic radiation. These are supplemented by em-
pirical relationships determined from observations of complex processes such as ice
crystal formation in clouds; turbulent mixing, and waves in both air and water; biologi-
cal processes; sea-ice growth; and glacier movement.
The main components within a climate model include
• atmosphere (simulates winds, temperatures, clouds and precipitation, turbu-
lent mixing, transport of heat, water, trace chemicals and aerosols around the
globe),
• land surface (simulates surface characteristics such as vegetation, snow cover,
soil water, rivers, ice sheets, and carbon storage),
29
OCR for page 36
A N AT I O N A L S T R AT E G Y F O R A D VA N C I N G C L I M AT E M O D E L I N G
and Intercomparison (PCMDI) at Lawrence Livermore National Laboratory has been
instrumental in developing CMIP, including archiving, analysis, and quality control
of model output, although CMIP now has broad international institutional support.
CMIP has developed into a vital community-based infrastructure in support of climate
model diagnosis, validation, intercomparison, documentation, and data access.
What Can Climate Models Do Well?
Climate models have evolved into remarkably sophisticated tools for addressing a
diverse range of scientific and societally relevant issues. Their fidelity can be assessed
by comparing them statistically with such observations (Box 1.2) as the mean sea-
sonal cycle, seasonal extremes of temperature, rain and snowfall, and other routinely
measured quantities around the globe, as well as statistics of the El Niño-Southern
Oscillation and other important forms of climate variability and the observed changes
of climate over the past century and across previous eras. The evolution of climate
models over the past 50 years and the diversity of models used for different purposes
and across different space and time scales are discussed in more detail in Chapter 3.
Climate models skillfully reproduce important, global- to continental-scale features of
the present climate, as assessed in more detail by IPCC (2007c, Chapter 11). For in-
stance, over most parts of the globe, the simulated seasonal-mean surface air temper-
ature is within 3°C of observations (IPCC, 2007c), compared to an annual cycle that can
exceed 50°C in places, and simulated seasonal-mean precipitation has typical errors
of 50 percent or less on regional scales of 1,000 km or larger that are well resolved by
these models (Pincus et al., 2008). In the oceans, projected seasonal-mean sea-surface
temperatures are within 1-2°C of those observed over most of the globe, and major
ocean current systems like the Gulf Stream are correctly positioned (IPCC, 2007c). The
simulated seasonal patterns of sea-ice extent, snow cover, and cloudiness are also in
broad agreement with observations (IPCC, 2007c; Pincus et al., 2008). Swings in Pacific
sea-surface temperature, winds, and rainfall associated with El Niño are simulated by a
number of climate models with fairly realistic amplitude, location, and period (Achuta-
Rao and Sperber, 2006; Neale et al., 2008). Other forms of natural climate variability,
such as the year-to-year range of seasonally averaged temperature or rainfall over
regions of 1,000 km or larger in size and their spatial patterns of year-to-year variabil-
ity, are also simulated reasonably well (Gleckler et al., 2008). Simulation of the statistics
of extreme hot and cold spells has also improved (IPCC, 2007c), especially in models
using grid spacings of less than 100 km. In many ways climate models have become
remarkably accurate tools for simulating observable statistical aspects of the Earth
system (see Chapter 3 for more details of historical model improvements).
36
OCR for page 37
Introduction
Climate models do have well-known limitations for simulating the current climate,
stemming from both the coarseness of their grid spacing and the challenge of encap-
sulating the complex physical interactions between all parts of the climate system.
For instance, the grid of current climate models cannot represent fine-scale details
of mountain ranges important for simulating snowpack, rainfall, and glaciation in
such regions, details of coastal processes such as oceanic upwelling or tidal currents,
or hurricanes and severe thunderstorms. Tropical rainfall and many cloud processes
rely on interactions between very small scale air motions and other processes such
as condensation or freezing that are also not straightforward to represent in current
climate models. Other limitations include a lack of fully coupled land-ice or ocean
biogeochemistry models in many simulations, which are areas of active research but
which are just starting to be included in climate simulations. Furthermore, credible
simulations of some processes, such as the formation of continental ice sheets, would
require model runs of tens of thousands of years that are not yet feasible on current
computers.
The main concern among scientists, decision makers, and the interested public is the
extent to which climate projections can be trusted based on model simulations for
the next decades, the next century, and beyond. Here the crucial problem is that hu-
man greenhouse gas and aerosol emissions are quickly moving the climate outside its
natural range over at least the past few million years, so it is doubtful that the past can
act as a guide to the future. Furthermore, theory, observations, and climate models all
point to strong positive internal feedbacks within the climate system that increase its
response to changes in its composition. How can we be sure our best climate models
can reliably simulate not only the current climate, but also how human influences
(presupposing we know what they will be) will change climate?
The best indicators are (i) the ability of models to simulate observed climate change
of the past 150 years and especially the more rapid and more comprehensively
measured changes of the past 30 years, and (ii) the spread in projections made using
different climate models or model versions, both taken in the context of paleoclimate
observations and simulations that suggest circumstances that may favor abrupt or
rapid changes in climate regime. Comparisons of multiple state-of-the-art models
against one another (and observations) advance understanding of the climate sys-
tem and help build trust in model projections. Intermodel differences provide a lower
bound on the uncertainty of climate projections. They may miss sources of error
common to all current models; one might hope that, as climate models become more
comprehensive, the likelihood of such errors diminishes as long as the model compo-
nents and their interactions are carefully tested against observations. Chapter 10 of
37
OCR for page 38
A N AT I O N A L S T R AT E G Y F O R A D VA N C I N G C L I M AT E M O D E L I N G
BOX 1.2 HOW DO CLIMATE MODELS GET EVALUATED?
As the IPCC report Climate Change 2007: The Physical Science Basis notes, “there is consider-
able confidence that climate models provide credible quantitative estimates of future climate
change, particularly at continental scales and above” (IPCC, 2007c). There are three primary
reasons for this confidence: (1) As noted above, the fundamentals of a climate model are based
on established physical laws, such as laws of conservations of energy, mass, and momentum. (2)
Climate model simulations are routinely and extensively assessed by being compared with ob-
servations of the atmosphere, ocean, cryosphere, and land surface (Figure 1). (3) Climate models
are able to reproduce features of past climates and climate changes (e.g., the warming of the
past century, and the mid-Holocene warming of the Northern Hemisphere 6,000 years ago).
FIGURE 1 Climate model development and testing involves multiple stages and the contributions of the
model development community, the model user/evaluation community, and the data community. SOURCE:
Jakob, 2010.
38
OCR for page 39
Introduction
BOX 1-2 CONTINUED
Current climate models are calibrated during their development process to match observa-
tions within reasonable uncertainty ranges. However, the warming to date due to greenhouse
gas increases has been partially compensated by an uncertain amount of cooling caused by
human-induced enhancement of light scattering by aerosols and by their effect on clouds; this
compensation has been estimated to be from 20 to 70 percent (with 90 percent confidence)
based on a range of observational and model-based studies (IPCC, 2007d). Over the 21st century,
global aerosol emissions are expected to not increase further, but greenhouse gas emissions
are likely to accelerate for at least the next few decades, so this compensation will become less
significant. Because of the uncertain cooling by aerosols the current warming cannot be used
to constrain the “climate sensitivity.” Thus, the simulated 21st-century global-average warming
varies across the international suite of climate models with a range of approximately 30 percenta
as is further discussed in Chapter 4.
Models provide quantitative estimates of future climate change, but with significant sources
of uncertainty—lack of knowledge, or imperfect knowledge about specific quantities or the
behavior of a system. These include the uncertainty in the “forcing” on the climate system from
future greenhouse gas and aerosol emissions, as well as natural processes such as volcanic erup-
tions and solar variability, used as inputs to climate models; the uncertainty in the climate system
response to this forcing; the uncertainty from natural internal variability of the climate system;
the uncertainty from incomplete representations of known but complicated and small-scale
processes (such as cumulus clouds) and of poorly understood processes (such as ice nucleation
in clouds); and the uncertainty from “unknown unknowns” (see Chapter 6 for more information
on uncertainty).
a More specifically, the interquartile range is 30 percent of the mean, where the interquartile range is
a measure of statistical dispersion, and measures the difference between the 75th percentile and the 25th
percentile of the data.
39
OCR for page 40
A N AT I O N A L S T R AT E G Y F O R A D VA N C I N G C L I M AT E M O D E L I N G
IPCC (2007c) discusses some other strategies that are also used to estimate or bound
model uncertainties.
Figure 1.7 shows which aspects of climate can be most robustly predicted, separated
by phenomenon and time scale, based on such assessments. In general, climate models
more robustly predict trends at larger space and time scales, and they predict tempera-
ture trends more reliably than precipitation trends. They all project a reduction in sum-
mer sea-ice extent, but not as large as that observed in recent years. They robustly pre-
dict the contribution to global sea-level rise from heat uptake in the oceans, but most do
not include a representation of ice-sheet melt and the disintegration of the tongues of
large glaciers that may considerably accelerate sea-level rise over the next century. They
agree that the polar regions will become wetter and that the subtropics will become
drier, but they do not agree on which regions of the subtropics will experience strong
drying. As climate models become more comprehensive and their grid scale becomes
finer, they can provide meaningful projections of more parts of the climate response
and their possible feedbacks on the overall climate system, but this does not necessar-
FIGURE 1.7 Time and space scales of key climate phenomena. Color coding shows relative reliability
of climate model simulations of these phenomena (or their statistics in the present climate, for climate
variability/extremes).
40
OCR for page 41
Introduction
ily reduce projection uncertainty about some aspects of climate change. Indeed, global
climate sensitivity, defined as the global warming simulated by a climate model in re-
sponse to a sustained doubling of atmospheric CO2 concentrations, still shows a similar
30 percent spread3 across leading models as it did 20 years ago.
Climate Information Delivery to Users
Although a number of aspects of the climate system can be projected with some
degree of confidence, this climate information may not be useful for making deci-
sions. As climate models have become more ambitious, so have their users. Many users
of climate model outputs need to make decisions on how or whether to respond to
climate change, in some cases within institutions where the reality or importance of
climate change is not universally acknowledged. Users consider the information from
the climate models a valuable commodity, but they are not always sure what data
are available to them or how to best use them to inform their decisions. The research
community, both by limited capacity and by culture, is often hard pressed to respond
to the desires of the user community for new types of model output at high time and
space resolution. Quantifying uncertainty in climate projections is still a multifaceted
research problem, making communication of relevant uncertainties with diverse user
groups challenging, especially when these uncertainties are perceived to be discour-
agingly large or the climate model output is only part of a modeling chain.
WHY THIS STUDY?
With many studies and reports showing that there will likely be significant impacts
as a result of climate change (IPCC, 2007a,b,c; NRC, 2010a,b,d,e, 2011a), now is an
appropriate time to examine the capabilities of the nation’s climate modeling enter-
prise to ensure that it is advancing adequately. The modeling community has already
developed plans to make continued progress over the next 3-5 years. However, both
the climate science and applications communities would enormously benefit from
a major advance in improving the usefulness of climate projections, especially on
regional space scales and decadal time scales and including trends in extreme events.
Is this possible? Is this likely? How can the United States best position itself to advance
and better use climate models? What resources and planning will that take? The need
has arisen for a forward-looking, comprehensive, strategic assessment of how best to
improve the United States’ capabilities to simulate past, present, and future climate on
local to global scales and at decadal to centennial time scales.
3 This is a 30 percent interquartile spread; see previous footnote for definition of interquartile.
41
OCR for page 42
A N AT I O N A L S T R AT E G Y F O R A D VA N C I N G C L I M AT E M O D E L I N G
In recognition of this need, the Committee on a National Strategy for Advancing Cli-
mate Modeling was tasked by NOAA, NASA, DOE, NSF, and the intelligence community
to produce a high-level assessment, providing a strategic framework to guide progress
in the nation’s climate modeling enterprise over the next 10-20 years (see Appendix A
for the full statement of task).
STRUCTURE OF THIS REPORT
In response to the statement of task for this committee (Appendix A), this study has
built upon recent efforts to engage and coordinate the national and international
climate modeling community, recent NRC and interagency reports that have made
recommendations about both U.S. climate modeling and its role in the broader and
more diverse climate research and applications communities, and recent actions and
progress by federal agencies and other domestic groups. Ultimately, the report at-
tempts to provide a coherent set of recommendations understandable to nonexperts
(Box 1.3 includes the definition of a number of key terms), and to set out a compre-
hensive, unified, and achievable vision for climate modeling for the next decade and
beyond that can form the basis of a national strategy that advances climate models,
climate observations,4 and user needs.
To obtain advice from a broad spectrum of climate modelers, researchers using
climate model output, and the diverse and growing community of users of climate
model outputs and projections, the committee convened a 50-person community
workshop to engage with leaders from the modeling and user communities. During
day-long open sessions at four other meetings, the committee heard from other stake-
holder groups, both nongovernmental and from various levels of government, that are
trying to use climate projections for long-term planning (Appendix B has more detail
on the information-gathering process). The presentations and discussions encom-
passed global and regional models, downscaling, computing and data, user needs and
education, the role of the private sector, and cultivating a coordinated national model-
ing and user community that spans many goals and applications.
4 One cannot consider advancing climate modeling without attention to the supporting climate obser-
vations, both space-based and in situ, needed to initialize, force, and validate climate models, as well as for
monitoring climate variability and change. The United States currently does not have a coordinated climate
observing system, or a strategy that could lead to a coherent system, across both in situ and remotely sensed
observations. As noted in the report Improving the Effectiveness of U.S. Climate Modeling (NRC, 2001b): “the
lack of a suitable sustained observing system for climate limits progress in climate modeling.” This statement
still rings true today, and therefore this report only discusses observations at a high level.
42
OCR for page 43
Introduction
The committee was charged with examining “decadal to centennial” time scales
(Appendix A) but decided to extend the report to also touch on shorter time scales,
including intraseasonal to interannual (ISI) time scales, even though ISI climate predic-
tion has recently been assessed by another NRC report (NRC, 2010c). Four motivations
for this were the following: (1) Seasonal to interannual prediction is a valuable test of
climate models, because phenomena that evolve on this time scale like El Niño have
been well observed for more than 25 years, during which they have gone through
enough cycles to allow the seasonal prediction skill of climate models to be tested
and compared. (2) Decadal prediction of climate is a natural extension of interannual
climate prediction, because it also requires a detailed initial knowledge of the ocean
state. (3) For many users, simulation of climate variability about the long-term trends
we project is also very important; ISI simulations observationally test aspects of the
skill of climate models in predicting this variability. (4) Currently, ISI climate prediction
is a nexus between U.S. operational weather and climate forecasting at short time
scales (e.g., as performed at NCEP) and research-oriented climate modeling at long
time scales. Hence, it may be a fruitful arena to explore closer interactions between the
operational and research modeling communities.
This report is structured in three sections. In addition to the introduction in this chap-
ter, this first section reviews the history of previous reports as context for this report
(Chapter 2). Building on that background material, the second section of the report
examines a number of the issues that are currently facing the U.S. climate modeling
community. These issues include climate model hierarchy, resolution, and complex-
ity (Chapter 3); scientific frontiers in climate modeling (Chapter 4); integrated climate
observations (Chapter 5); characterization, quantification, and communication of
uncertainty (Chapter 6); the climate model development workforce (Chapter 7); the
relationship of U.S. climate modeling efforts with international efforts (Chapter 8); and
operational climate prediction systems (Chapter 9).
This final section of the report examines several key issues in the U.S. climate model-
ing enterprise where the committee presents its primary recommendations and an
overarching national strategy for advancing climate modeling in the United States
over the next two decades. These issues include the challenges and opportunities
related to computational infrastructure (Chapter 10), unified climate modeling (Chap-
ter 11), interfacing with the trained climate model user and educational communities
(Chapter 12), and optimizing U.S. institutional arrangements (Chapter 13). A number
of specific recommendations are presented throughout the text. These recommenda-
tions are synthesized into an overarching strategy in the final chapter of the report
(Chapter 14).
43
OCR for page 44
A N AT I O N A L S T R AT E G Y F O R A D VA N C I N G C L I M AT E M O D E L I N G
BOX 1.3 DEFINITION OF KEY TERMS
Boundary conditions: External data input into climate models that define conditions that are
fixed relative to the dynamic elements of those models. In the case of Earth system models, the
boundary conditions define the orbit of Earth, the land/ocean cover, the height of the moun-
tains, drainage basins and paths of rivers, and the radiation from the sun, among other things.
See also Forcings.
Climate models and Earth system models: Climate models are computer codes that encapsulate
the physical laws governing the motions and cycles of energy and water in the atmosphere,
ocean, and land surface, including sea ice and snow. Earth system models are climate models
that additionally incorporate representations of the chemical and biological processes that con-
trol the cycling of human-produced and natural aerosols, as well as biogeochemical substances
including carbon, nitrogen, and sulfur. Some Earth system models also represent ice sheets and
climate-induced changes in the distribution of different types of vegetation.
Climate predictions and projections: Climate predictions are model simulations that are started
from our best estimate at the state of the climate system at a particular time. Climate projec-
tions, on the other hand, are simulations started from a statistically representative initial state.
Both predictions and projections are made using estimates of future values of the forcings. The
goal of projection is to look at the statistics of the simulated climate and how they change; the
goal of prediction is to forecast the evolution of the actual climate state, including variations in
El Niño or the Atlantic meridional overturning circulation.
Common modeling framework: A group of programs that provides a high-performance, flexible
software infrastructure, which enables climate models to run on very large parallel computers
and that supports coupling diverse, modular climate model components.
Data assimilation: The process of making best use of observational data to provide an estimate
of the state of the system that is compatible with a given model and that is better than could
be obtained using just the data or the model alone.
Forcings: External data input into climate models that drive climate variations and change (e.g.,
greenhouse gas concentrations, volcanic aerosols, and solar irradiance variations).
Model fidelity: The measure of agreement between the statistical distributions of a climate vari-
able or group of variables as simulated by a model compared with observations (e.g., the seasonal
and geographical root-mean-square difference between simulated and observed rainfall over
1980-2010).
Model forecast skill: The typical accuracy of a forecast, e.g., as measured by the agreement be-
tween realistically initialized model predictions of some variable (e.g., winter-mean surface
air temperature over Kansas based on model predictions from 6 months before) and their
corresponding verifications over some period. The relation of fidelity to skill is similar to that
between prediction and projection. In particular, model fidelity (correctly predicting the statisti-
cal distribution of this quantity) need not imply model skill (skillfully predicting warm winters
when they are observed).a
44
OCR for page 45
Introduction
BOX 1-3 CONTINUED
Multimodel ensemble: A set of simulations from several different models, forced by the same ex-
ternal forcing. Considerable evidence suggests that an average over simulations from different
models produces a better match to observational climatological distributions than similar-sized
averages of simulations from any single model.
Perturbed physics experiments: Multiple simulations from the same models using a plausible range
of parameters or representations of physical processes. These simulations make it possible to
analyze the sensitivity of simulation results to some of the choices made in model development.
Operational climate prediction: Distinct from climate model research and development, opera-
tional climate prediction is a regularly scheduled, user-driven, product-oriented process that
conforms to a specified schedule of generation and delivery of products and that depends on
dedicated computing and information delivery resources with failsafe contingency plans.
Parameterization: The process of representing the effects of processes other than resolved-scale
fluid motion (e.g., cumulus cloud dynamics and microphysics; land-surface or sea-ice modeling;
or transfer of heat, salt, and nutrients in unresolved oceanic eddies) in a climate model by using
the resolved fields whose time evolution is predicted by the model.
Reanalysis: The process of reassimilating historical observations of atmospheric and oceanic
quantities such as temperature, pressure, wind, humidity, current, and salinity using fixed state-
of-the-art models and data assimilation techniques to produce long time series of global fields.
Regional climate models: Climate models that are restricted to a portion of the globe so as to
reduce the computational cost and thereby increase the spatial resolution, and which use the
output of a coarser-resolution global climate model at their boundaries. Such models are often
used in “downscaling,” the process of representing global climate model output at the relatively
small spatial scales that are more relevant to decision makers. Regional climate models some-
times include greater scientific complexity that can inform particular applications and decision
makers.
Seamless prediction: Viewing weather and climate prediction as problems that share common
processes and dynamics and that can be addressed using modeling approaches that span a
broad range of time scales and spatial resolutions.
Tuning: The process of adjusting the values of parameters used in climate models to achieve the
best fit to observations in a dependent control data set. The values are adjusted only within the
range of observational uncertainty of those parameters.
Uncertainty: Lack of knowledge or imperfect knowledge about specific quantities or the behavior
of a system.
Unified modeling across time scales: The ultimate realization of seamless prediction whereby a sin-
gle climate model is used to predict the weather, seasonal climate, and decadal climate change.
a There is some evidence that model fidelity and prediction skill are related (see DelSole and Shukla,
2010).
45
OCR for page 46
