|
||||||||||||||||
|
||||||||||||||||
The following HTML text is provided to enhance online readability. Many aspects of typography translate only awkwardly to HTML. Please use the page image as the authoritative form to ensure accuracy. Page 111 6
|
||||||||||||||||
 |
 |
Page 111 |
 |
 |
 |
|
| ||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||
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 111
Page 111
6
Crosscutting Issues
Certain issues cut across all the disciplines involved in
climate research. Among them are the nature of our available
climate information, modeling efforts, prediction efforts,
detection and attribution issues, and linkages across time scales.
Progress in these areas will go far to advance our understanding of
climate change and variability over decade-to-century time
scales.
Climate Information
Fundamental to our understanding of climate change is the
information we use to investigate and study such change. For
dec-cen time scales, the information is available through a variety
of forms, and each of these must be exploited if we are to make
significant headway. We have: (1) instrumental data for the past
100 years or so, which will be improved and extended into the
future; (2) proxy indicators of climate change, both historical
records and paleoclimate evidence; (3) analysis products, the
results of applying some form of analysis to observations to yield
a more focused picture or consistent interpolations; and (4) the
output of models, our only tool for estimating future climate
states. Each of these information products is briefly discussed
here, with suggestions for future additions to our climate
database.
Instrumental Data
To build the requisite understanding of climate change and
variability, we must first recognize our current state of
understanding and its limitations. The inadequacies of the current
instrumental data available for exploring climate change and
variability on dec-cen time scales are readily exposed when we try
to use them to answer some of our most fundamental questions. For
example: Is the planet getting warmer? Is the hydrologic cycle
changing? Is the atmosphere-ocean circulation changing? Are the
weather and climate becoming more extreme or variable? Is the
radiative forcing of the climate changing? These questions cannot
be answered definitively, because there is no global climate
observing system that gathers all the information needed. Each of
these apparently simple questions is actually quite complex, both
because of its multivariate aspects, and because the spatial and
temporal sampling required to adequately address it must be
considered on a global scale.
A brief review of our ability to answer these questions reveals
many areas of success, but also some glaring inadequacies that must
be addressed if we are to understand and predict climate change,
and to refine this database for future generations. The basic
problems are the shortness and inaccuracy of the instrumental
record and its lack of spatial coverage, together with the
difficulty of interpreting the paleoclimatic proxy record, which is
discussed in the next section. What we can do to improve our
ability to detect and monitor climate change is outlined at the end
of this section, following the discussion of the fundamental
questions.
Is the Planet Getting Warmer?
Measurements show that near-surface air temperatures are
increasing. Best estimates suggest that the overall warming has
been around 0.5ºC since the late nineteenth century (IPCC,
1996a). Nonetheless, many questions have arisen regarding the
adequacy of these estimates (IPCC, 1996a), and the relative
scarcity of global measurements throughout the century is only the
first. Changes in the methods of measuring land and marine
surface-air temperatures from ships, buoys, and land-surface
stations; changes in instrumentation, instrument exposures, and
sampling times; urbanization effects—these are but a few of
the time-varying biases that have plagued the interpretation of the
surface air-temperature records for the twentieth century. Only by
also considering other temperature-sensitive variables—e.g.,
snow cover, glaciers, sea level, and even some proxy non-real-time
measurements such as ground temperatures from bore-holes—can
we be confident that the planet has indeed warmed. The measurements
we rely on to calculate global
OCR for page 112
Page 112
changes of temperature were never collected for that purpose;
they were made primarily to aid in navigation, agriculture,
commerce, and, in recent decades, weather forecasting. Thus, many
uncertainties remain about important details of the temperature
increase. The IPCC (1996a) has summarized known changes in the
temperature record; this summary is presented in graphic form in
the upper panel of Figure 6-1.
Recent global-scale measurements of layer-averaged atmospheric
temperatures and sea surface temperatures from instruments aboard
satellites have greatly aided our ability to monitor global
temperature change (Spencer and Christy, 1992a,b; Reynolds, 1988),
but the situation is far from satisfactory (Hurrell and Trenberth,
1996). Changes in satellite temporal sampling (e.g., orbital
drift), changes in atmospheric composition (e.g., volcanic
emissions), and techni-
Figure 6-1
Schematic of observed variations of selected climate indicators. Upper panel, temperature indicators; lower
panel, hydrologic indicators. (From IPCC, 1996a; reprinted with permission of the Intergovernmental Panel on Climate Change.)
OCR for page 113
Page 113
cal difficulties related to overcoming surface-emissivity
variability limit our ability to produce highly reliable products
of near-surface global temperature change. Nonetheless, space-based
measurements have shown that stratospheric temperatures have
decreased over the past two decades, though perhaps not as much as
suggested by measurements from weather balloons. (It is now known
that the data from these balloons high in the atmosphere have an
inadvertent temporal bias, because of improvements in shielding
from direct and reflected solar radiation (Leurs and Eskridge,
1995).)
Even if the instrumental records of ground, atmosphere, and sea
surface temperatures were accurate enough, the length of the
records would still be an issue. To be certain that the system is
getting warmer, we need a record long enough to enable us to
distinguish a steady trend of warming from long-period variations,
which we may be seeing only partially. Identifying and
understanding dec-cen variability is crucial to determining whether
the planet is getting warmer.
Is the Hydrologic Cycle Changing?
The source term for the hydrologic water balance, precipitation,
has been measured for over two centuries in some locations. Even
today, however, it is acknowledged that in many parts of the world
we still cannot reliably measure true precipitation (Sevruk, 1982;
IPCC, 1996a). For example, annual biases of more than 50 percent
are not uncommon in cold climates (Karl et al., 1995), and even for
more moderate climates precipitation is believed to be
underestimated by 10 to 15 percent (IPCC, 1992). Improvements in
instrumentation have also introduced time-varying biases (Karl et
al., 1995). Satellite-derived measurements of precipitation are the
only ones that provide large-scale ocean coverage. Although
comprehensive estimates have been made of large-scale spatial
precipitation variability over the oceans, where few measurements
exist, problems inherent in developing such estimates limit our
confidence in using them to identify global-scale decadal changes.
For example, even the recent work of Spencer (1993) in estimating
worldwide ocean precipitation using a microwave sounding unit
aboard the NOAA polar orbiting satellites has several limitations:
The observations are limited to ocean coverage (and hindered by the
requirement of an unfrozen ocean), do not adequately measure solid
precipitation, have low spatial resolution, and are affected by the
diurnal sampling inadequacies associated with polar orbiters (e.g.,
limited overflight capability).
Documentation of past changes in land-surface precipitation has
been compared with other hydrologic data, such as changes in
streamflow, to ascertain its robustness. The lower panel of Figure
6-1 summarizes some of the more important known changes in
precipitation, such as the increase in the middle to high latitudes
and the decrease in the subtropics. There is also evidence to
suggest that much of the increase in mid- to high-latitude
precipitation arises from increased autumn and early-winter
precipitation in much of North America and Europe. Color plate 5
depicts the spatial aspects of the changes in precipitation during
this century; rather large-scale coherent patterns are
apparent.
Other changes related to the hydrologic cycle are also
summarized in Figure 6-1. Confidence is low for many of the changes
noted, which is particularly distressing given the important role
of clouds and water vapor in climate-feed-back effects. Records of
cloud amount are the result of surface-based (human) observations
(now being replaced by automated measurements in the United States)
and satellite measurements. Neither surface-based nor space-based
datasets have proven to be entirely satisfactory for detecting
changes in clouds. Polar-orbiting satellites have enormous
difficulties related to sampling aliasing and satellite drift
(Rossow and Cairns, 1995). For human observations, changes in
observer schedules, observing biases, and incomplete sampling have
created serious problems in data interpretations, now compounded by
the change to automated measurements at many stations. Nonetheless,
there is still some indication (but low confidence) that global
cloud amounts have tended to increase. This finding is supported by
reductions in evaporation (as measured by pan evaporimeters) over
the past several decades in Russia and the United States, and by a
worldwide reduction in land-surface diurnal temperature range.
Moreover, an increase in water vapor has been documented over much
of North America and in the tropics (IPCC, 1996a).
Water vapor is the most important greenhouse gas in the
atmosphere, so changes in water vapor, especially at upper levels
of the troposphere, are very important for understanding climate
change. The measurement of changes in atmospheric water vapor is
hampered by data-processing and instrumental difficulties for both
weather-balloon and satellite retrievals. Satellite data also
suffer from discontinuities among successive satellites and from
errors introduced by changes in orbits and calibrations.
Upper-tropospheric water vapor is believed to be a particularly
important climate-feedback quantity, but little can be said as yet
about how it has varied over the course of the past few decades,
how it responds to climate change, or how successful models are in
simulating these changes.
Is the Atmosphere-Ocean Circulation
Changing?
A number of the responses presented above suggest that the
atmosphere and ocean circulation systems have been changing. This
evidence is surprisingly meager, however. Daily analyses of
circulation are performed routinely, but the analysis schemes have
changed over time, so they are of limited use for monitoring
climate change. Moreover, even the recent re-analysis efforts by
the world's major numerical weather-prediction centers, for which
the analysis scheme is fixed over the historical record, contains
time-
OCR for page 114
OCR for page 115
OCR for page 116
OCR for page 117
OCR for page 118
OCR for page 119
OCR for page 120
OCR for page 121
OCR for page 122
OCR for page 123
Representative terms from entire chapter:
ice cores
Page 114
varying biases; some of the data carry embedded biases, and the
data mix changes over the course of the re-analysis (Trenberth and
Guillemot, 1998). Even less information is available on measured
changes and variations in ocean circulation. Only recently has a
single, coarse snapshot of the ocean been taken in the World Ocean
Circulation Experiment (WOCE).
Are the Weather and Climate Becoming
More Extreme or Variable?
Perhaps one of society's greatest concerns about weather and
climate is their extremes. Only a limited quantity of reliable
information is available about large-scale changes in extreme
weather or climate variability, for two reasons. First, there is
inadequate monitoring of the necessary quantities. Second, access
to weather and climate data held by the world's national weather
and environmental agencies is prohibitively expensive. The
time-varying biases that affect climate means are even more
difficult to effectively eliminate from the extremes of the
distributions of various weather and climate elements. There are a
few areas, however, where regional and global changes in weather
and climate extremes have been reasonably well documented.
Interannual temperature variability has not changed
significantly over the past century. On shorter time scales and
higher frequencies, however (e.g., days to a week), there is some
evidence for a decrease in high-frequency temperature variability
across much of the Northern Hemisphere (Karl et al., 1996). Related
to this decrease has been a tendency for fewer low-temperature
extremes to occur, but widespread changes in extreme high
temperatures have not been noted.
Trends in intense rainfalls have been examined for a variety of
countries. There is some evidence for an increase in intense
rainfalls (in the United States, tropical Australia, Japan, and
Mexico), but analyses are far from complete and the record contains
many discontinuities. The strongest increases in extreme
precipitation are documented in the United States and
Australia.
There are grounds for believing that intense tropical-cyclone
activity has decreased in the North Atlantic, the one basin for
which we have reasonably consistent tropical-cyclone data
throughout the twentieth century. Even here, though, it is
difficult to be certain of the tropical-cyclone strengths reflected
in data obtained prior to World War II. Elsewhere, tropical-cyclone
data do not reveal any long-term trends, or if they do, the trends
are most likely to be the result of inconsistent analyses. Changes
in meteorological assimilation schemes have introduced very
difficult problems in interpreting changes in extratropical cyclone
frequency. In some regions, however, such as the North Atlantic, a
clear trend toward increased storm activity has been noted. This
tendency has also been apparent in significant increases in wave
heights in the northern half of the North Atlantic. In contrast,
decreases in storm frequency and wave heights have been noted in
the southern half of the North Atlantic over the past few decades.
These changes are also reflected in the prolonged positive
excursions of the NAO since the 1970s.
Is the Radiative Forcing of the Planet
Changing?
Without an adequate time history of the important agents of
climate change—that is, those factors that affect the
radiative-heat balance of the planet—it is impossible to
understand global change. The atmospheric concentration of CO2, an important greenhouse gas because of
its long atmospheric residence time and relatively high atmospheric
concentration, has increased substantially over the past few
decades. This rise is quite certain; it is revealed by precise
measurements made at Mauna Loa Observatory since the late 1950s
(Keeling et al., 1976, 1989), at the South Pole since the mid-1960s
(Keeling and Whorf, 1994), and at a number of other stations around
the world that began operating in subsequent decades (WMO, 1984).
Since CO2 is a long-lived
atmospheric constituent and it is well mixed through the
atmosphere, a moderate number of well-placed stations can provide a
very robust estimate of global changes in carbon dioxide as long as
they operate for the primary purpose of monitoring
seasonal-to-decadal changes.
To understand the causes of the increase in atmospheric carbon
dioxide, however, we must understand how the carbon cycle operates
and how the anthropogenic carbon budget is balanced. Understanding
the carbon cycle, which is discussed in greater detail in the first
part of Chapter 5, requires estimates of anthropogenic sources of
carbon: the emissions resulting from fossil-fuel burning and
production and cement production, as well as the net emission from
changes in land use, such as deforestation. These estimates are
derived from a combination of modeling, sample measurements,
high-resolution satellite imagery, and sophisticated analysis of
different types of CO2 and climate
records (Keeling et al., 1996a; Dettinger and Ghil, 1998).
Understanding the carbon budget also requires measuring carbon
storage in the atmosphere, the ocean uptake, and uptake by forest
regrowth; assessing the effect on vegetation of CO2 and nitrogen fertilization; and taking
into account climate-feedback effects, such as the increase in
vegetation resulting from increased temperatures. At present many
of these factors are still uncertain, because so few sustained
ecosystem measurements have been made. It is clear, however, that
anthropogenic emissions are the primary cause of the atmospheric
increase of CO2. Indeed, a major
unresolved issue is why the atmospheric concentration of CO2 is not even higher than observed.
Several other radiatively important anthropogenic atmospheric
trace constituents have been measured over the past few decades.
These measurements have confirmed significant increases in
atmospheric concentrations of methane
Page 115
(CH4), nitrous oxide (N2O), and the halocarbons (including the
stratospheric-ozone-destroying agents, the chlorofluorocarbons and
the bromocarbons). Because of their long lifetimes, implying
spatial homogeneity in the atmosphere, a few well-placed,
high-quality in situ stations have been able to provide good
estimates of global changes in the concentration of these
constituents. Stratospheric ozone depletion has been monitored by
satellite and by in situ ozonesondes. Both observing systems have
been crucial in ascertaining changes in stratospheric ozone. (Note
that ozone was originally of interest not because of its role as a
radiative-forcing agent, but because of its ability to absorb UV
radiation before it reached the Earth's surface.) The combination
of the surface- and space-based observing systems has yielded much
more precise measurements than either system could have provided
alone. Over the past few years it has been possible to improve the
ozonesonde and satellite data by using information about past
calibration methods, in part because differences in trends between
the two observing systems could be identified.
Color plate 6 depicts the IPCC (1996a) best estimate of the
radiative forcing associated with various atmospheric constituents.
Unfortunately, measurements of most of the forcings other than
those already discussed have low or very low confidence, both
because of our uncertainty about their role in the physical climate
system and because we have not adequately monitored their change.
For example, our current estimates of changes in sulfate-aerosol
concentrations are derived from model estimates of source
emissions, not from measured atmospheric concentrations. Monitoring
sulfate aerosol is complicated because its short atmospheric
lifetime causes its concentration to vary spatially. Another
example of low confidence is measurements of solar irradiance.
These measurements have been taken by balloons and rockets for
several decades, but continuous measurements of
top-of-the-atmosphere solar irradiance did not begin until the late
1970s with the Nimbus 7 and the Solar Maximum Mission
satellites.
Significant absolute differences in total irradiance are found
between different satellites' measurements, emphasizing the
critical need for overlap between satellites and for absolute
calibration of the irradiance measurements to determine decadal
changes (NRC, 1994). Spectrally resolved measurements will be a key
element in our ability to model the effects of solar variability,
but at present no long-term commitment has been made to take such
measurements. Another important forcing currently estimated through
measured, modeled, and estimated changes in optical depth is that
related to the aerosols sporadically injected high into the
atmosphere by major volcanic eruptions. However, aerosols of
volcanic origin usually persist in the atmosphere for at most a few
years. Improved measurements of the size distribution, composition,
and radiative properties of volcanic aerosols will help us better
understand this agent of climate change.
What Can We Do to Improve Our Ability
to Detect Climate and Global Changes?
Even after extensive re-working of past data, in many instances
we are incapable of resolving important aspects of climate and
global change. Better quality and continuity, and fewer
time-varying biases, will be required of virtually every monitoring
system and dataset if we expect to conclusively answer questions
about how the planet has changed. Our inability to do so now is
often a result of having to rely on observations that were never
intended to be used to monitor the physical characteristics of the
planet over the course of decades. Long-term monitoring capable of
resolving dec-cen changes requires different strategies of
operation.
In situ measurement systems are in a state of decay or decline,
or are undergoing poorly documented change. Surface-based automated
measurement is being introduced without adequate precautions to
explore and record the differences between the old and new
observing systems. Satellite-based systems alone cannot provide all
the measurements necessary for detecting changes. Much wiser
implementation and monitoring practices must be adhered to for both
space-based and surface-based observing systems if we are to
adequately understand global changes. A number of steps can be
taken to improve our ability to monitor climate and global
change:
• When changes are made to existing environmental
monitoring systems, or new observing systems are introduced,
standard practices should include an assessment of the impact of
these changes on our ability to monitor environmental variations
and alteration.
• For critical environmental variables, it should be
standard practice to overlap measurements in time and space when a
new observing system replaces an old one.
• Information on instrument calibration and validation, as
well as the history of an observing station or platform, are
essential for data interpretation and use. Changes in instrument
sampling time, local environmental conditions, and any other
factors pertinent to the interpretation of the observations and
measurements should be recorded as a mandatory part of the
observing routine and be archived with the original data. The
algorithms used to process observations need to be well documented,
and accessible to the scientific community. Documentation of
changes and improvements in the algorithms should be carried along
with the data throughout the archiving process.
• Timely, regular assessments of the quality and
homogeneity of the instrumental databases are needed, and should
include all data used to monitor past and present environmental
variations and change. Special attention should be given to the
long-term, high-resolution instrumental data required to identify
change or variations in the occurrence of extreme environmental
events.
• Societal and policymaking requirements for knowledge or
observations of environmental variations and change
Page 116
should be taken into account in laying out a strategy for a
global, comprehensive system for observing climate.
• Stations, platforms, and observation systems with long,
uninterrupted records should be maintained. Every effort should be
made to protect the datasets that document long-term homogeneous
observations, particularly those encompassing a century or more.
Priorities for sites or observation systems should be assigned on
the basis of their contribution to long-term monitoring of each
element of the climate system.
• In the design and implementation of new environmental
observing systems, highest priority should be given to data-poor
regions, regions sensitive to change, and key measurements that
currently have inadequate temporal resolution. In addition, the
appropriateness of the variables to be measured should be
verified.
• Network designers, operators, and instrument engineers
must be provided with long-term environmental monitoring
requirements when they begin to design an observing system. Most
observing systems now in place were designed for purposes other
than long-term monitoring, and many of them do not acquire
information in a suitable form. Instruments must have adequate
precision, and their biases must be small enough, to resolve the
environmental variations and changes that are of primary interest
to climate research.
• Much of the development of new observational
capabilities, and much of the evidence supporting the value of
these observations, stemmed originally from research needs or
research-oriented programs. Stable, long-term commitments to these
observation systems, and a clear plan for their transition from
research to operations, are two requirements for the development of
adequate long-term environmental monitoring capabilities.
• Data-management systems that facilitate the use and
interpretation of observational data are essential. Freedom of
access, low cost, mechanisms that encourage use (directories,
catalogs, browsing capabilities, and availability of metadata,
including station histories, algorithm accessibility,
documentation, and so on), and quality control should guide data
management. International cooperation is critical for effective
management and exchange of data used to monitor long-term
environmental variability and change.
An ongoing study by the NRC Panel on Climate Observing Systems'
Status is in the process of identifying and characterizing existing
and emerging factors that could lead to a deterioration in the
quality of climate data or in its availability from operational and
research observational networks. The study will also make
recommendations for maintaining the quality and availability of
climate data.
How Can Other Observations Address
Climate Models?
While detection of climate change is, as discussed above, a
critical problem for instrumental observations, it is not the only
one. Ultimately, the problem of assessing the importance of climate
variability that occurs over decades and longer, and particularly
the task of predicting it, will rest on using physically based
models. It is unlikely that these models can be adequately
validated through comparison of predicted and observed climate
change alone. Such validation would require observing a large
number of examples to separate random unmodeled noise from
simulation skill, and each example would require decades to
observe. Confidence in any understanding of climate variability on
dec-cen time scales must rest on the verified realism of the
components of our models. This confidence cannot be gained from
theoretical arguments alone—each model component must be
tested against appropriate observations. Much of this report
addresses a large number of processes that affect climate and must
be included in models and, therefore, must be verified before the
models can be trusted to simulate climate variability. It would be
exhausting to reiterate these processes and to point out what
observations are needed to verify their model representations.
Rather, it will suffice to emphasize that quantitative verification
is needed, and to review some of the kinds of observations and
studies that will yield the needed climate information:
• Accurate measurements for developing and verifying model
components. A model that quantitatively simulates many of the
important climate processes deserves greater confidence than one
that does not. Thus accurate measurements of such quantities as
air-sea fluxes of heat, water, and momentum at specific sites; the
directional and wavelength distribution of atmospheric radiation
under various meteorological and cloud conditions; the flow of
water from the Pacific to Indian Oceans through the Indonesian
Seas; or the flux of ice from Antarctica into the Southern Ocean
would provide valuable tests of climate models. Measurements of
this type are most likely to come from research projects aimed at
specific processes. This approach was first applied to relatively
small-scale processes that could be studied by a single discipline
in a few weeks of extensive observation. TOGA COARE, however,
provided an example of applying the same philosophy to a
large-scale meteorological and oceanographic process. Studies of
this type may provide the climate information best tuned to
addressing specific climate phenomena.
• Surveys of climate indicators for assessing overall model
performance. The large-scale spatial and seasonal variations of
properties distributed by the climate system provide critical tests
of the transport processes simulated by climate models. For
example, the distribution and rates of accumulation of
anthropogenic or chemically reactive gases and aerosols in the
atmosphere, or the distribution of inorganic carbon and freshwater
in the ocean, are fields whose simulation critically tests both the
source/sink and transport processes in models. Diagnostic
quantities such as the ratios of certain oceanic properties can in
some cases be an even more useful
Page 117
tool for assessments of model simulations than the use of
directly observed quantities. Global surveys of these property
fields are large undertakings, so it will be useful to identify a
few key property fields that will provide the most stringent and
useful tests, and to develop ways of measuring them.
Proxy Data
If we had only the recent instrumental record as our guide, we
might consider the climate system to be relatively stable on
decade-to-century time scales. Instrumental records of atmospheric
and oceanic conditions are too short to adequately define dec-cen
variability: Atmospheric records rarely predate the current
century, and high-quality ocean records are limited to recent
decades. Moreover, the instrumental record is spatially biased;
multidecadal temperature records from the oceans are concentrated
in shipping lanes, and long atmospheric records originate mostly in
western Europe and North America. The existing instrumental record
of climate is thus insufficient to reveal most natural modes of
climate variability on multidecadal to multicentury time scales.
Models can simulate the climate systems over these time scales, but
they must rely on the questionable assumption that the processes
they incorporate behave and interact similarly over a range of time
scales that far exceeds the instrumental baseline.
Fortunately, long-term records of climate exist in paleoclimatic
archives worldwide. They offer the opportunity to extend our
observational baseline into the dec-cen range of the spectrum of
climate variability. Such records provide new information on the
natural variability and sensitivity of climate, and they constitute
an observational basis for evaluating the behavior of the numerical
models used for climate prediction.
Poleoclimatic Contributions
Paleoclimatic records contribute in important ways to a better
understanding of dec-cen climate variability. Simply extending the
record of climate at a particular location makes possible
evaluation of natural variability over time scales not represented
by instrumental data. For example, proxy records of moisture
balance from California and the Great Plains reveal that the
droughts of this century pale by comparison with droughts over the
rest of the millennium as regards both amplitude and duration (Muhs
and Maat, 1993; Madole, 1994; Stine, 1994; Laird et al., 1996).
Such long-term records allow us to place this century's warming in
perspective—in particular, to evaluate whether it is
unprecedented (Bradley and Jones, 1993; Briffa et al., 1995; Jacoby
et al., 1996).
We can use information on past climate variability to assess
whether the short-term (seasonal-to-interannual) modes and patterns
observed today (see Chapter 3) have the same spatial patterns and
global teleconnections as long-term (dec-cen) modes. For example,
seasonal-to-interannual modes such as ENSO are apparently modulated
on dec-cen time scales (Cole et al., 1993; Dunbar et al., 1994).
The African and Asian monsoons probably respond in phase to
large-scale forcing over multicentury time scales (Overpeck et al.,
1996), even though over shorter periods the Asian monsoon weakens
and East African rainfall increases during ENSO warm extremes. In
addition, proxy records of snow accumulation (Alley et al., 1993)
and temperature from ice cores (Dansgaard et al., 1993) can provide
some insight into how rapidly climate regime changes can occur.
Paleoclimate reconstructions also allow us to observe the
response of climate to changes in forcings or boundary
conditions—for example, the relationship between drought and
solar variability, or the behavior of ENSO during periods of warmer
background SST or lower sea level. By improving our understanding
of the natural variability and sensitivity of climate over dec-cen
time scales, paleoclimatic reconstructions enable us to evaluate
model behavior over these time scales (see, e.g., Knutson et al.,
1997). Finally, paleoclimatic archives preserve records of climate
forcings as well as responses (Zielinski et al., 1994; Lean et al.,
1995). Reconstructions of volcanic aerosol loading or solar
variability (for instance) can be incorporated into model
simulations to test the system's responses (Rind and Overpeck,
1993). Fields of SSTs can also be interpolated from point
reconstructions for use in initializing transient simulations with
atmospheric GCMs.
Sources of Proxy Data
Many geologic and biologic archives preserve useful information
on previous climate conditions. The primary sources of information
useful on dec-cen time scales are described below. Certain of these
archives (ice cores, coral reefs, and old-growth trees) are under
threat of destruction from human and environmental influences,
which will spell the loss of valuable climate information.
Ice cores have provided important records of quantities
of atmospheric constituents from both high-latitude and
high-altitude regions. Temperature can often be derived from the
isotopic content of the ice (e.g., Thompson et al., 1995) on the
basis of the relationship between the stable isotopic content of
precipitation and the temperature of condensation, although
large-scale climate features also control significant isotopic
variance over dec-cen time scales (White et al., 1997a). Borehole
temperature measurements permit verification of the
temperature-isotope relationship for periods when precipitation,
season, or other factors alter this dependence (Cuffey et al.,
1995). Accumulation rates allow reconstruction of the hydrologic
balance (Meese et al., 1994). Certain ice cores preserve pristine
air bubbles throughout, making possible unique observations of past
changes in atmospheric greenhouse-gas concentrations (Raynaud et
al.,
Page 118
1993; Etheridge et al., 1996) and yielding chemical ratios that
reflect biogeochemical processes (Whung et al., 1994). Ice cores
preserve records of aerosol loading and chemistry (Zielinski et
al., 1994; O'Brien et al., 1995) that reflect both source strengths
and transport patterns, and ice-core pollen records track regional
vegetation changes (Thompson et al., 1995). Record lengths range
from millennia in many tropical sites to hundreds of millennia in
the coldest regions; resolution is decadal or less at the base of
the oldest cores, because the more deeply buried ice thins under
pressure, but the resolution can be seasonal in records of recent
centuries.
Corals represent a multivariate record of tropical
surface-ocean variability. Oxygen isotopic variations in the
geochemistry of coral skeletons track SST (Fairbanks and Dodge,
1979; Dunbar et al., 1994; Gagan et al., 1994; Wellington et al.,
1996) and salinity where such variations are strong (Cole and
Fairbanks, 1990; Linsley et al., 1994). Certain metals that
substitute for calcium in the aragonite lattice also reflect
temperature (Beck et al., 1992; Shen and Dunbar, 1996; Shen et al.,
1996; Mitsuguchi et al., 1996), while others are incorporated in
proportion to their concentration in surface water, which may be
governed by processes such as upwelling, runoff, or wind mixing
(Shen et al., 1987, 1992; Lea et al., 1989). In some locations,
coral growth rates reflect SST (Lough et al., 1996), and
fluorescent bands in their skeletons may track river discharges
(Isdale, 1984). Radiocarbon concentrations in corals reflect
surface water14C variations, which
make possible the reconstruction of aspects of ocean circulation
over recent centuries (Druffel, 1987, 1997; Druffel and Griffin,
1993). Coral record lengths range from 100 to 800 years, with
weekly to quarter-year resolution depending on growth rate; shorter
records at similar resolution can be found in fossil sequences over
the past 103 to 105 years (Beck et al., 1997; Gagan et
al., 1998).
Tree-ring climate reconstructions are generally derived
from suites of many (10-50) cores from a single site, cross-dated
to provide absolute age control and pooled to eliminate noise
associated with individual tree responses (Fritts, 1976; Cook and
Kairiukstis, 1990). Statistical calibration with local climate,
including independent validation intervals, yields quantitative
records of tree sensitivities to aspects of climate like summertime
drought (Meko et al., 1993; Hughes and Graumlich, 1996), summer
temperature (Briffa et al., 1990, 1995; Cook et al., 1991), or
annual temperature (Jacoby and D'Arrigo, 1989; Jacoby et al.,
1996). Different aspects of tree growth, including ring widths and
the density of various aspects of the wood architecture, can be
measured and calibrated against climate. Tree-ring reconstructions,
which are already widely available over northern mid-latitudes, can
be pooled and gridded to generate reconstructed fields of such
parameters as drought (Cook et al., in press).
Sediments in oceans and freshwater environments preserve
a wealth of information on global, regional, and local climate
variations; when sedimentation rates are sufficiently high, dec-cen
climate variability is recorded. Climate parameters interpreted
from sedimentary evidence include a wide range of physical,
ecological, hydrologic, and chemical aspects particular to a given
site. In anoxic environments, where the uppermost layers of
sediment are not disturbed by burrowing organisms, the sediments
are laid down in annual couplets (varves) that make possible
extremely high-resolution chronology and interpretations (see,
e.g., Overpeck, 1996; Hughen et al., 1996; Behl and Kennett, 1996).
Where sedimentation rates are high but varves do not form, decadal
resolution is achievable (Hodell et al., 1995; Laird et al., 1996).
Sediment records with this resolution can span on the order of
104 years. Signs of lake-level
changes, including geomorphologic features and submerged forest
remains, provide another type of sedimentary evidence for dec-cen
variability of hydrologic controls on lake levels (Stine, 1994).
These are not continuous records, but they offer a snapshot of
hydrologic conditions useful in addressing century-scale climatic
change.
Historical information about climate offers another
source of information on past variability in attributes deemed
important to a given society. Often these proxies are related to
practical considerations, such as droughts, harvest records, fish
catch, death records, the freezing of waterways, or the level of a
river (Quinn, 1993; Frenzel et al., 1994). Although such records
require a transfer function to produce strictly climatological
quantities, their immediate relevance to societally important
attributes can also be considered an advantage.
Other proxy sources of climatic information may prove
useful for understanding dec-cen variability in specific locations,
or as new archives are explored and calibrated. For example, relict
tree stumps rooted in modem Sierra Nevada wetlands (Stine, 1994)
and recently active dune fields in Colorado grasslands (Muhs and
Maat, 1993; Madole, 1995) provide snapshots of considerably dryer
weather in North America during the Holocene. Dating of tropical
and temperate mollusc remains from archaeological sites in Peru has
led Sandweiss et al. (1996) to conclude that ENSO variability was
not present until about 5000 BP. A combination of reconstructions
from ice cores and findings at archaeological sites in Greenland
support the theory that Norse settlements were abandoned because of
colder summertime temperatures associated with the early stages of
the transition into the Little Ice Age which gripped the North
Atlantic region for several centuries (Pringle, 1997).
Calibration of Proxy Records
The extraction of climatological quantities from proxy data
requires an understanding of how the signal is incorporated into
the proxy, and of any competing influences on the record.
Calibration procedures vary among the different archives, and
include both process-based and statistical approaches. For tree
rings, precise chronologies permit the compositing of many
individual records to reduce noise;
Page 119
standard practices include a model-development process that
incorporates both calibration procedures and an independent period
of validation against instrumental data. Although the processes
that control tree growth are understood to the extent that certain
relationships can be expected, statistical calibration provides the
primary basis for most tree-ring reconstructions of climate. For
most other types of archives, poorer replication and chronological
uncertainties mean that compositing dozens of records is not
feasible. Thus a more process-based approach is usually taken,
involving such steps as automatic weather stations set up at
ice-coring sites, automated temperature monitors and
water-collection programs at coral sites, and sediment traps in
regions of sediment coring. The resulting measurements indicate
which processes are relevant to the incorporation of the
paleoclimate signal, and statistical methods are used to evaluate
the degree to which the chosen interpretation is correct.
Complications specific to calibration of paleoclimatic records
include geochronology, biology, and the seasonality of response.
All paleoclimatic data must contend with the issue of assigning
accurate ages; annual precision is achievable in many cases where
annual layers are deposited (particularly tree rings, but also many
corals and ice cores). Age uncertainties for all methods need to be
quantifiable in order to determine the limits of useful
interpretations. Many paleoclimate archives are living organisms,
whose biology can affect how a climate signal is recorded.
Organisms may stop growing during a time of climate-induced stress,
or long-term non-climatic growth trends may exist and must be
removed. Biological processes can cause a consistent offset of a
carbonate skeleton's geochemistry from the thermodynamically
expected concentrations of isotopes and metals. Finally, many
proxies reflect a seasonally specific or weighted response, based
on nonconstant growth or deposition rates throughout the year or
increased sensitivity during certain seasons. These potential
complications need to be recognized if optimal reconstructions are
to be developed.
Data Products
Once calibration has been established, paleoclimate
interpretations produce a variety of types of data products. The
simplest records are histories of past variations at a single site.
These histories may be a snapshot or window on the past that
reveals a scenario different from the modem climate, like the tree
stumps in a lake bed or dune fields under modem grasslands
mentioned earlier. A more typical reconstruction from ice and
sediment cores, corals, and tree rings is a continuous, well-dated
time series from which information about seasonal to centennial
modes of temporal variability can be extracted. Just like a single
instrumental temperature record, a single paleoclimatic record is
of limited value in understanding large-scale climate variability.
However, if the site is sensitive to large-scale variability, or if
many sites are developed, indices of large-scale modes such as the
NAO or ENSO can be reconstructed (Cook et al., 1997).
Taking this approach a step further, the synthesis of results
from many sites can provide a basis for interpolating spatial
fields of climate reconstructions. The development of paleoclimatic
data into fields of climatic quantities improves the interface
between these data and GCM simulations, which produce, and are
usually forced by, such spatial fields. For example, CLIMAP (1981)
produced SST and ice-extent maps for the last glacial maximum that
have been used extensively as boundary conditions and even
validation fields for GCM simulations of that period; similar
approaches can be used for studies of dec-cen variability to assess
the climate response to large forcing changes. Cook et al. (in
press) describe how 388 drought-sensitive tree-ring chronologies
from the continental United States can be combined into a
reconstruction of the annual Palmer Drought Severity Index (PDSI)
for the past 300 years. Overpeck et al. (1997) present a
multi-proxy 400-year temperature reconstruction from the Arctic
that confirms recent unprecedented warming associated with dramatic
environmental changes. EOF-based interpolation techniques can be
used to fill gaps between single-site reconstructions in order to
generate fields from sparser datasets (Kaplan et al., in press;
Mann et al., 1998), although this approach does assume that
dominant EOF patterns are consistent through time. A global gridded
annual-mean time series of, for example, surface temperature and
precipitation for the past 1,000-2,000 years from reconstructed
paleodata would allow the application of standard statistical
techniques for comparison with modem data.
Summary
Proxy data provide a unique contribution to the objectives of a
research program aimed at understanding climate change on dec-cen
time scales. Indeed, there are few other continuous sources of
observations on climate variability that extend beyond the middle
of the nineteenth century. Paleoclimatic reconstructions provide a
test bed for numerical climate models, and suggest new conceptual
models for long-term climate variations. To take full advantage of
this information resource will require broadening support for
climate research in the following ways:
• Cross-disciplinary interaction among those climate
scientists using proxy and instrumental data, and those developing
and using numerical models, needs to be fostered. The combination
of instrumental, paleoclimatic, and modeling approaches can provide
answers to questions about calibration, limitations, and
interpretation that cannot otherwise be resolved for more than the
last hundred years or so.
• Climate-monitoring programs can provide important process
and calibration baselines for paleoclimatic interpre-
Page 120
tations, particularly if such programs are designed from the
start to include aspects of climate that proxy data can record.
• In some parts of the world, potential sources of
paleoclimatic information are disappearing as tropical glaciers
melt and old-growth trees and coral-reef environments are
exploited. Quick efforts to sample these endangered resources will
pay off in new understanding.
• Searches for new proxy records and refinements of
existing ones are needed to expand spatial coverage, provide
measures of additional climate parameters, and clarify the
limitations of the various types of proxies.
• Open access to paleoclimate data can be assured through
continued support for database activities, both to encourage data
submission and to integrate a variety of proxy information into a
single database for climate-variability studies. The World Data
Center-A for Paleoclimatology archives paleodata and makes it
available to the climate-science community. As can be seen from its
website at <<http://www.ngdc.noaa.gov/paleo/paleo.html>,
the WDC-A is continually developing this resource.
• The collection, processing, and interpretation of proxy
data to yield a final climatological product require considerable
effort. These efforts should be vigorously pursued in order to turn
our currently sparse, though tantalizing, picture of past climate
changes into a more complete set of time-varying spatial fields.
This is the only currently available means by which we can advance
our understanding of dec-cen variability from an observational
perspective, and it should be fully exploited.
Analysis Products and Model
Output
A special hybrid climate-information product involves the
interpolation of climatic datasets into spatially consistent
fields, through the use of climate modeling and assimilation
techniques. In essence, the inaccurate, irregularly spaced, and
often sparse observations are blended with a model simulation to
produce a globally consistent sequence of climatological fields. In
particular, major numerical and simulation centers, such as the
European Centre for Medium-Range Weather Forecasting, NOAA's
National Centers for Environmental Prediction, and NASA's Data
Assimilation Office, have been carrying out retrospective analysis
(''re-analysis'') projects over the past few decades, using a
single model and data-assimilation scheme (Kalnay et al., 1996;
Todling et al., 1998) to reconstruct climate evolution since World
War II. These re-analysis products can be used as spatially and
temporally interpolated data. Alternatively, statistical techniques
such as optimal interpolation can be used, in which the spatial
covariance structure is created from the observed data and then
used to fill in spatial gaps that preserve this structure (assuming
that the structure has not changed).
These methods allow us to expand the spatial and temporal
coverage of the data in a way that is completely consistent with
physical laws and observed data relationships. In this manner, the
data are more accessible for incorporation into models or
comparison with model output, facilitating the integration of
models and observations that will be critical to the success of
dec-cen-scale research. To the same end, it is important that
observational data and model output be subjected to the same kind
of processing (e.g., smoothing or gridding) so that they can be
compared on an equal footing. Note too that if the instrumental
record is to grow in a regular and systematic way, a single
re-analysis will not suffice. A new re-analysis of the entire
record must be performed every 5 to 10 years, using the best
available model and any additional data that have become available
since the previous re-analysis.
Coupled-Model Development and
Infrastructure
All scientific understanding is crystallized, and all
predictions are made, using models. They may be descriptive and
intuitive, or more formal mathematical ones. Climate dynamics
currently employs models that embody equations reflecting our
admittedly incomplete knowledge of the physical, chemical, and
biological laws that govern each of the climate system's
subsystems, and also their interactions. These models vary in their
degree of completeness and detail; some are very simple,
mechanistic ones, whereas others are highly elaborate and require
huge resources in people and computing power. Models are used to
assimilate incomplete, imperfect, and irregularly distributed
observations; to simulate known climate phenomena; to discover or
help understand new, as yet unknown phenomena; and to predict the
climate system's evolution over time. This section briefly reviews
existing models, and then describes a model configuration that
should permit achievement of the major goals of a research program
in decade-to-century-scale climate variability.
Existing Models
Models for each of the climate system
components—atmosphere, biosphere, cryosphere, and
hydrosphere—and their predictive capabilities have been
presented in the relevant sections of Chapter 4. This information
is merely integrated and summarized here as a basis for building
the model configuration described in the next subsection.
The modeling of each one of the climate system components is at
a slightly different point in its evolution, both from simple to
more detailed models and from the study of one model or class of
models to an organic integration of all relevant models. Only a
combined analysis of all these models with each other and with the
observations can provide all the different types of information
needed about the (sub)-system. The models in use range from scalar
evolution equations to general-circulation models. A scalar
equation can
Page 121
describe, for instance, the evolution of global surface air
temperature due to various heat sources and sinks and climatic
feedbacks, both positive and negative. The GCMs use numerically
solvable forms of the most complete systems of evolution equations
that describe local variations in the temperature, pressure,
velocity, and composition of the atmosphere, ocean, and other
subsystems. They have been successfully utilized for data
assimilation, simulation of various atmospheric and oceanic
phenomena, and weather prediction.
No single class of model can satisfy all needs. The GCMs'
spatially detailed, local description of phenomena and their
short-term variability limits these models' ability to provide such
a description for sufficiently long time intervals and sufficiently
broad ranges of the many unknown parameters that enter the
governing equations. Simple models are flexible enough to explore
broad swaths of parameter space and study variability on the long
time scales of interest for dec-cen climate variability, but they
can be evaluated against observations only through the use of
intermediate models with greater spatial and physical detail. Such
intermediate models describe variations in latitude or height only,
in latitude and height, or in latitude and longitude, while
averaging or integrating over the directions not explicitly
included.
For the atmosphere, experience with GCMs as well as with the
systematic use of the full hierarchy of models mentioned above is
extensive, and can provide a pattern for similar usage in the other
subsystems. Experience with ocean, cryosphere, biosphere, and
continental hydrosphere models in climate studies is considerably
less, but is rapidly reaching a similar level of
sophistication.
Models that consider the tropical ocean coupled with the global
atmosphere have played a key role in the advancement of coupled
models, and have also played an important role in the TOGA program.
TOGA has clearly had a great impact on the rapid and successful
development of such models (NRC, 1996). A full hierarchy of
TOGA-type models, from simple through intermediate to coupled GCMs
(CGCMs), is now being used in data assimilation, in simulation of
air-sea interactions, and in experimental but routine
seasonal-to-interannual prediction. CGCMs and simple and
intermediate coupled models for the global or mid- to
-high-latitude oceans are also evolving rapidly. Models for snow,
surface hydrology, and near-surface vegetation have already
evolved, at least in part, as modules or appendices of atmospheric
GCMs.
A Target Modeling Structure
As noted earlier, the enterprise of monitoring, understanding,
and eventually predicting climate variability on dec-cen time
scales is inseparable from that of modeling. Thus, the organized
structure (or hierarchy) of coupled models described below is part
of the process of building our understanding, and should evolve
with it. A substantial part of the structure needs to be set up
early in this process, and most of it should be in place by the
time a concerted scientific program is ramped up.
At the top of this organized structure there should be at least
one modular CGCM, with a very flexible interface between the
subsystem models, or "modules," of which it is composed, and an
architecture that permits an easy interchange of the modules. The
complexity inherent in the climate system's behavior on the dec-cen
scale will require the participation of the entire climate-dynamics
community, at universities and at government laboratories, in the
modeling enterprise. The interchangeable-module architecture is
certainly one that would permit all members of this community, from
graduate students to senior researchers, to distill their evolving
knowledge of the subsystems into improved modules, and to use the
full power of the CGCM in a configuration well adapted to their
specific area of application. While a modular approach would have
significant advantages, it is recognized that to obtain maximum
computational efficiency, the module's computer-program codes may
have to be optimized for specific machines.
This modular CGCM (or CGCMs) needs to be supported by a full
hierarchy of simple and intermediate models of each subsystem and
of subsets of the entire system. For example, an intermediate model
of the global ocean or of a particular ocean basin may be coupled
to a simple model of the global atmosphere, as one rung in such a
hierarchy. This complementarity is required by the learning process
involved in developing better modules for CGCMs, as well as by the
need to evaluate CGCM performance and build knowledge of climate
variability and its predictability on dec-cen time scales.
This target structure, in turn, requires adequate resources in
data, people, and computing. The observations described in the
preceding section are necessary for model evaluation and eventual
climate prediction. Observations have to be combined, by the
process of data assimilation using the models, to provide
physically consistent descriptions of past and present climate
change, as well as to permit the prediction of the future from an
optimally described present.
A new generation of scientists needs to be educated to deal with
the enormous complexities of modeling the entire climate system and
predicting its behavior. These scientists will have to master not
only the traditionally relevant disciplines of dynamic meteorology
and physical oceanography, but also important aspects of analytic
chemistry and biochemistry, ecology, and other biological
disciplines. They must be at ease with small and large
models—that is, have both theoretical and numerical skills, as
well as a high degree of sophistication in the use of the most
modern computing, communications, and observing technology. To
achieve this, their education will have to bridge not only
disciplines but also institutions.
The resources available for climate research will have to
Page 122
include computing, communications, and multimedia support at the
highest levels of performance provided by current technology. A
minimal network should incorporate at least one fully dedicated
device capable of the highest throughput available at a single site
at any given time. It should include communication links capable of
sustaining information flow at a level consistent with this device.
These resources will enable the various members of the community to
operate as if housed on a single campus. Powerful workstations will
be needed at many locations to support work on: simple and
intermediate models; module development, testing, and applications;
and analysis of simulations and experimental predictions with the
modular CGCM (or CGCMs). Both central and distributed memory will
have to be sufficient for the observations gathered, as well as for
model simulations, model-assimilated datasets, and experimental
predictions. Finally, hypertext information exchange—text,
datasets, still and moving images—should be supported to
facilitate the integration by many scientists of the rich
information gathered and generated about the climate system, its
components, and its decade-to-century-scale variability.
Detection, Attribution, and
Simulation
The global-mean surface air temperature has increased by about
0.5ºC since the mid-nineteenth century, as was indicated in
Figure 2-6. This increase is not monotonic; there is a fairly rapid
rise from 1910 to 1940 and again from the mid-1970s to the present,
and some decrease in the 1940s and 1960s-1970s. (The increase was
more uniform in the Southern Hemisphere than in the Northern
Hemisphere, where the cooling between 1960 and 1980 was more
pronounced.) It has been suggested that the overall warming trend
can be attributed to the observed increase in greenhouse gases,
including CO2; the lack of
monotonicity in the increase must then be interpreted as the result
of natural climate variability or other forcings. Recent modeling
studies indicate that it may be necessary to consider in particular
the cooling effect of anthropogenic sulfate aerosols, which reflect
solar radiation, in order to explain quantitatively the magnitude
of the observed warming (IPCC, 1996a). However, the inclusion of
sulfate aerosols is not sufficient to account for all of the
variations in globally averaged surface temperature over the last
century, especially the rapid warming during the 1920s and early
1930s.
One of the most important goals of dec-cen research is the
reliable projection of future climate change (e.g., global warming)
through the use of coupled ocean-atmosphere-land models. The best
way of lending credence to future model predictions of climate is
to successfully simulate past climate change. In order to assess
how realistic the sensitivity of a climate model is to external
forcing, it is necessary to simulate observed long-term change of
climate by driving the model with time series of actual thermal
forcings, such as increased concentrations of greenhouse gases and
aerosols in the atmosphere. The performance of the model can then
be evaluated by comparing the simulated and the observed long-term
climate changes. The large uncertainties about the believability of
model projections of future climate change must be reduced by
assessing the models through the hindcast of climate, as described
above. To carry out such a hindcast, long-term observational data
for driving the model are essential. The highest priority should be
given to:
(1) Obtaining reliable, long-term observations of factors that
change thermal forcing, which drives the long-term variations of a
climate model. These factors include not only the increase in
greenhouse gases, but also changes in solar irradiance and in
aerosol loading in the atmosphere.
(2) Making reliable, long-term observations of a carefully
chosen set of basic climatic variables. Important variables include
spectra of radiative fluxes at the top of the atmosphere, as well
as satellite-observable radiances that are indicators of levels of
cloudiness, snow cover, sea ice, vegetation, and possibly soil
wetness. Other excellent candidates for long-term observation are
sea level, surface salinity, and the water-mass structure of the
oceans. The monitoring of variables such as total carbon content,
alkalinity, and the partial pressure of CO2 will also be valuable for reliably
projecting the future increase of atmospheric CO2.
(3) Reconstructing past changes of the variables listed above.
This task will require comprehensive compilations of historical and
other proxy data, with careful attention to their quality.
In order to evaluate a climate model through the hindcast
strategy described above, it is also necessary to distinguish the
anthropogenic change of climate from its natural, internally
generated fluctuations. For this purpose, we need to improve our
knowledge of internally generated (rather than externally forced)
climate variability, with a view to separating the anthropogenic
changes from the natural variations. The latter seem to contain
quasi-periodic, and hence predictable, components with interannual
and interdecadal periods (Plaut et al., 1995; Mann et al.,
1995b).
Several groups have recently achieved some success in exploring
this topic by analyzing the climate variability from very-long-term
(thousand-year) integrations of coupled ocean-atmosphere-land
models. For example, Stouffer et al. (1994) and Manabe and Stouffer
(1996) have found that (with the notable exception of the tropical
eastern Pacific, where SST anomalies are underestimated) their
model approximates the standard deviation of annual mean surface
air temperature and its geographical distribution. The model also
simulates the broad-band spectrum of ob-
Page 123
served global-mean surface air temperature, from interannual to
interdecadal time scales. However, it fails to reproduce the
warming trend of centennial time scale (i.e., ~0.5ºC per
century) that has been observed since the end of the last century.
If the model is assumed to be realistic—in spite of its
failure to reproduce the quasi-periodic components of the natural
variability—this result suggests that the observed
centennial-scale warming trend is not generated within the climate
system by nonlinear interaction among the atmosphere, ocean, and
continental surface. Instead, the trend must be caused by a
sustained trend in natural and/or anthropogenic thermal forcing,
such as changes in solar irradiance, greenhouse gases, and aerosol
loading in the atmosphere. Essentially similar results have also
been obtained from the long-term integration of a coupled model at
the U.K. Meteorological Office (see, e.g., Mitchell et al.,
1995).
Identifying predominant patterns associated with the natural,
internally generated climate variability would aid in the detection
of patterns of anthropogenic change (see, for example, Barnett and
Schlesinger, 1987; Hasselmann, 1993; and Santer et al., 1995). If
the effect of sulfate aerosols is considered together with the
effect of greenhouse gases in GCMs, the spatial distribution of the
model-generated change of atmospheric temperature over the decadal
time scale appears to become more realistic (Santer et al., 1996).
These and other recent results (see, e.g., IPCC, 1996a, for an
overview) are leading to a more reliable estimate of the
anthropogenically induced climate change, as well as of the natural
variability caused by mechanisms internal to the climate
system.
In the future, major effort will need to be devoted to
observational and modeling studies of internally generated climate
variability, so that this variation can be distinguished from
anthropogenic climate change. Records from past observations of
both ocean and the atmosphere should be compiled and analyzed for
variables such as concentration of greenhouse gases in ice cores,
sea-level pressure, surface and subsurface temperature and salinity
in oceans, air temperature and humidity at the surfaces, and
temperature and geopotential height at selected pressure levels in
the atmosphere. It is also essential to improve model
parameterizations of various feedback processes, in particular
those involving cloud, snow, and sea-ice cover, all of which
substantially affect incoming solar and/or outgoing terrestrial
radiation at the top of the atmosphere. Other factors of critical
importance are cumulus convection and land-surface heat and water
budgets. Greater use of data from remote sensing and in situ
measurements of radiative emissions and river runoff facilitate
evaluating and improving the parameterizations of the important
processes identified above.
Linkage Across Time Scales
As was noted in Chapter 4, there are practical reasons for
dividing the study of climate variations by the time scales on
which they occur. The climate system clearly evolves over a
continuum of time scales, however, and no "spectral gap" in nature
justifies such a separation. To advance our understanding of
overall climate change and variability most efficiently, it is
important that we explicitly recognize those processes that cannot
easily be categorized by scale, and that we particularly emphasize
those mechanisms which affect climate variability and change over a
range of time scales.
A few specific examples of climate-variability patterns and
their possible causal mechanisms that appear on more than one time
scale are (1) the interdecadal variability of ENSO, in amplitude,
periodicities, and warm or cold anomaly distribution; (2) the North
Atlantic Oscillation (NAO) and its purely atmospheric, purely
oceanic, or coupled mechanisms; and (3) changes in the carbon
cycle, over land, ocean, and the tropical or mid-latitude or polar
regions. The modes of variability that cross two or more time
scales can arise either from the intrinsically broad-band behavior
in time of a specific spatial mode, or from the nonlinear coupling
between narrow-band spatio-temporal modes that share certain
regional characteristics. Which one of these overall types of
behavior is at the root of a given dec-cen climate phenomenon has
important implications for its predictability.
Currently, the national and international organizations devoted
to the study of physical climate are structured to address
separately the high-frequency variability (GEWEX),
seasonal-to-interannual variability (GOALS),
decade-to-century-scale variability (CLIVAR DecCen), and millennial
and longer-scale variability (e.g., PAGES). Each of these groups
has identified a suite of high-priority issues that must be
addressed. Many of the detailed processes involved in these issues
are common to all four units. For example, improved understanding
of air-sea exchanges is of fundamental importance to the study of
climate, regardless of time scale. Similarly, the patterns of
climate variability and the coupled modes are of equal importance
to all groups, because their regional manifestations occur on a
broad range of time scales. Issues related to these common
processes and patterns warrant particular attention, and a dec-cen
program that is highly coordinated with GOALS and GEWEX would
enable them to be studied most effectively. Furthermore, the
physically based studies of climate must be fully integrated with
those investigating the chemical-biological aspects, which are
currently being addressed by elements of the International
Geosphere-Biosphere Programme.