An Overview of Dangerous Climate Change

Stephen H. Schneider and Janica Lane

Stanford University, Stanford, California
Abstract
This paper briefly outlines the basic science on climate change, as well as the IPCC assessments on emissions
scenarios and climate impacts, to provide a context for the topic of key vulnerabilities to climate change. A
conceptual overview of dangerous climate change issues and the roles of scientists and policy makers in this
complex scientific and policy arena are suggested, based on literature and recent IPCC work in progress.
Literature on assessments of dangerous anthropogenic interference with the climate system is summarized,
with emphasis on recent probabilistic analyses.

Introduction
Europes summers to get hotter, The Arctics ominous thaw, Study shows warming trend in Alaskan
Streams, Lake Tahoe Warming Twice as Fast as Oceans, Global Warming Seen as Security Threat, Global
warming a bigger threat to poor, Tibets glaciers heading for meltdown, Climate change affects deep sea
life, UK: Climate change is costing millions. These are just a few of the many headlines that crossed the wires
in 2004, and they have elicited widespread concern, even in the business community. 2004 is thought to have
been the fourth warmest year on record and the worst year thus far for weather-related disasters claims. Munich
Re, the largest reinsurer in the world, recently stated that it expects natural-disaster-related damages to
increase exponentially in the near future, and it attributes much of these damages to anthropogenic climate
change. Thomas Loster, a climate expert at Munich Re, says: We need to stop this dangerous experiment
humankind is conducting on the Earths atmosphere.
Dangerous has become something of a clich when discussing climate change, but what exactly does it
mean in that context? This paper will explore some basic concepts in climate change, how they relate to what
might be dangerous, and various approaches to characterizing and quantifying dangerous anthropogenic
interference [DAI] with the climate system [68]. It will also outline and differentiate the roles of scientists and
policymakers in assessing what dangerous climate change entails. It will discuss current scientific attempts at
assessing elements of dangerous climate change and will suggest ways in which such science can be
translated into policy. It will also state explicitly that determination of acceptable levels of impacts or what
constitutes danger are deeply normative, involving value judgments that must be made by decision makers,
though scientist have a major role in providing analysis and context.

Climate Change: A Brief Primer

We will begin by stressing the well-established principles in the climate debate before turning to the
uncertainties and more speculative, cutting-edge science debates. First, the greenhouse effect is empirically
and theoretically well-established. The gases that comprise Earths atmosphere are semi-transparent to solar
energy, allowing about half of the incident sunlight to penetrate the atmosphere to Earth's surface. The surface
absorbs the heat, heats up and/or evaporates liquid water into water vapor, and also re-emits energy upward as
infrared radiation. Certain naturally-occurring gases and particles particularly clouds absorb most of the
infrared radiation. The infrared energy that is absorbed in the atmosphere is re-emitted, both up to space and
back down towards the Earths surface. The energy channeled towards the Earth causes its surface to warm
further and emit infrared radiation at a still greater rate, until the emitted radiation is in balance with the
absorbed portion of incident sunlight and the other forms of energy coming and going from the surface. That is
what accounts for the 33 C difference between the Earths actual surface air temperature and that which is
measured in space as the Earths radiative temperature. Nothing so far is controversial. More controversial is

the extent to which non-natural (i.e., human) emissions of greenhouse gases have contributed to climate
change, how much we will enhance future disturbance, and what the consequences of such disturbance could
be for social and environmental systemsin short, the extent to which human alterations could risk DAI.
It is also well-known that humans have caused an increase in radiative forcing. In the past few centuries,
atmospheric carbon dioxide has increased by more than 30 percent. The reality of this increase is undeniable,
and virtually all climatologists agree that the cause is human activity, predominantly the burning of fossil fuels.
To a lesser extent, deforestation and other land-use changes and industrial and agricultural activities like
cement production and animal husbandry also contribute to greenhouse gas buildups. Most mainstream climate
scientists agree that there has been an anomalous rise in global average surface temperatures since the time of
the Industrial Revolution. Earths temperature is highly variable, with year-to-year changes often masking the
overall rise of approximately 0.7 C that has occurred since 1860, but the twentieth century upward trend is
obvious, as shown in Figure 1, below. Especially noticeable is the rapid rise at the end of the twentieth century.
Indeed, the period from the 1980s onwards has been estimated to be the warmest period in the last 2,000
years. Mann and Jones, 2003 [35]; Mann, Bradley, and Hughes, 1998 [34]; and Mann, Bradley, and Hughes,
1999 [33] have attempted to push the Northern Hemisphere temperature record back 1,000 years by performing
a complex statistical analysis involving some 112 separate indicators related to temperature. Although there is
considerable uncertainty in the millennial temperature reconstruction, the overall trend shows a gradual
temperature decrease over the first 900 years, followed by a sharp upturn in the twentieth century. That upturn
is, of course, a compressed representation of the real (thermometer-based) surface temperature record of the
last 150 years. Though there is some on-going dispute about details in the medieval period (e.g., [70]), many
replicate studies confirm the basic picture of unusual warming in the past three decades compared to the past
millennium [71].

1.

Figure 1 Explaining temperature trends using natural and anthropogenic forcing.

It is likely that human activities have caused a discernible impact on observed warming trends. There is a
high correlation between increases in global temperature and increases in carbon dioxide and other greenhouse
gas concentrations during the era, from 1860 to present, of rapid industrialization and population growth. As
correlation is not necessarily causation, what other evidence is there about anthropogenic CO2 emissions as a
direct cause of recent warming? As the data summarized in Figure 1 suggest, it looks increasingly like the
answer is yes. The figure suggests that the best explanation for the global rise in temperature seen thus far is
obtained from a combination of natural and anthropogenic forcings. Although substantial, this is still

circumstantial evidence. However, many recent fingerprint analyses have reinforced these conclusions (i.e.,
[58], [19], [48], [54], and [57]).

Climate Change Scenarios

Since the climate science and historical temperature trends summarized by the IPCC clearly show highly
likely direct cause and effect relationships, we must now ask how climate may change in the future. Scientists,
technologists, and policy analysts have invested considerable effort in constructing storylines of plausible
human demographic, economic, political, and technological futures from which a range of emissions scenarios
can be described, the most well-known being the Intergovernmental Panel on Climate Changes (IPCC) Special
Report on Emissions Scenarios (SRES), published in 2000. One grouping is the A1 storyline and scenario
family, which describes a future world of very rapid economic growth, global population that peaks in midcentury and declines thereafter, and, in several variations of it, the rapid introduction of new and more efficient
technologies. Major underlying themes are convergence between regions, capacity-building, and increased
cultural and social interactions, with a substantial reduction in regional differences in per capita income. A1 is
subdivided into A1FI (fossil intensive), A1T (high-technology), and A1B (balanced), with AIF1 generating the
most carbon dioxide (CO2) emissions and A1T the least (of the A1 storyline, and the second lowest emissions
of all six marker scenarios). But even in the A1T world, CO2 still at least doubles by 2100, as is true for all of the
SRES scenarios.
For a contrasting vision of the worlds social and technological future, SRES offers the B1 storyline, which is
(marginally) the lowest-emissions case of all the IPCCs scenarios. The storyline and scenario family is one of a
converging world with the same global population as A1, peaking in mid-century and declining thereafter, but
with more rapid change in economic structures towards service and information economies, which is assumed
to cause a significant decrease in energy intensity. The B1 world finds efficient ways of increasing economic
output with less material, cleaner resources, and more efficient technologies. Many scientists and policymakers
have doubted whether a transition to a B1 world is realistic and whether it can be considered equally likely when
compared to the scenarios in the A1 family. Unfortunately, the IPCC did not discuss probabilities of each
scenario (e.g., see the debate summarized by [14]). Figure 2 is illustrative of the SRES scenarios.

Climate Change Impacts

After producing the SRES scenarios, the IPCC released its Third Assessment Report (TAR) in 2001, in
which it estimated that by 2100, global average surface temperatures would rise by 1.4 to 5.8C. While warming
at the low end of this range would likely be relatively less stressful, it would still be significant for some unique
and valuable systems [25]. Warming at the high end of the range could have widespread catastrophic
consequences, as a temperature change of 5 to 7C on a globally averaged basis is about the difference
between an ice age and an inter-glacial period occurring in merely a century [7]. In addition, if the IPCCs
projections prove reasonable, the global average rate of temperature change over the next century or two would
likely exceed the average rates sustained over the last 10,000 years [63].
Based on these temperature forecasts, the IPCC has produced a list of likely effects of climate change,
most of which are negative (see [25]). These include more frequent heat waves (and less frequent cold spells);
more intense storms (hurricanes, tropical cyclones, etc.), and a surge in weather-related damage; increased
intensity of floods and droughts; warmer surface temperatures, especially at higher latitudes; more rapid spread
of disease; loss of farming productivity and/or movement of farming to other regions, most at higher latitudes;
rising sea levels, which could inundate coastal areas and small island nations; and species extinction and loss
of biodiversity. On the positive side, the literature suggests longer growing seasons at high latitudes and the
opening of navigation in the normally ice-plagued Arctic. Weighing these pros and cons is the normative (valueladen) responsibility of policymakers, as discussed extensively below.
The IPCC also suggested that, particularly for rapid and substantial temperature increases, climate change
could trigger surprises: rapid, non-linear responses of the climate system to anthropogenic forcing, thought to
occur when environmental thresholds are crossed and new (and not always beneficial) equilibriums are
reached. Schneider et al. (1998) [64] took this a step further, defining imaginable surprises events that could
be extremely damaging but are not truly unanticipated which could include a large reduction in the strength or
possible collapse of the North Atlantic thermohaline circulation (THC) system, which could cause significant

cooling in the North Atlantic region, with both warming and cooling regional teleconnections up- and
downstream of the North Atlantic; and deglaciation of polar ice sheets like Greenland or the West Antarctic,
which would cause (over many centuries) many meters of additional sea level rise on top of that caused by the
thermal expansion from the direct warming of the oceans [59]. There is also the possibility of true surprises,
events not yet currently envisioned [64]. However, in the case of true surprises, it is still possible to formulate
imaginable conditions for surpriselike rapidly forced climate change, since the faster the climate system is
forced to change, the higher the likelihood of triggering abrupt non-linear responses [28]. Potential climate
change, and more broadly, global environmental change, faces both types of surprises because of the
enormous complexities of the processes and interrelationships involved (such as coupled ocean, atmosphere,
and terrestrial systems) and our insufficient understanding of them individually and collectively (e.g., [21]).

Source: IPCC Synthesis Report, 2001.

Figure 2 SRES emissions scenarios.

Many systems have been devised for categorizing climate change impacts. IPCC (2001b) [25] has
represented impacts as reasons for concern, as in Figure 3, below. These impacts are: risks to unique and
threatened systems, risks associated with extreme weather events, the distribution of impacts (i.e., equity
implications), aggregate damages (i.e., market economic impacts), and risks of large-scale singular events
(e.g., surprises). Leemans and Eickhout (2004) [32] have also suggested including risks to global and local

ecosystems as an additional reason for concern, though this could be partially represented under the first
reason for concern. The figure, also known as the burning embers diagram, shows that the most potentially
serious climate change impacts (the red colors on the figure) typically occur after only a few degrees Celsius of
warming.
Parry et al.s (2001) [49] millions at risk work suggests another approach. The authors estimate the
additional millions of people who could be placed at risk as a result of different amounts of global warming. The
risks Parry et al. focus on are hunger, malaria, flooding and water shortage. Similarly, the 2002 Johannesburg
World Summit on Sustainable Development (WSSD) came up with five key areas to target for sustainable
development: water, energy, health, agriculture, and biodiversity (WEHAB). These categories, with the addition
of coastal regions (as proposed by [51]), are also well-suited to grouping climate change impacts.

Source: IPCC, 2001b

Figure 3 IPCC reasons for concern about climate change impacts.

In looking at climate impacts from a justice perspective, Schneider and Lane (2005) [61] propose three
distinct areas in which climate change inequities are likely to be significant: inter-country equity,
intergenerational equity, and inter-species equity. (Schneider and Lane and others have also suggested intranational inequity of impacts.) Another justice-oriented impacts classification scheme is Schneider et al.s (2000)
[62] five numeraires: market system costs in dollars per ton Carbon (C); human lives lost in persons per ton C;
species lost per ton C; distributional effects (such as changes in income differentials between rich and poor) per
ton C; and quality of life changes, such as heritage sites lost per ton C or refugees created per ton C. Lane,
Sagar, and Schneider (2005) [31] propose examining not just absolute costs in each of the five numeraires, but
relative costs as well in some of them:
we should consider market-system costs relative to a countrys GDP, species lost relative
to the total number of species in that family, etc. Expressing impacts through the use of such
numeraires will capture a richer accounting of potential damages and could help merge the
often-disparate values of different groups in gauging the seriousness of damages. In other
cases, such as human lives lost, we believe that the absolute measure remains more
appropriate.

It our strong belief that such broad based, multi-metric approaches to impacts categorization and
assessment are vastly preferable to focusing solely on market categories of damages, as is often done
by traditional cost-benefit analyses. One-metric aggregations can underestimate the seriousness of
climate impacts. Evidence for this was gathered by Nordhaus (1994a) [41], who surveyed conventional
economists, environmental economists, atmospheric scientists, and ecologists about estimated climate
damages. His study reveals a striking cultural divide across natural and social scientists who
participated in the study. Conventional economists surveyed suggested that even extreme climate
change (i.e., 6C of warming by 2090) would not likely impose severe economic losses, implying it is
likely to be cheaper to emit more in the near term and worry about cutting back later, using additional
wealth gained from near-term emitting to fund adaptation later on. Natural scientists estimated the total
economic impact of extreme climate change, much of which they assigned to non-market categories, to
be 20 to 30 times higher than conventional economists projections. In essence, the natural scientists
tended to respond that they were much less optimistic that humans could invent acceptable substitutes
for lost climatic services (see[56]).
Because they typically only measure market impacts, traditional cost-benefit analyses are often
considered skewed from a justice perspective. In a traditional CBA, the ethical principle is not even
classical Benthamite utilitarianism (greatest good for the greatest number of people), but an aggregated
market power form of utilitarianism (greatest good for the greatest number of dollars in benefit/cost
ratios). Thus, an industrialized country with a large economy that suffered the same biophysical climate
damages as an unindustrialized nation with a smaller economy would be considered to have suffered
more by virtue of a larger GDP loss and would, in the aggregate-dollars-lost metric, be more important
to rescue and/or rehabilitate, if possible. Even more problematic, what if an industrial northern country
experienced a monetary gain in agriculture and forestry from global warming due to longer growing
seasons, while at the same timeas much of the literature suggestsless developed southern
countries suffered from excessive heating that amounted to a monetary loss of the same dollar value as
the gain in the north? This could hardly be viewed as a neutral outcome despite a net welfare change
of zero (derived from summing the monetary gain in the north and the loss in the south). Very few would
view a market-only valuation and global aggregation of impacts in which the rich get richer and the poor
get poorer as a result of climate change as an ethically neutral result.
Under the framework of the five numeraires and other systems that rely on multiple metrics, the interests of
developing countries and the less privileged within nations would be given a greater weight on the basis of the
threats to non-market entities like biodiversity, human life, and cultural heritage sites. Take the example of
Bangladesh: Assume that rising sea levels (caused by climate change) lead to the destruction of lives, property,
and ecosystems equivalent to about 80% of the countrys GDP. While the losses would be indisputably
catastrophic for Bangladesh, they would amount to an inconsequential 0.1% of global GDP [25], causing a
market-aggregation-only analysis to classify the damage as relatively insignificant, though a reasonable
interpretation of many would be that such a loss clearly qualifies as DAI. Those considering multiple
numeraires, on the other hand, would argue that this is clearly unfair, as the loss of life, degraded quality of life,
and potential loss of biodiversity are at least as important as aggregate market impacts.

Dangerous Climate Change

What, then, is dangerous climate change? The term was introduced in the 1992 United Nations
Framework Convention on Climate Change (UNFCCC), which calls for stabilization of greenhouse gases to
prevent dangerous anthropogenic interference with the climate system [68]. The Framework Convention
further suggests that: Such a level should be achieved within a time frame sufficient

to allow ecosystems to adapt naturally to climate change;

to ensure that food production is not threatened and;
to enable economic development to proceed in a sustainable manner.

While it seems that some of the impacts of climate change discussed thus far suggest that dangerous levels of
climate change may occur, the UNFCCC never actually defined what it meant by dangerous.

Many metrics for defining dangerous have been introduced in recent years, and most focus on the
consequences (impacts) of climate change outcomes. From an equity perspective, it can be argued that any
climate change that impacts more upon those who contributed the least to the problem is less just and thus
arguably more dangerous and could have repercussions that extend beyond environmental damages (to
security, health, and economy, for example). Along similar lines, scientists defined dangerous anthropogenic
interference at COP10 in Buenos Aires in December 2004 by assessing the key vulnerabilities with regard to
climate change. In the IPCC TAR, vulnerability was described as a consequence of exposure, sensitivity, and
adaptive capacity (Glossary, [25]). As proposed in Working Group 2s working draft on Chapter 19 for the
forthcoming IPCC Fourth Assessment Report (AR4), key vulnerabilities are defined as:
(a) a valued property of the coupled human-natural system that is associated with large
impact risks (such as the sensitivity of the thermohaline ocean circulation to
anthropogenic climate change), or
(b) an impact risk that is considered salient to specific sectors, regions or social groups (such
as increased flood risks in coastal regions or the extinction of species). Key vulnerabilities
describe those interactions between elements of the climate system, climate-sensitive
resources and the services provided by them that may involve significant adverse
outcomes that are considered in the literature to be significant enough in terms of their
ecological, social, and economic implications to be relevant to the determination of DAI.
[23]
(Please note, that all references to the current text or perspectives of Zero Order Drafts of the AR4 do
not necessarily represent what will emerge from the many rounds in the next few years of review and
governmental approvals of content. They cannot be quoted as IPCC results, and are only useful here to
illustrate conceptual issues.)
The notion of key vulnerabilities was derived partly from the discussion on concepts of danger that
occurred at the European Climate Forums (ECF) symposium on Key vulnerable regions and climate change
in Beijing in October 2004 and was presented at COP 10. The ECF symposium identified three concepts of
danger:

Determinative dangers are, on their own, enough to define dangerous levels of climate change. The
ECFs list of determinative dangers resulting from climate change include: circumstances that could
lead to global and unprecedented consequences, extinction of iconic species or loss of entire
ecosystems, loss of human cultures, water resource threats, and substantial increases in mortality
levels, among others.
Early warning dangers are dangers already present in certain areas that are likely to spread and
worsen over time with increased warming. These dangers could include Arctic Sea ice retreat, boreal
forest fires, and increases in frequency of drought, and they could become determinative over time or
taken together with other dangers.
Regional dangers are widespread dangers over a large region, most likely related to food security,
water resources, infrastructure, or ecosystems. They are not considered determinative, as they are
largely confined to a single region. [12]

Dessai et al. (2004) [10] also focus on vulnerabilities as an indicator of dangerous climate change. They
have separated definitions of danger into two categories: those derived from top-down research processes and
those derived from bottom-up methods. The more commonly used top-down approach determines physical
vulnerability based on hierarchical models into which are inputted different scenarios of socio-economic change,
whereas the bottom-up approach focuses on the vulnerability and adaptive capacity of individuals or groups,
which leads to social indications of potential danger like poverty and/or lack of access to healthcare, effective
political institutions, etc.
While scientists have many ideas about what vulnerabilities may be considered dangerous, it is a common
view of most natural and social scientists that it is not the direct role of the scientific community to define what
dangerous means. Rather, it is ultimately a political question because it depends on value judgments about
the relative salience of various impacts and how to face climate change-related risks and form norms for
defining what is acceptable [60, 37]. In fact, the notion of key vulnerabilities itself is also a value judgment, and

different decision makers at different locations and levels are likely to perceive vulnerabilities and the concept of
dangerous in distinct ways. Dessai et al. (2004) [10] explain the juxtaposition of science and value judgment
by assigning two separate definitions for risk internal and external. External risks are defined via scientific risk
analysis of system characteristics prevalent in the physical or social worlds. Internal risk, on the other hand,
defines risk based on the individual or communal perception of insecurity. In the case of internal risk, in order for
the risk to be real, it must be experienced. Of course, these two definitions are intertwined in complex ways.
Decision makers perceptions of risk are partly informed by the definitions and guidance provided by scientific
experts, and societal perceptions of risk may also influence scientific research.

The Role of Science

Ultimately, scientists cannot make expert value judgments about what climate change risks to face and
what to avoid, as that is the role of policy makers, but they can help policymakers evaluate what dangerous
climate change entails by laying out the elements of risk, which is classically defined as probability x
consequence. They should also help decision-makers by identifying thresholds and possible surprise events,
as well as estimates of how long it might take to resolve many of the remaining uncertainties that plague climate
assessments.
There is a host of information available about the possible consequences of climate change, as described in
our discussion of the SRES scenarios and of the impacts of climate change, but the SRES scenarios do not
have probabilities assigned to them, making risk management difficult. Some would argue that assigning
probabilities to scenarios based on social trends and norms should not be done (e.g., [15]), and that the use of
scenarios in and of itself derives from the fact that probabilities cant be analytically estimated. In fact, most
models dont calculate objective probabilities for future outcomes, as the future has not yet happened and
objective statistics are impossible in principle before the fact. However, modelers can assign subjective
confidence levels to their results by discussing how well established the underlying processes in a model are, or
by comparing their results to observational data for past events or elaborating on other consistency tests of their
performance (e.g., [14]). It is our belief that qualified assessment of (clearly admitted) subjective probabilities in
every aspect of projections of climatic changes and impacts would improve climate change impact
assessments, as it would complete the risk equation, thereby giving policymakers some idea of the likelihood of
threat associated with various scenarios, aiding effective decision-making in the risk-management framework.

Uncertainties
A full assessment of the range of climate change consequences and probabilities involves a cascade of
uncertainties in emissions, carbon cycle response, climate response, and impacts. We must estimate future
populations, levels of economic development, and potential technological props spurring that economic
development, all of which will influence the radiative forcing of the atmosphere via emissions of greenhouse
gases and other radiatively active constituents. At the same time, we also must deal with the uncertainties
associated with probabilities generated with carbon cycle modeling, and, equally important, confront
uncertainties surrounding climate sensitivity estimated from climate models tested on paleoclimatic situations,
as well as perform other validation exercises.
Figure 4 shows the explosion that occurs as the different elements of uncertainty are combined. This
should not be interpreted as a sign that scientists cannot assign a high degree of confidence to any of their
projected climate change impacts, but rather that the scope of possible consequences is quite wide. There are
many projected effects, on both global and regional scales, that carry high confidence estimates, but the figure
suggests that there still are many impacts to which we can only assign low confidence ratings and others that
have not yet been postulatedi.e., surprises and irreversible impacts.
One other aspect of Figure 4 needs mentioning: Current decision makers aware of potential future risks
might introduce policies to reduce the risks over timealso know as reflexive responseswhich would be
equivalent to a feedback that affects the size of the bars on Figure 4 merely because the prospects for risks
created precautionary responses. That possibility is partly responsible for the attitudes of some who are
reluctant to assign probabilitieseven subjective onesto the components of Figure 4. If no probabilities are
associated with scenarios, then the problem still remains for decision makers to weigh the importance of climate
risks against other pressing social issues competing for limited resources.

emission
scenarios

carbon cycle
response

global climate
sensitivity

regional
climate
change
scenarios

range of
possible
impacts

Source: Modified after Jones, 2000, and the "cascading pyramid of uncertainties" in Schneider, 1983.

Figure 4 Explosion of uncertainty.

Various classification schemes have been generated to categorize different types of uncertainties prevalent
in scientific assessment (e.g., [76], [20], [64], [39], [55], [11], [36]). In the AR4, one classification scheme for
uncertainties includes the following categories: lack of scientific knowledge, natural randomness, social choice,
and value diversity [23].
The plethora of uncertainties inherent in climate change projections clearly makes risk assessment difficult.
In this connection, some fear that actions to control potential risks could produce unnecessary loss of
development progress, especially if impacts turned out to be on the benign side of the range. This can be
restated in terms of Type I and Type II errors. If governments were to apply the precautionary principle and act
now to mitigate risks of climate change, theyd be said to be committing a Type I error if their worries about
climate change proved unfounded and anthropogenic greenhouse gas emissions didnt greatly modify the
climate and lead to dangerous change. A Type II error would occur if serious climate change did occur, yet
insufficient hedging actions had been taken as a precaution because uncertainty surrounding the climate
change projections was used as a reason to delay policy until the science was more certain. Researchers,
understandably, often are wary of Type I errors, as they are the ones making the projections and dont like to be
responsible for actions that turn out to be unnecessary. Decision makers, and arguably most individuals, on the
other hand, might be more worried that dangerous outcomes could be initiated on their watch (Type II error),
and thus may prefer some hedging strategies.

Vulnerability measurements
The climate science community has been asked to provide decision makers with information that may help
them avoid Type 2 errors (e.g., avoid DAI). One way is through studies providing quantitative measures of key
vulnerabilities. In contemplating quantitative values for human vulnerabilities, studies have addressed monetary
loss [42, 43, 16, 30] and a wide range of population-related metrics, including loss of life [74], risk of hunger as
measured by the number of people who earn enough to buy sufficient cereal [50], risk of water shortage as
measured by annual per capita water availability [3], mean number of people vulnerable to coastal flooding [40],
number of people prone to malaria infection or death [67, 69], and number of people forced to migrate as a
result of climate change [9].
Non-human quantitative analyses have also been performed. They have calculated potential numbers of
species lost [66], numbers of species shifting their ranges [48, 54], and absolute or relative change in range of
species or habitat type. Leemans and Eickhout (2004) [32] note that after 1 to 2 C of warming, most species,
ecosystems, and landscapes have limited capacity to adapt. Rates of climate change also influence adaptive
capacity of social and (especially) natural systems.
Another quantitative measure of vulnerability is the five numeraires, discussed above, as it encompasses
both human and non-human metrics of impacts. Each numeraire may be reported separately, or they can be
aggregated. Any aggregation should be accompanied by a traceable account of how it was obtained [38].

Thresholds
Another important step toward achieving the goal of informing decision makers is identifying climate
thresholds or limits. One classification scheme lists three categories of thresholds relevant in the context of
Article 2 of the UNFCCC: systemic (natural) thresholds, normative (social) impact thresholds, and legal limits. A
systemic threshold is a point at which the relationship between one or more forcing variables and a valued
system property becomes highly negative or non-linear [23]. Normative thresholds have been divided into two
categories by Patwardhan et al. (2003) [51]. Type I normative thresholds are target values of linear or other
smooth changes that after some point would lead to damages that might be considered unacceptable by
particular policy-makers [51]. Type II normative thresholds are linked directly to the key intrinsic processes of
the climate system itself (often non-linear) and might be related to maintaining stability of those processes or
some of the elements of the climate system [51]. Examples are presented in Table 1, below. Legal limits are
manmade constraints like environmental standards placed upon certain factors that are thought to play a part in
unfavorable outcomes. They can be influenced by normative thresholds, as well as cost and other factors.
Vulnerability

Global Mean Limit

References

3 oC in 100 yr

ONeill and Oppenheimer

(2002) [44]

700ppm CO2

Keller et al. (2004) [29]

2 oC, 450ppm CO2

ONeill and Oppenheimer

(2002) [44]

2-4 oC,
<550ppm CO2

Oppenheimer and Alley (2004,

2005) [46, 47]

1 oC

Hansen (2004) [17]

Shutdown of thermohaline
circulation

Disintegration of West
Antarctic ice sheet (WAIS)

Disintegration of Greenland
ice sheet

Smith et al. (2001) [65]

Widespread bleaching of coral
reefs

>1 C

Broad ecosystem impacts with

limited adaptive capacity
(many examples)

1-2oC

Leemans and Eickhout

(2004) [32], Hare (2003) [18],
Smith et al. (2001) [65]

450-650ppm

Parry et al. (2001) [49]

Large increase of persons-atrisk of water shortage in

vulnerable regions
Increasingly adverse impacts,
most economic sectors

>3-4oC

ONeill and Oppenheimer

(2002) [44]

Hitz and Smith (2004) [22]

Source: IPCC Fourth Assessment Report, Chapter 19, Zero Order Draft.

Table 1 Proposed numerical values of Dangerous Anthropogenic Interference.

Extensive literature relating to Type II thresholds, also referred to as Geophysical and Biological
Thresholds, has arisen in recent years. The literature has attempted to incorporate Type II thresholds into
integrated assessment and decision-making, both on global scales (e.g., [1], [6], [75], [60], [21], [8], [59]) and on
regional scales (e.g., [53]). The next step involves associating specific climate parameters with thresholds. For
example, ONeill and Oppenheimer (2002) [44] have given values of carbon dioxide concentration and global
temperature change that they believe may be associated with Type II thresholds corresponding to the
disintegration of the West Antarctic ice sheet (WAIS), collapse of thermohaline circulation, and widespread
decline of coral reefs. Oppenheimer and Alley (2004) [46] also proposed a range of threshold values for
disintegration of the WAIS, and Hansen (2004) [17] and Oppenheimer and Alley (2005) [47] discuss
quantification of thresholds for loss of WAIS and Greenland ice sheets. Due to large uncertainties in models
and in interpretation of paleoclimatic evidence, a critical issue in all of the above studies is whether the values
selected correspond to actual geophysical or biological thresholds or simply represent convenient and
subjective judgments about levels or risk.
Type I thresholds, perhaps more accurately called socioeconomic limits, generally dont involve the largescale discontinuities implied in the word threshold, with the exception of a possible collapse of an atoll society
to cope with climate-change-induced sea level rise [9]. Again, there is extensive literature on Type I thresholds.
Parry et al. (2001) [49], Arnell (2000) [4], Arnell et al. (2004) [5], Parry (2004) [50], and other studies view
climate change impacts in terms of changes in the size of vulnerable populations, typically as a result of climatechange-induced food shortages, water shortages, malaria infection, and coastal flooding.
We present a simple example as another approach to the problem of joint probability of temperature rise to
2100 and the possibility of crossing dangerous warming thresholds. Instead of using two probability
distributions, an analyst could pick a high, medium, and low range for each factor. For example, a glance at the
cumulative probability density function of Andronova and Schlesinger (2001) [2] included below shows that
the 10th percentile value for climate sensitivity is 1.1 oC for a doubling of CO2. 1.1 oC is, of course, below the 1.5
o
C lower limit of the IPCCs estimate of climate sensitivity and the temperature projection for 2100. But this 10th
percentile value merely means that there is a 10% chance climate sensitivity will be 1.1 oC or less that is, a
90% chance climate sensitivity will be 1.1 oC or higher. The 50th percentile result that is, the value that climate
sensitivity is as likely to be above as below is 2.0 oC. The 90th percentile value is 6.8 oC, meaning there is a
90% chance climate sensitivity is 6.8 oC or less, but there is still a very uncomfortable 10% chance it is even
higher than 6.8 oC a value well above the top figure in the IPCC range for climate sensitivity (4.5 oC).

Figure 5 Probability density function (A) and cumulative density function (C).
Using these three values (6.8 oC, 2.0 oC, and 1.1 oC) for high, medium, and low climate sensitivity can
produce three alternate projections of temperature over time (using a simple mixed-layer climate model), once
an emissions scenario is given. In the example below, these three climate sensitivities are combined with two of
the SRES storylines: the fossil fuel intensive scenario (A1FI) and the high-technology scenario (A1T), where
development and deployment of advanced lower carbon-emitting technologies dramatically reduces the longterm emissions. These make a good comparison pair since they almost bracket the high and low ends of the six
SRES representative scenarios range of cumulative emissions to 2100, and since both are for the A1 world,
the only major difference between the two is the technology componentan aspect decision makers have the

capacity to influence via policies and other measures. Therefore, asking how different the projected climate
change to 2100 is for the two different scenarios is a very instructive exercise in exploring in a partial way the
likelihood of crossing dangerous warming thresholds.
Well use a conservative estimate of 3.5 oC for this dangerous threshold since 3.5 oC was the highest
number projected for the 2100 temperature rise in the IPCCs Second Assessment Report (SAR) and because
the IPCC Working Group 2 TAR suggested that after a few degrees, many serious climate change impacts
could be anticipated. However, 3.5 oC is a very conservative number, since the IPCC noted that some unique
and valuable systems could be lost at warmings any higher than 1-1.5 oC. In essence, the threshold for what
is dangerous depends not only on the probabilities of factors like climate sensitivity and adaptive capacity, but
on value judgments as to what is acceptable given any specific level of warming or damageand who suffers
the damage or pays the adaptation costs. The figure below presents the results.
The most striking feature of both figures A and B (A is for the A1FI scenario and B the A1T), below, is the
top 90th percentile line, which rises very steeply above the other two lines below it. That is because of the
peculiar shape of the probability density function for climate sensitivity in the cumulative probability density
function figure, above it has a long tail to the right due to the possibility that aerosols have been holding back
not-yet-realized heating of the climate system.

Figure 6 Three climate sensitivities and two scenarios.

This simple pair of figures shows via a small number of curves the amount of temperature change over time
for three climate sensitivity probabilities (10th, 50th, and 90th percentile). However, it does not give probabilities
for the emissions scenarios themselves; only two are used to bracket uncertainty, and thus no joint probability
can be gleaned from this exercise. The problem with this is that the likelihood of threshold-crossing occurrences
is quite sensitive to the particular selection of scenarios and climate sensitivities used. This adds urgency to
assessing the relative likelihood of each such entry (scenario and sensitivity) so that the joint distribution has a
meaning consistent with the underlying probabilistic assessment of the components. Arbitrary selection of
scenarios or sensitivities will produce conclusions that could easily be misinterpreted by integrated assessors
and policymakers as containing expert subjective probabilistic analysis when, in fact, they do not until a
judgment is formally made about the likelihood of each storyline or sensitivity.
Such joint probability analyses are the next step. A group at MIT has already made an effort at it (see [72]),
as have Wigley (2005) [73], Rahmstorf and Zickfeld (2005) [52], and Mastrandrea and Schneider (2004) [37].

We will summarize here Mastrandrea and Schneider (2004) [37], which estimates the probability of DAI and the
influence of climate policy in reducing the probability of DAI.

Climate Science and Policy Crossroads

In defining their metric for DAI, Mastrandrea and Schneider estimate a cumulative density function (CDF)
based on the IPCCs burning embers diagram by marking each transition-to-red threshold and assuming that
the probability of dangerous change increases cumulatively at each threshold temperature by a quintile, as
shown by the thick black line in Figure 7. This can be used as a starting point for analyzing dangerous climate
change.

Source: Modified by Mastrandrea and Schneider, 2004.

Figure 7 Cumulative density function for Dangerous Anthropogenic Interference.

From Figure 7, Mastrandrea and Schneider identify 2.85C as their median threshold for dangerous
climate change, which may still be conservative. Mastrandrea and Schneider apply this median 2.85C
threshold to three key parameters climate sensitivity, climate damages, and the discount rate all of which
carry high degrees of uncertainty and are crucial factors in determining the policy implications of global climate
change. To perform these calculations, they use Nordhaus (1994b) [42] DICE model because it is well known
and is a relatively simple and transparent integrated assessment model (IAM), despite its limitations. Using an
IAM allows for exploration of the impacts of a wide range of mitigation levels on the potential for exceeding a
policy-relevant threshold such as DAI. Mastrandrea and Schneider focus on two types of model output: i) global
average surface temperature change in 2100, which is used to evaluate the potential for DAI; and ii) optimal
carbon taxes.
They begin with climate sensitivity, typically defined as the amount that global average temperature is
expected to rise for a doubling of CO2 from pre-industrial levels. The IPCC estimates that climate sensitivity
ranges between 1.5 oC and 4.5 oC, but it has not assigned subjective probabilities to the values within or outside

of this range, making risk analysis difficult. However, recent studies, many of which produce climate sensitivity
distributions wider than the IPCCs 1.5 oC to 4.5 oC range, with significant probability of climate sensitivity above
4.5 oC, are now available. Mastrandrea and Schneider use three such probability distributions: the combined
distribution from Andronova and Schlesinger (2001) [2], and the expert prior (F Exp) and uniform prior (F Uni)
distributions from Forest et al. (2001) [13]. They perform a Monte Carlo analysis sampling from each climate
sensitivity probability distribution separately, without applying any mitigation policy, so that all variation in results
will be solely from variation in climate sensitivity. The probability distributions they produce show the percentage
of outcomes resulting in temperature increases above their 2.85 oC dangerous threshold (Figure 8A).
Mastrandrea and Schneiders next simulation is a joint Monte Carlo analysis looking at temperature
increase in 2100 with climate policy, varying both climate sensitivity and the climate damage function, their
second parameter (Figure 8B). For climate damages, they sample from the distributions of Roughgarden and
Schneider (1999) [56], which produce a range of climate damage functions both stronger and weaker than the

Source: Mastrandrea and Schneider, 2004.

Figure 8. Climate sensitivity-only and joint (climate sensitivity and climate damages) Monte Carlo
analyses.
Notes: Panel A displays probability distributions for each climate sensitivity distribution for the climate sensitivity-only Monte
Carlo analyses with zero damages. Panel B) displays probability distributions for the joint (climate sensitivity and climate
damage) Monte Carlo analyses. All distributions indicate a 3-bin running mean and the percentage of outcomes above the
median threshold of 2.85C for dangerous climate change (P{DAI}), and the joint distributions display carbon taxes
calculated in 2050 (T2050) by the DICE model using the median climate sensitivity from each climate sensitivity distribution
and the median climate damage function for the joint Monte Carlo cases. Comparing the joint cases with climate policy
controls, b), to the climate sensitivity-only cases with negligible climate policy controls, a), high carbon taxes reduce the
potential (significantly in two out of three cases) for DAI. (However, this case uses a PRTP of 0%, implying a discount rate
of about 1%. With a 3% PRTPa discount rate of about 6%this carbon tax is an order of magnitude less, and the

reduction in DAI is on the order of 10%. See the supplementary on-line materials of Mastrandrea and Schneider, 2004 [37]
for a full discussion.)

original DICE function. As shown, aside from the Andronova and Schlesinger climate sensitivity distribution,
which gives a lower probability of DAI under the single (climate sensitivity-only) Monte Carlo analysis, the joint
runs show lower chances of dangerous climate change as a result of the more stringent climate policy controls
generated by the model due to the inclusion of climate damages. Time-varying median carbon taxes are over
$50/Ton C by 2010, and over $100/Ton C by 2050 in each joint analysis. Low temperature increases and
reduced probability of DAI are achieved if carbon taxes are high, but because this analysis only considers one
possible threshold for DAI (the median threshold of 2.85 oC) and assumes a relatively low discount rate (about
1%), these results cannot fully describe the relationship between climate policy controls and the potential for
dangerous climate change. They are given to demonstrate a framework for probabilistic analysis, and the
highly model-dependent results are not intended to be taken literally.
Because the analysis above only considers Mastrandrea and Schneiders median threshold (DAI[50%]) of
2.85C, Mastrandrea and Schneider continue their attempt to characterize the relationship between climate
policy controls and the potential for dangerous climate change by calculating a series of single Monte Carlo
analyses varying climate sensitivity and using a range of fixed damage functions (rather than just the median
case). For each damage function, they perform a Monte Carlo analysis sampling from each of the three climate
sensitivity distributions discussed above. They then average the results for each damage function, which gives
the probability of DAI at a given 2050 carbon tax under the assumptions described above, as shown in Figure 9,
below. Each band in the figure corresponds to optimisation around a different percentile range for the
dangerous threshold CDF, with a lower percentile from the CDF representing a lower temperature threshold
for DAI. At any DAI threshold, climate policy works: higher carbon taxes lower the probability of future
temperature increase, and thus reduce the probability of DAI. For example, if climate sensitivity turns out to be
on the high end and DAI occurs at a relatively low temperature like 1.476C (DAI[10%]), then there is nearly a
100% chance that DAI will occur in the absence of carbon taxes and about an 80% chance it will occur even if
carbon taxes were $400/ton, the top end of Mastrandrea and Schneiders range. If we inspect the median
(DAI[50%]) threshold for DAI (the thicker black line in Figure 9), we see that a carbon tax by 2050 of $150$200/Ton C will reduce the probability of DAI to nearly zero, from 45% without climate policy controls (for a 0%
PRTP, equivalent to a discount rate of about 1%).

Source: Mastrandrea and

Schneider, 2004.

Figure 9 Carbon taxes in 2050 and the probability of DAI.

Notes: Each band represents a different percentile range for the DAI threshold CDFa lower percentile from the CDF
representing a lower temperature threshold for DAI. At any threshold, climate policy controls significantly reduce the
probability of DAI. At the median DAI threshold of 2.85C (the thicker black line above), a 2050 carbon tax of >$150/Ton C is
necessary to virtually eliminate the probability of DAI.

Lastly, Mastrandrea and Schneider run Monte Carlo analyses varying climate sensitivity at different values
for the PRTP, which illustrates the relationship between the discount rate and the probability of DAI at different
temperature threshold values, as shown in Figure 10, below. As expected, increasing the discount rate shifts
the probability distribution of future temperature increase upwards; a lower level of climate policy controls
becomes optimal and thus increases the probability of DAI. At the median threshold of 2.85C for DAI (the
thicker black line in Figure 10), the probability of DAI rises from near zero with a 0% PRTP to 30% with a 3%
PRTP. A PRTP of 3% is the value originally specified in Nordhaus DICE model. At PRTP values greater than
1%, the optimal outcome becomes increasingly insensitive to variation in future climate damages driven by
variation in climate sensitivity.

Source: Mastrandrea
and Schneider, 2004.

Figure 10 PRTP and the probability of DAI.

Notes: Increasing the PRTP (and hence the discount rate) reduces the present value of future climate damages and
increases the probability of DAI. At the median threshold of 2.85C for DAI (thicker black line above), the probability of
DAI rises from near zero with a 0% PRTP to 30% with a 3% PRTP, as originally specified in the DICE model.

While Mastrandrea and Schneiders results using the DICE model do not provide us with confident
quantitative answers, they still demonstrate three very important issues: (1) that DAI can vary significantly,
depending on its definition; (2) that parameter uncertainty will be critical for all future climate projections, and
most importantly for this volume on the benefits of climate stabilization policies; (3) that climate policy controls
(i.e., optimal carbon taxes) can significantly reduce the probability of dangerous anthropogenic interference.
This last finding has considerable implications for introducing climate information to policymakers. We agree
with Mastrandrea and Schneider that presenting climate modeling results and arguing for the benefits of climate
policy should be framed for decision makers in terms of the potential for climate policy to reduce the likelihood
of exceeding a DAI threshold. While Mastrandrea and Schneiders quantitative results should not be taken
literally, the framework and methods for assessing DAI that they use should be taken seriously, as it is an
effective method for conceptualizing climate change policy decisions.

The Value Judgments

Despite the uncertainties surrounding climate change probabilities and consequences, policymakers must
still produce value judgments about what climate change risks to face and what to avoid. They must use all
expert information available, like that above, to decide the relative allocation of limited resources on avoiding

potential DAI versus healthcare or education or a host of other worthy causes. It is our personal value judgment
that hedging against first decimal place odds of DAI is prudent, and we hope that as climate science progresses
and more information is available to policy makers, they will be more willing to risk Type I errors in the climate
change arena and will enact effective abatement and adaptation measures. This view is partly motivated by
Figure 11. Human actions over the next few generations can precondition climatic changes and impacts over
the next millennium.
The figure shows a cartoon of effects that can play themselves out over a millennium, even for decisions
taken within the next century. Such very long-term potential irreversibilities (significant increases in global
annual average surface temperature, sea level rise from thermal expansion and melting glaciers, etc.) that the
Figure depicts are the kinds of nonlinear events (exceeding Type 2 thresholds) that would likely qualify as
dangerous anthropogenic interference with the climate system [37, 44, 7]. Whether a few dominant countries
and/or a few generations of people demanding higher material standards of living and consequently using the
atmosphere as an unpriced waste dump to more rapidly achieve such growthoriented goals is ethical is a
value-laden debate that will no doubt heat up as greenhouse gas buildups continue.

Source: IPCC, 2001d.

Figure 11 Carbon dioxide concentration, temperature, and sea level rise.

References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.

13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.

Alley, R.B., J. Marotzke, W. D. Nordhaus, J. T. Overpeck, D. M. Peteet, R. A. Pielke Jr., R. T. Pierrehumbert, P. B.

Rhines, T. F. Stocker, L. D. Talley, J. M. Wallace, 2003: Abrupt Climate Change, Science 299(5615): 2005-2010.
Andronova, N.G. and M.E. Schlesinger, 2001: Objective Estimation of the Probability Density Function for Climate
Sensitivity, Journal of Geophysical Research 106: 22,60522,612.
Arnell, N. W., 2005: Climate change impacts on river flows in Britain: the UKCIP02 scenarios, Journal of the Chartered
Institution of Water and Environmental Management. In press.
Arnell, N.W., 2000: Thresholds and response to climate change forcing: the water sector, Climatic Change 46: 305316.
Arnell, N.W., R. Nicholls, M.J.L. Livermore, S.R. Kovats, P. Levy, M.L. Parry, and S. Gaffin, 2004: Climate and socioeconomic scenarios for climate change impacts assessments: characterising the SRES storylines, Global
Environmental Change 14: 3-20.
Azar, C. and K. Lindgren, 2003: Catastrophic events and stochastic cost-benefit analysis of climate change, Climatic
Change 56:245.
Azar, C. and H. Rodhe, 1997: Targets for Stabilization of Atmospheric CO2," Science 276: 1818-1819.
Baranzini, A., M. Chesney, and J. Morisset, 2003: The impact of possible climate catastrophes on global warming
policies, Energy Policy 31(8): 691-701.
Barnett, J. and W. M. Adger, 2003: Climate Dangers and Atoll Countries, Climatic Change 61: 321.
Dessai, S., W. N. Adger, M. Hulme, J. Khler, J. P. Turnpenny, and R. Warren, 2004: Defining and experiencing
dangerous climate change, Climatic Change 64(1-2): 11-25.
Dessai, S. and M. Hulme, 2003: Does climate policy need probabilities? Working Paper No. 34, Tyndall Centre for
Climate Change Research, Norwich, UK.
European Climate Forum and Potsdam Institute for Climate Impact Research, 2004: What is dangerous climate
change? Initial results of a [Oct. 27-30, 2004 Beijing] symposium on key vulnerable regions, climate change, and
Article 2 of the UNFCCC. Presented at the 10th Conference of the Parties, Buenos Aires, 14 December. Available online
at: http://www.european-climate-forum.net/pdf/ECF_beijing_results.pdf.
Forest, C.E., P. H. Stone, A. P. Sokolov, M. R. Allen, and M. D. Webster, 2001: Quantifying Uncertainties in Climate
System Properties with the Use of Recent Climate Observations, Science 295: 113-117.
Giles, J., 2002: When Doubt Is a Sure Thing, Nature 418: 476-478.
Grubler, A. and N. Nakicenovic, 2001: Identifying Dangers in an Uncertain Climate," Nature 412:15.
Hammitt, J.K. and A.I. Shlyakhter, 1999: The Expected Value of Information and the Probability of Surprise, Risk
Analysis 19(1): 135-52.
Hansen, J., 2004: Defusing the global warming time bomb, Scientific American 290(3): 68-77.
Hare, W., 2003: Assessment of knowledge on impacts of climate change-contribution to the specification of Art. 2 of
the UNFCCC, background report to the German Advisory Council on Global Change (WBGU) Special Report 94.
Hasselmann, K., 1997: On Multifingerprint Detection and Attribution of Anthropogenic Climate Change," Climate
Dynamics 13: 601-611.
Helton, J.C. and D.E. Burmaster, 1996: Treatment of aleatory and epistemic uncertainty in performance assessments
for complex systems, Reliability Engineering & System Safety 54: 91-94.
Higgins, P. A. T., M. D. Mastrandrea, and S. H. Schneider. 2002: Dynamics of Climate and Ecosystem Coupling:
Abrupt Changes and Multiple Equilibria, Philosophical Transactions of the Royal Society of London (Series B-Biological
Sciences) 357: 647-655.
Hitz, S. and J. Smith, 2004: Estimating global impacts from climate change, Global Environmental Change 14(3): 201218.
IPCC, 2004: Chapter 19 of Working Group IIs contribution to the IPCC Fourth Assessment Report. Zero order draft.
IPCC, 2001a: The Scientific Basis, Summary for Policy Makers - Contribution of Working Group I to the Third
Assessment Report of the Intergovermental Panel on Climate Change, ed. J.T. Houghton, Y. Ding, D.J. Griggs, M.
Noguer, P.J. van der Linden, X. Dai, K. Maskell, and C.A. Johnson. Cambridge: Cambridge University Press, 881 pp.
IPCC, 2001b: Impacts, Adaptation, and Vulnerability - Contribution of Working Group II to the IPCC Third Assessment
Report, ed. J.J. McCarthy, O.F. Canziani, N.A. Leary, D.J. Dokken, K.S. White. Cambridge: Cambridge University
Press, 1032 pp.
IPCC, 2001c: Mitigation - Contribution of Working Group III to the Third Assessment Report of the Intergovernmental
Panel on Climate Change, ed. B. Metz, O. Davidson, R. Swart, and J. Pan. Cambridge: Cambridge University Press,
752 pp.
IPCC, 2001d: Synthesis Report 2001- Contribution of Working Groupe I, II, and III to the Third Assessment Report of
the Intergovernmental Panel on Climate Change, ed. R.T. Watson and the Core Writing Team. Cambridge: Cambridge
University Press, 397 pp.
IPCC, 1996: IPCC Second Assessment Climate Change 1995: Summaries for Policymakers of the three Working
Group reports, ed. B. Bolin et al. Cambridge: Cambridge University Press, 73 pp.

29. Keller, K., B. M. Bolker, and D. F. Bradford, 2004: Uncertain climate thresholds and optimal economic growth, Journal
of Environmental Economics and Management 48(1): 723-741.
30. Keller, K., K. Tan, F.M.M. Morel, and D.F. Bradford, 2000: Preserving the ocean circulation: Implications for climate
policy, Climatic Change 47(1-2): 17-43.
31. Lane, J., A. Sagar, and S.H. Schneider, 2005: Equity implications of climate change impacts and policy, Tiempo.
Submitted.
32. Leemans, R. and B. Eickhout, 2004: Another reason for concern: regional and global impacts on ecosystems for
different levels of climate change, Global Environmental Change (Part A) 14: 219-228.
33. Mann, M.E., R.S. Bradley, and M.K. Hughes, 1999: Northern Hemisphere Temperatures During the Past Millennium:
Inferences, Uncertainties, and Limitations, Geophysical Research Letters 26(6): 759.
34. Mann, M.E., R.S. Bradley, and M.K. Hughes, 1998: Global-scale temperature patterns and climate forcing over the
past six centuries, Nature, 392: 779-787.
35. Mann, M.E. and P.D. Jones, 2003: Global Surface Temperatures Over the Past Two Millennia," Geophysical Research
Letters 30(15): 5-1 5-4.
36. Manning, M. and M. Petit, 2004: A Concept Paper for the AR4 Cross Cutting Theme: Uncertainties and Risk.
Suggestions for IPCC AR4.
37. Mastrandrea, M.D. and S.H. Schneider, 2004: Probabilistic Integrated Assessment of 'Dangerous' Climate Change,"
Science 304: 571-575.
38. Moss, R.H. and S.H. Schneider, 2000: Uncertainties in the IPCC TAR: Recommendations to Lead Authors for More
Consistent Assessment and Reporting", Guidance Papers on the Cross Cutting Issues of the Third Assessment Report
of the IPCC, ed. R. Pachauri, T. Taniguchi, and K. Tanaka. Geneva, Switzerland: World Meteorological Organization,
33-51.
39. New, M. and M. Hulme, 2000: Representing uncertainty in climate change scenarios: a Monte-Carlo approach,
Integrated Assessment 1:203-213.
st
40. Nicholls, R.J., 2004: Coastal flooding and wetland loss in the 21 century: changes under the SRES climate and socioeconomic scenarios, Global Environmental Change 14(1): 69-86.
41. Nordhaus, W.D., 1994a: Expert Opinion on Climatic Change," American Scientist 82:45-51
42. Nordhaus, W.D., 1994b: Managing the Global Commons: The Economics of Climate Change. Cambridge, MA: MIT
Press.
43. Nordhaus, W.D. and J. Boyer, 2000: Warming the world: Economic models of global warming. Cambridge, MA: MIT
Press.
44. O'Neill, B. C. and M. Oppenheimer, 2002 Climate change - dangerous climate impacts and the Kyoto protocol,
Science 296 (5575): 1971-1972.
45. ONeill, B.C. and M. Oppenheimer, 2004: Climate Change Impacts Sensitive to Path to Stabilization, Proceedings of
the National Academy of Sciences 101: 16,411-16,416.
46. Oppenheimer, M. and R.B. Alley, 2005: Ice Sheets, Global Warming, and Article 2 of the UNFCCC, Climatic Change,
in press.
47. Oppenheimer, M. and R.B. Alley, 2004: The West Antarctic Ice Sheet and Long Term Climate Policy, Climatic Change
64: 1-10.
48. Parmesan, C. and G. Yohe, 2003: A Globally Coherent Fingerprint of Climate Change Impacts Across Natural
Systems," Nature 421: 37-42.
49. Parry, M., N. Arnell, T. McMichael, R. Nicholls, P. Martens, S. Kovats, M. Livermore, C. Rosenzweig, A. Iglesias, and G.
Fischer, 2001: Millions at risk: defining critical climate change threats and targets, Global Environmental Change (Part
A) 11(3): 181-183.
50. Parry, M., 2004: Global impacts of climate change under the SRES scenarios, Global Environmental Change 14(1): 1.
51. Patwardhan, A., S.H. Schneider, and S.M. Semenov, 2003: Assessing the Science to Address UNFCCC Article 2: A
concept paper relating to cross cutting theme number four. Suggestions for IPCC AR4.
52. Rahmstorf, S., and K. Zickfeld, 2005: Thermohaline Circulation Changes: A Question of Risk Assessment, Climatic
Change, in press.
53. Rial, J. A., R.A. Pielke Sr., M. Beniston, M. Claussen, J. Canadell, P. Cox, H. Held, N. De Noblet-Ducoudr, R. Prinn, J.
F. Reynolds, J. D. Salas, 2004: Nonlinearities, feedbacks and critical thresholds within the Earths climate system,
Climatic Change 65:11.
54. Root, T.L., J.T. Price, K.R. Hall, S.H. Schneider, C. Rosenzweig, and J. A. Pounds, 2003: 'Fingerprints' of Global
Warming on Animals and Plants," Nature 421: 57-60.
55. Rotmans, J. and M.B.A. van Asselt, 2001: Uncertainty in integrated assessmemt modelling: A labyrinthic path,
Integrated Assessment 2: 43-55.
56. Roughgarden, T. and S.H. Schneider, 1999: Climate Change Policy: Quantifying Uncertainties for Damages and
Optimal Carbon Taxes," Energy Policy 27: 415-429.
57. Santer, B.D., R. Sausen, T. M. L. Wigley, J. S. Boyle, K. AchutaRao, C. Doutriaux, J. E. Hansen, G. A. Meehl, E.
Roeckner, R. Ruedy, G. Schmidt, and K. E. Taylor, 2003: Behavior of Tropopause Height and Atmospheric

58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.

Temperature in Models, Reanalyses, and Observations: Decadal Changes, Journal of Geophysical Research 108 (D1):
1-1 to 1-22.
Santer, B.D., T.M.L. Wigley, and P.D. Jones, 1993: "Correlation Methods in Fingerprint Detection Studies," Climate
Dynamics 8: 265-276.
Schneider, S.H., 2004: Abrupt non-linear climate change, irreversibility and surprise, Global Environmental Change
14: 245-258.
Schneider, S.H. and C. Azar, 2001: Are Uncertainties in Climate and Energy Systems a Justification for Stronger
Near-term Mitigation Policies? Proceedings of the Pew Center Workshop on The Timing of Climate Change Policies,
ed. E. Erlich, 85-136. Washington D.C., 11-12 October 2001.
Schneider, S.H. and J. Lane, 2005: Dangers and thresholds in climate change and the implications for justices, in
Justice in Adaptation to Climate Change, ed. W.N. Adger. In press.
Schneider, S.H., K. Kuntz-Duriseti, and C. Azar, 2000: Costing Non-linearities, Surprises and Irreversible Events,"
Pacific and Asian Journal of Energy 10(1): 81-106.
Schneider, S.H. and T.L. Root, 2001: Climate Change: Overview and Implications for Wildlife", in Schneider, S.H. and
T.L. Root (eds.) Wildlife Responses to Climate Change: North American Case Studies, Washington D.C.: Island Press,
1-56.
Schneider, S.H., B.L. Turner, and H. Morehouse Garriga, 1998: Imaginable Surprise in Global Change Science,"
Journal of Risk Research 1(2): 165-185.
Smith, J. B., H.-J. Schellnhuber, and M.M.Q. Mirza, 2001: Vulnerability to Climate Change and Reasons for Concern: A
Synthesis, in Climate Change 2001: Impacts, Adaptation, and Vulnerability, ed. J. J. McCarthy et al. Cambridge:
Cambridge University Press, 913967.
Thomas, C. D. et al., 2004: Extinction risk from climate change, Nature 427: 145.
Tol, R.S.J. and H. Dowlatabadi, 2002: Vector-borne diseases, development, and climate change, Integrated
Environmental Assessment 2:173-181.
United Nations, 1992: United Nations Framework Convention on Climate Change. Bonn, Germany: United Nations, 33
pp.
van Lieshout, M., R.S. Koavates, M.T. J. Livermore, and P. Martens, 2004: Climate change and malaria: analysis of the
SRES climate and socio-economic scenarios, Global Environmental Change 14: 87-99.
von Storch, H., E. Zorita, J. Jones, Y. Dimitriev, F. Gonzlez-Rouco, S. Tett, 2004: Reconstructing Past Climate from
Noisy Data, Sciencexpress 30 September.
Wahl, E.R. and C. Ammann, 2005: "Stationarity and Fidelity of Simulated El Nio-Southern Oscillation Climate Proxies
over the Last Millenium in Forced Transient AOGCM Output," Climate Dynamics, in preparation.
Webster, M., C. Forest, J. Reilly, M. Babiker, D. Kicklighter, M. Mayer, R. Prinn, M. Sarofim, A. Sokolov, P. Stone, and
C. Wang, 2003: "Uncertainty Analysis of Climate Change and Policy Response," Climatic Change 61: 295-320.
Wigley, T. M. L., 2005: Choosing a stabilization target for CO2, Climatic Change, in press.
World Health Organization, 2003: Climate Change and Human Health Risks and Responses, ed. A.J. McMichael,
D.H. Campbell-Lendrum, C.F. Corvaln, K.L. Ebi, A. Githeko, J.D. Scheraga and A. Woodward. France, 250pp.
Wright, E. L. and J. D. Erickson, 2003: Incorporating catastrophes into integrated assessment: science, impacts, and
adaptation, Climatic Change 57:265.
Wynne, B., 1992: Uncertainty and environmental learning: Reconceiving science and policy in the preventive
paradigm, Global Environmental Change 2: 111-127.