An Environmental Risk Assessment/Management Framework for Climate Change Impact Assessments

Abstract

This paper presents an environmental risk assessment/risk management framework to assess the impacts of climate change on individual exposure units identified as potentially vulnerable to climate change. This framework is designed specifically to manage the systematic uncertainties that accompany the propagation of climate change scenarios through a sequence of biophysical and socio-economic climate impacts. Risk analysis methods consistent with the IPCC Technical Guidelines for Assessing Climate Change Impacts and Adaptations are set within a larger framework that involves stakeholders in the identification, assessment and implementation of adaptation measures. Extensive consultation between parties occurs in a flexible structure that embeds scientific methods of risk analysis within a broad setting of social decision-making. This format is consistent with recent forms of environmental risk assessment/management frameworks. The risk analysis links key climatic variables expressed as projected ranges of climate change with an upper and lower limit, with impact thresholds identified collaboratively by researchers and stakeholders. The conditional probabilities of exceeding these thresholds are then assessed (probabilities using this method are conditional as the full range of uncertainty for the various drivers of climate change, and their probability distributions, remains unknown). An example based on exceeding irrigation demand limited by an annual farm cap is used to show how conditional probabilities for the exceedance of a critical threshold can be used to assess the need for adaptation. The time between the identification of an acceptable level of risk and its exceedance is identified as a window of adaptation.The treatment of risk consists of two complementary actions, adaptation to anticipated changes in climate and the mitigation of climate change through reductions in greenhouse gas emissions. Both of these actions will reduce the risk of critical thresholds being exceeded. The potential of this framework for addressing specific requirements of the United Nations Framework Convention for Climate Change is discussed.

Full text

Natural Hazards 23: 197–230, 2001.
© 2001 Kluwer Academic Publishers. Printed in the Netherlands. 197
An Environmental Risk Assessment/Management
Framework for Climate Change Impact
Assessments
ROGER N. JONES
CSIRO Atmospheric Research, PB1 Aspendale Victoria 3195, Australia
e-mail: roger.jones@dar.csiro.au
Abstract. This paper presents an environmental risk assessment/risk management framework to as-
sess the impacts of climate change on individual exposure units identified as potentially vulnerable to
climate change. This framework is designed specifically to manage the systematic uncertainties that
accompany the propagation of climate change scenarios through a sequence of biophysical and socio-
economic climate impacts. Risk analysis methods consistent with the IPCC Technical Guidelines for
Assessing Climate Change Impacts and Adaptations are set within a larger framework that involves
stakeholders in the identification, assessment and implementation of adaptation measures. Extensive
consultation between parties occurs in a flexible structure that embeds scientific methods of risk
analysis within a broad setting of social decision-making. This format is consistent with recent forms
of environmental risk assessment/management frameworks. The risk analysis links key climatic
variables expressed as projected ranges of climate change with an upper and lower limit, with impact
thresholds identified collaboratively by researchers and stakeholders. The conditional probabilities of
exceeding these thresholds are then assessed (probabilities using this method are conditional as the
full range of uncertainty for thevarious drivers of climate change, and their probability distributions,
remains unknown). An example based on exceeding irrigation demand limited by an annual farm
cap is used to show how conditional probabilities for the exceedance of a critical threshold can be
used to assess the need for adaptation. The time between the identification of an acceptable level
of risk and its exceedance is identified as a window of adaptation. The treatment of risk consists
of two complementary actions, adaptation to anticipated changes in climate and the mitigation of
climate change through reductions in greenhouse gas emissions. Both of these actions will reduce
the risk of critical thresholds being exceeded. The potential of this framework for addressing specific
requirements of the United Nations Framework Convention for Climate Change is discussed.
Key words: climate impacts, risk assessment, climate change, risk management
1. Introduction
Environmental risk management or risk assessment is the process of identifying,
evaluating, selecting, and implementing actions to reduce risk to human health
and to ecosystems (USPCC RARM, 1997). The increasing importance of envir-
onmental quality to the economy, human health and ecosystems has influenced
a number of recently formulated national environmental risk assessment/risk man-

198 ROGER N. JONES
agement frameworks, particularly as command-and-control regulation has failed to
deliver adequate outcomes (Power and McCarty, 1998). The terms risk assessment
and risk management are both used to describe the whole framework, or specific
parts of the framework, where risk assessment encompasses an analysis phase and
risk management an implementation phase (Beer and Ziolkowski, 1996). Due to
these inconsistent terminologies, the terms risk assessment and management are
considered here as synonymous. This paper follows the Australian and New Zeal-
and Risk Management Standard (Standards Australia and Standards New Zealand,
1999) in ascribing the terms risk analysis where a level of risk is assessed, and risk
treatment where a level of risk is reduced through planned intervention.
Risk itself is defined by the US Presidential/Congressional Commission on
Risk Assessment and Risk Management (USPCC RARM) as the probability that
a substance or situation will produce harm under specified conditions. Risk is a
combination of two factors:
•The probability that an adverse event will occur.
•The consequences of the adverse event (USPCC RARM, 1997).
Risk analysis is the process of assessing these two factors. Risk treatment (some-
times referred to as risk management) is applied to reduce the consequences of
adverse events identified by risk analysis. Climate change due to the enhanced
greenhouse effect can be defined as an environmental risk in two senses: (1) the
environment is directly exposed to risk from climate change, and (2) environmental
change resulting from climate change may threaten human activities.
In the international arena, climate change is addressed by the United Nations
Framework Convention for Climate Change (FCCC), at the policy level, and by
the Intergovernmental Panel on Climate Change (IPCC), who provide research
assessments to aid policymakers and the wider community in implementing the
FCCC. The ultimate objective of the FCCC is to stabilise greenhouse gases at
a level that would ultimately prevent dangerous anthropogenic interference with
the climate system. The IPCC describes its role as to assess on a comprehensive,
objective, open and transparent basis the scientific, technical, and socio-economic
information relevant to understanding the risk of human-induced climate change,
its potential impacts, and options for adaptation and mitigation (IPCC, 1993). As
part of this role, impact assessment endeavours to understand the risk of climate
change to biophysical and socio-economic systems. The assessment of adaptation
and mitigation options is one part of risk treatment, and the implementation of
adaptation and mitigation options, which is beyond the brief of the IPCC but is
required of ratified nations by Articles 3 and 4 of the FCCC, is the other part of
risk treatment.
Since the IPCC was formed in 1988, many different methods have been applied
to climate change impact assessments. Factors that affect the choice of such meth-
ods include the goal of the assessment, the exposure units or units to be studied,
the availability of data, the choice of models suitable for the projection of future
outcomes, and the time frame involved. An exposure unit is defined as the sector,

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 199
location or activity being assessed for impacts under climate change (see Carter et
al., 1994).
A major limitation of impact assessments is the use of plausible climate scen-
arios with no information as to their likelihood (Jones, 2000a). This limitation
is due to the high uncertainty associated with projections of climate change in
previous assessments of the IPCC. As a result, adaptation options typical of the
Second Assessment Report (Watson et al., 1996) and the subsequent regional im-
pacts report (Watson et al., 1998) tend to be broad and generic. The propagation
of uncertainties throughout an impact assessment is also a factor. Where a range
of uncertainty bounded by low and high scenarios of climate change is applied
to models of a system sensitive to climate, the outcomes also show low and high
extremes that are magnified by the uncertainties within the impact system. Con-
sequently, such outcomes are usually too broad to be of use for planning adaptation
options. For instance, Schreider et al. (1996) in applying “most dry” and “most
wet” scenarios to two Australian catchments based on the results from a suite of
global climate models (GCMs), produced an increase in droughts of 35% and in
floods of 50% in 2030 respectively. This result shows that streamflow is highly
sensitive and vulnerable to climate change, i.e., it establishes that a hazard exists,
but does not quantify the risk that is faced.
A growing number of studies have presented themselves as risk assessments of
climate change impacts. While it is not the purpose of this paper to assess whether
they have met certain ‘risk assessment’ criteria, many of these studies have been
limited by the lack of an established methodology. Such studies include global
sea-level rise (Titus and Narayan, 1996; Yohe and Schlesinger, 1998), forests
(Woodbury et al., 1998; Keinast et al., 1998; Keinast et al., 1999), mangroves
(Davis et al., 1994), integrated risk assessment (Shlyakhter et al., 1995; Morgan
and Dowlatabadi, 1996), hydro-electricity (Mimikou and Baltas, 1997), air qual-
ity (Bass and Brook, 1997), land-use change (Bass et al., 1997), rice pathogens
(Luo, 1995), agriculture (Kenny et al., 2000), vector-borne diseases (Martens et
al., 1995; Patz and Balbus, 1996; Patz et al., 1998) and irrigation demand (Jones,
1999, 2000b). Most of these studies, explicitly or implicitly, measure risk in terms
of thresholds (defined in the next section) and use as inputs two or more climate
scenarios or a projected range of uncertainty bound by low and high extremes.
This paper presents a risk assessment framework for assessing the ‘bottom up’
assessment of individual exposure units. The framework is simple in concept al-
though difficult to execute due to the complexity of climate change, the different
forcings that act on impacts, and the need to involve diverse stakeholder groups in
the assessment. The framework relies on the following propositions:
•Key climatic variables forcing an exposure unit at a particular location can be
identified and used as input for an impact model.
•Those variables can be expressed in terms of a projected range with high and
low extremes with a given probability distribution function (pdf).

200 ROGER N. JONES
•An impact threshold forced by those key climatic variables can be quantified
with reference to their projected ranges.
•A conditional probability of that threshold being exceeded can be calculated
on the basis of explicitly referenced assumptions linking key climatic variables
and an impact model.
•Adaptation and/or mitigation options can be assessed to reduce the exposure
of that threshold to climate change.
The risk analysis part of the framework is illustrated using a study of irrigation de-
mand in southeastern Australia undertaken by Jones (1999, 2000b) and of sea-level
rise (Jones et al., 2000). The framework is based on a development of the IPCC
Technical Guidelines for Assessing Climate Change Impacts and Adaptations
(Carter et al., 1994) while incorporating many features typical of environmental
risk assessment frameworks. Specific examples of stakeholder interactions are not
given although the framework is influenced by previous interactions of the author
with stakeholder groups.
2. Context
The IPCC investigation and reporting process produces climate projections from
climate models through a sequence of consequences beginning with scenarios of
greenhouse gas emissions, changes in atmospheric concentrations and changes in
radiative forcing. Impact assessments have focussed on projecting impacts through
an extension of that method. A succession of models isused, culminating in impact
models that simulate the biophysical, and sometime socio-economic, effects of
climate on the exposure unit being investigated (see Carter et al., 1994). This is a
linear, prescriptive framework, where a question is asked, the models are engaged
as above, and a projection through time or ‘snap-shot’ at a particular date is the
end result. This process results in an explosion of uncertainty, as incomplete know-
ledge and fundamental system uncertainty mean that singular predictions cannot
be made.
Risk assessment aims to manage such uncertainties. Rather than being the end
result, levels of impact are addressed in the initial stages of the assessment. These
levels of impact become the criteria against which risk is evaluated in the light
of system uncertainties. In the context of climate change impacts, these criteria are
referred to as thresholds – the point where a stimulus leads to a significant response
(Parry et al., 1996). If the uncertainties surrounding projected climate change and
its effect on particular impacts can be managed, then the probability of threshold
exceedance can be calculated and the consequences of that exceedance (i.e. risk)
can be assessed.

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 201
2.1. IMPACT THRESHOLDS
Swart and Vellinga (1994) suggested defining critical levels of climate change so
that a risk-based approach to impact assessment could be undertaken on a regional
basis. They proposed the integration of such regional assessments to determine
an appropriate level of stabilisation for the global climate system, thus avoiding
dangerous climate change as required by the FCCC. Carter et al. (1994) referred
to a critical threshold in terms of critical response to climate change and Parry
et al. (1996) developed this idea further. Jones and Pittock (1997) expanded the
concept of critical thresholds to cover both climate and levels of impact related
to climate, under the general term of impact thresholds. Impact thresholds are
defined as any degree of change that can link the onset of a given critical bio-
physical or socio-economic impact to a particular climate state or states (Pittock
and Jones, 2000). Kenny et al. (2000) also argue for a broader definition of critical
threshold that encompasses management, geography and risk, linking the concept
of a threshold explicitly to climate change where the concept of criticality may be
related to “dangerous interference”. All of these authors recognise that criticality
is a phenomenon limited in scale to a local or regional basis, whereas the reference
to dangerous anthropogenic interference with the climate system in the FCCC is
global. Therefore, it is not possible to directly link the concept of critical thresholds
to dangerous climate change at present, even though Swart and Vellinga (1994) and
Parry et al. (1996) suggest methods for doing so.
Impact thresholds can be grouped into two main categories: biophysical
thresholds that mark a physical discontinuity on a spatial or temporal scale, and
behavioural thresholds where reaching a particular state triggers a change in beha-
viour in the form of a social or economic outcome (R. A. Warrick, pers. comm.,
1998). Biophysical thresholds represent a distinct change in conditions, such as
the drying of a wetland, floods, breeding events. Climatic thresholds include frost,
snow and monsoon onset. Ecological thresholds include breeding events, local to
global extinction or the removal of specific conditions for survival. Behavioural
thresholds are set by benchmarking a level of performance. Exceeding a beha-
vioural threshold results in a change of legal, regulatory, economic or cultural
behaviour. Examples of agricultural thresholds include the yield per unit area of
a crop in weight, volume or gross income (Jones and Pittock, 1997).
The more recently developed risk assessment/management frameworks involve
greater social participation within an assessment, mainly through the formalised
involvement of stakeholders (Power and McCarty, 1998). This is a formal acknow-
ledgment that the functions of risk analysis and risk treatment cannot be separated,
and that those who treat risk must be involved in the formulation and analysis
stages of an assessment (and that those who analyse risk need to take account
of stakeholders’ needs). Recognising a parallel need in impact studies, Walsh et
al. (1999) and Pittock and Jones (2000) outline the importance of stakeholder or
user-defined thresholds in climate impact studies, especially where adaptation is

202 ROGER N. JONES
an outcome of the assessment. Kenny et al. (2000) refer to management thresholds
where stakeholders are aiming for a particular level of management.
User-defined thresholds are valuable for both the analysis and communication
of outcomes as they telescope the uncertainties in operational outcomes into a
single item. However, most behavioural thresholds are dynamic and may change
over time in response to socio-economic change. Stakeholders become responsible
for the management of the uncertainties associated with that threshold. The iden-
tification of impact thresholds in the early stage of an assessment will sharpen the
aims of the assessment and aid in the communication of the results.
The advantages and disadvantages of stakeholder-defined thresholds are as
follows:
Advantages
•Stakeholders have a frame of reference for the threshold.
•Stakeholders may already be adapting to current climatic, environmental and
economic conditions for the activity under assessment.
•The threshold definition process acts as a social, cultural and economic filter,
providing criteria for acceptability and non-acceptability, i.e. ownership of the
outcomes.
•Sharpens the focus of an assessment by telescoping operational uncertainties
towards a single point.
•Allows the choice of a series of thresholds of varying degrees of change, and
containing both positive and negative outcomes.
Disadvantages
•Thresholds based on outcomes at the interface of physical and socio-economic
systems can contain a great deal of complexity.
•Involving stakeholders is a time-consuming process requiring supplementary
skills beyond those required by conventional scientific research.
•The process may not achieve consensus, especially with diverse groups of
stakeholders - however, this disadvantage can be turned to advantage by testing
different assumptions if time, skill and resources permit.
•Thresholds containing value judgements, such as those affecting changes in
management and profitability, and those containing a precautionary factor
due to uncertainty will change in response to changes in social values and/or
knowledge systems.
3. The Framework
The risk assessment framework for climate impact assessment is based on the seven
steps of impact assessment of Carter et al. (1994; Figure 1). While this framework
has proved useful, it is structured mainly for projecting impacts, so is subject to

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 203
Figure 1. Seven steps of impact assessment (Carter et al., 1994).
the weaknesses discussed above. The guidelines allow for system feedbacks and
can incorporate new or updated scenarios (Carter et al., 1994) but are not spe-
cifically designed to carry out diagnostic assessments or to involve stakeholders
in the formulation of impact thresholds for risk analysis. The definition of critical
levels after Swart and Vellinga (1994) is suggested in the evaluation of adaptation
strategies (Steps 6 and 7: Carter et al., 1994) but as argued here, this task should
be undertaken early in a risk assessment.
Seven steps for undertaking a climate impact risk assessment are shown in
Figure 2 omitting the linkages between each step. These steps occur within the
context of IPCC and the FCCC processes as described earlier. The involvement of
stakeholders is central. These steps are as follows:
1. Identify the key climatic variables affecting the exposure units being assessed.
2. Create scenarios and/or projected ranges for key climatic variables.
3. Carry out a sensitivity analysis to assess the relationship between climate
change and impacts.
4. Identify the impact thresholds to be analysed for risk with stakeholders.
5. Carry out risk analysis.
6. Evaluate risk and identify feedbacks likely to result in autonomous adapta-
tions.
7. Consult with stakeholders, analyse proposed adaptations and recommend
planned adaptation options.
The order of these tasks is not prescriptive although some broad guidelines are
offered. These tasks are described in reference to studies on individual exposure
units, whether that is a sector, region or single activity.

204 ROGER N. JONES
Figure 2. Risk assessment framework for assessing climate change impacts shown without
explicit links between the steps.
3.1. IDENTIFY THE KEY CLIMATIC VARIABLES AFFECTING THE EXPOSURE
UNIT(S)BEING ASSESSED
The first step focuses on how climate affects the exposure unit being assessed.
However, before stakeholders can address the specific needs of the assessment,
they may need to have an overview of climate change as it relates to their area of
concern. This information can be sourced from publications produced under the
auspices of the IPCC. It may also be necessary to present a ‘primer’ describing
how climate research is carried out. Publications and briefings need to summarise
major uncertainties associated with climate change research and the climate change
scenarios affecting the exposure unit under assessment. Briefings will familiarise
researchers and stakeholders with each other, with each other’s knowledge bases
(including terminology and basic assumptions), and identify the largest inform-
ation gaps and needs. In both written and presented material, uncertainties and
caveats should be clearly stated.

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 205
If a region or sector is being investigated, the activities that are sensitive to cli-
mate need to be nominated, along with the key climatic or climate-related variables
that contribute to that sensitivity. Each climate variable should be identified, along
with its relevant aspects, such as possible changes in average, or through events
of a given magnitude, frequency or duration, taking account of climate variability
and extremes from daily to decadal time scales where relevant. These nominations
can be made in report form and/or in a workshop setting. Climate sensitivity can
be nominated on the basis of experience, previous research or may require new
research where climate is known to be a strong driver but the exact mechanism
remains unknown. Ideally, the key climatic variables should be explicitly dealt
within the impact model to be used in risk analysis, or be readily represented by
a dependent variable. Too little information at this stage will prohibit a numerical
risk analysis, allowing risk to be assessed only in a general manner.
3.2. CREATE SCENARIOS AND/OR PROJECTED RANGES FOR KEY CLIMATIC
VARIABLES
The aim of the risk analysis is to quantify the relationship between impact
thresholds and the uncertainty space created from the combination of key climatic
variables under climate change. To do this, scenarios or projected ranges for key
climatic variables need to be constructed. These can range from qualitative es-
timates of change in global mean to quantitative estimates of regional change for
incorporating changes to climate variability. Projected ranges of change contain an
upper and lower limit and an explicitly defined probability distribution function
(Jones, 2000a). Scenarios derived from General Circulation Model (GCM) output
can be treated as individual samples with a given probability (the default assump-
tion is of equal probability), or combined to form projected ranges as in CSIRO
(1996). The degree of quantification that is possible will influence the choice of risk
analysis method. If the aim is to analyse the probability of threshold exceedance,
then projected ranges are required for each of the key climatic variables quan-
tifying upper and lower limits, with clear statements of the assumptions used in
constructing those ranges. For further guidance on the evaluation of climate model

206 ROGER N. JONES
data and construction of climate scenarios, the reader is directed to The IPCC Data
Distribution Centre Guidelines on the use of Scenario Data for Climate Impact and
Adaptation Assessment (IPCC-TGCIA, 1999).
Ranges of global temperature and sea-level rise created from simple climate
models that use parameters derived from the behaviour of more complex models
have been published by the IPCC (1996a). Projected ranges of regional change can
be created from a series of scenarios of direct GCM output, patterns of local change
from individual GCMs scaled by global warming, expert analysis, or integrated
assessment models. Methods demonstrating such procedures are described in Titus
and Narayan (1996), Hulme and Brown (1998), Jones (1999, 2000a and b), Hulme
and Carter (1999), Visser et al. (2000) and New and Hulme (2000).
Because of uncertainty surrounding climate change scenarios, there is no estab-
lished best-practice methodology for creating ranges and probability distributions,
therefore uncertainty analysis testing these assumptions may be an end in itself. For
instance, Bayesian-type analysis can be carried out where assumptions are made
about the range and pdf, then tested in a risk analysis to determine the effect of such
assumptions on the conditional probabilities of exceedance for an impact threshold
(e.g. Jones et al., 2000; New and Hulme, 2000). This procedure can assess the
effect of different sampling strategies used in constructing the projected range,
thus determining how important a particular uncertainty may be (Jones, 2000a).
Extreme single-event probabilities are a special case. The tails of distribution
for a climate average, such as regional temperature, will be very unlikely if all
contributing uncertainties are allowed for. This would mean that most or all of the
contributing uncertainties provided a collection of ‘worst cases’, e.g. high climate
sensitivity, high regional sensitivity, high emission scenario and a high radiative
response of the earth’s climate. The issue of how such extremes should be viewed in
terms of risk is addressed in the Discussion but would become apparent in climate
observations sooner, rather than later. A specific single event, such as the collapse
of the West Antarctic Ice-Sheet, or North Atlantic Ocean deep water formation is
a special case, and is not catered for by this framework. A more likely event is a
rapid regional change in regional climate, such as those observed throughout the

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 207
Holocene, which are poorly simulated by GCMs. This possibility is common to all
impact assessments, and requires further development. As this framework is in its
formative stages, such phenomena cannot be accommodated as yet.
3.3. CARRY OUT A SENSITIVITY ANALYSIS
Two broad types of sensitivity analysis need to be carried out in a risk assessment:
model sensitivity and the sensitivity of impacts to climate. Sensitivity analysis is
carried out in the model testing stage to assess model errors, model sensitivity
to initial assumptions and the model’s role as a source of uncertainty and in its
propagation. A sensitivity analysis is also used to survey the broad changes that
an impact may be subject to under climate change. This latter analysis is comple-
mentary to the task of constructing ranges for key climatic variables and can be
used to scope the likely risks. Carter et al. (1994) describe several different types
of sensitivity analysis that are dependent on the detail of the available information
and the method of assessment being performed.
Model sensitivities and testing are not discussed here, and the reader can refer
to Morgan and Henrion (1990) and Carter et al. (1994) for detail and methods.
However, as uncertainty management is the raison d’etre of risk assessment, ex-
treme care must be exercised through out an assessment, so that uncertainties are
identified, the nature of their propagation throughout the assessment is understood
and that they are communicated as part of the results.
A guided sensitivity analysis is carried out by applying artificial scenarios of
key climatic variables to an impact model using the likely ranges of climate model
outputs as a guide (cf. IPCC-TGCIA, 1999). A numerical sensitivity analysis is
carried out by applying arbitrary changes to the key climatic variables contributing
to the impact model(s) to be used in a risk analysis. The range of arbitrary changes
applied should exceed the projected range for each key climatic variable. During
this step the results of a sensitivity analysis carried out this way can be portrayed
using a response surface, which forms an integral part of a risk response surface
(Jones, 2000b). Higher-order response surfaces may be necessary although com-
municating these in simple terms becomes an even greater task than it is for the

208 ROGER N. JONES
simple surfaces portrayed in this paper. Many impact assessments that are driven
by arbitrary, but plausible, changes to climate inputs can be considered little more
than guided sensitivity analyses (Jones, 2000a).
3.4. IDENTIFY IMPACT THRESHOLDS
As discussed earlier, the identification of impact thresholds facilitates impact as-
sessment by providing: an agreed frame of reference linking different knowledge
systems; an agreed point of reference against which to measure future uncertainty;
the ability to apply a level of complexity that matches the available knowledge
and resources; and an improved focus, by placing the greatest weight on the
information most relevant to stakeholders.
Where thresholds are associated with current management goals, they can also
contain the basis for adaptation to climate change.
TableI identifies thresholds that have either been used in climate change impact
assessments or that are currently in use and are applicable to a risk assessment
under climate change. Impact thresholds vary widely and contain diverse assump-
tions, represent many different types of output and exhibit varying degrees of
complexity. The simplest thresholds are rules of thumb, where a rough link between
climate and a particular outcome has been recognised and is used in practical
situations. Examples are threshold amounts of rainfall creating soil moisture levels
suitable for crop sowing, days of good snow cover in alpine tourist areas and so on.
These thresholds become formalised where a more objective measure is desired.
For instance, a commonly used threshold for drought is the Palmer Drought Index
that takes soil moisture into account (Palmer, 1965). This and other measures are
often used to determine officially declared droughts, which in turn allows govern-
ment assistance to be given to primary producers. In Australia, drought is now
recognised as a normal part of climate variability and assistance is given only in
exceptional circumstances, giving rise to the construction of a new, much more
complex set of thresholds (see White and Karssies, 1997).
An impact threshold may appear very simple in concept but can conceal a great
deal of complexity. For instance, many biophysical thresholds describing aspects

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 209
Table I. Examples of thresholds for a number of different sectors that have been used in climate change impact assessments or are suitable for such assessments
Sector/Activity Threshold Source
Ecology
Species or community abundance Vulnerable Endangered Species Protection Act
Endangered 1992 (Australia)
Extinct
Species distribution Climate profile shifts beyond current distribution Dexter et al. (1995), Chapman and
Quantified change in core climatic distribution Milne (1998)
Climatic thresholds affecting distribution
Ecological processes Critical levels of mean browsing intensity Keinast et al. (1999)
Flooding events affecting frequency of waterbird Johnson (1988)
breeding events
Climatic threshold between ecogeomorphic Lavee et al. (1998)
systems
Mass bleaching events on coral reefs Hoegh-Guldberg (1999)
Phenology Winter chill – e.g. frequency of occurrence below Hennessy and Clayton-Greene (1995),
daily min. temp. threshold Kenny et al. (2000)
Cumulative degree days for various biological Spano et al. (1999)
thresholds
Daylength/temperature threshold for breeding Reading (1998)
Temperature threshold for coral bleaching Huppert and Stone (1998)
Alpine Tourism Days of snow cover delineating good, moderate Gyalistras, pers. comm (1998)
and poor seasons

210 ROGER N. JONES
Table I. Continued
Sector/Activity Threshold Source
Hydrology
Waterquality Regulated water quality standards for factors Widespread and locally specific.
such as salinity, dO, nutrients, turbidity.
Water supply Regulated and/or legislated annual supply at Jones (1999, 2000b)
system, district at farm level
Stress threshold for water storages Lane et al. (1999)
Streamflow Maintenance or low flow event frequency and E.g., Australian Rainfall and Runoff
duration (Pilgrim, 1987); Panagoulia
Controlled surcharge event (control flood) and Dimou (1997)
Uncontrolled surcharge event
Catastrophic flow
Flooding Expected Monetary Value Criteria
1 in 100 year flood
Probable Maximum Flood
Maximum historical flood
Drought Palmer drought severity index Palmer (1965)
Drought Exceptional Circumstances White and Karrsies (1999)
Hydroelectric power Current mean and minimum energy supply Mimikou and Baltas (1997)
Agriculture
Animal Health Level of animal mortality (heat and cold stress)
Heat and cold stress (level of production)
Annual cost of disease prevention/production
losses ratio or cost/benefit

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 211
Table I. Continued
Sector/Activity Threshold Source
Animal production Carrying capacity (head of stock per ha) Hall et al. (1998)
“Safe” carrying capacity of rangeland
Crop production Accumulated degree days to fruit and/or harvest Kenny et al. (2000)
Economic Net/Gross income per ha/farm/region/nation
Human Health
Vector Borne Diseases Aggregate epidemic potential Patz et al. (1998)
Climatic envelope of disease vector McMichael (1996)
Critical density of vector to maintain virus Jetten and Focks (1997)
transmission
Thermal stress Heat and cold temperature levels and duration McMichael (1996)
Infrastructure Economic “write off”, e.g. replacement less costly
than repair
Infrastructure condition falling below given
standard
Land degradation Threshold for overland flow erosion Tucker and Slingerland (1997)

212 ROGER N. JONES
of plant phenology are part of everyday life (e.g. flowering, seed set) but are dif-
ficult to describe physically, as is evapotranspiration. Behavioural thresholds are
often associated with a linear process at the physical scale, where a consequent
change in behaviour introduces non-linearity. Water quality standards that demand
a management response if a certain level is crossed are an example. Threshold
complexity may require further research to integrate physical, economic and so-
cial behaviour to quantify baseline conditions before changes can be properly
investigated (Kenny et al., 2000). For example, the “safe” carrying capacity for
rangeland grazing in Queensland, Australia (Hall et al., 1998) required a great deal
of investigation involving both historical assessments and process studies prior to
investigating possible threats to grazing under climate change (G. M. McKeon,
pers. comm., 1998). Thresholds can also be dynamic, as mentioned earlier, and
may change over time. Some of these changes may be anticipated by the analysis
of autonomous adaptations (Section 3.6), especially those linked to socio-economic
change. However, the risk assessment framework allows thresholds of any type to
be used as long as they can be quantitatively linked to key climatic or climate-
related variables. A simple heuristic threshold may be sufficient, the assessment
perhaps recommending whether or not a more detailed threshold is required on the
basis of the risk analysis step.
Unless thresholds are so widely used as to be obvious to all parties involved in
a risk assessment, it is advisable to set thresholds through an organised process.
This can be done after any of steps 1 to 3 (e.g. providing general scenarios, scen-
arios of key climatic variables, or based on the results of a sensitivity assessment),
depending on how much information needs to be made available before a choice
can be made. It is not advisable to identify thresholds any later in an assessment.
Although stakeholders are encouraged to nominate thresholds they are comfortable
with, technical skills are needed to assess whether that threshold can be represen-
ted using the models and resources available. This may require some negotiation
between modellers and stakeholders to reach a suitable outcome.
3.5. CARRY OUT RISK ANALYSIS
The risk analysis step aims to calculate the risk of exceedance of given impact
thresholds. This step outlines the procedure presented by Jones (1999, 2000b) and
discussed in Pittock and Jones (2000). The analysis framework uses the three tools
developed in steps 1 to 4, namely:

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 213
1. impact threshold(s) (step 4) based on the key climatic variables identified in
step 1,
2. projected ranges of uncertainty for those key climatic variables (the scenarios
developed in step 2), and
3. a sensitivity analysis undertaken with impact models using those key climatic
variables (step 3).
The example presented by Jones (1999, 2000b) aims to illustrate how a risk ana-
lysis technique can be applied to climate impact assessment using a threshold
approach. The context of the example and a general description of the technique
are given here. The reader is directed to Jones (2000b) for a detailed explanation
of how the technique is applied.
This analysis was conducted for irrigation demand at the farm scale for a grazed
perennial pasture in northern Victoria, Australia. A simple bucket-type soil mois-
ture model was constructed from observed data and irrigator behaviour, and used
to determine the irrigation demand sufficient to meet the annual water right. This
is termed the farm cap. Where soil moisture fell below the wilting point for the
dominant pasture species, irrigation to field capacity of soil moisture was simu-
lated. The total irrigation for each season was calculated then compared with the
farm cap to determine the annual frequency of exceedance. In seasons when the
farm cap is exceeded, the irrigator has to limit water use through one or more
of several mechanisms, such as irrigating less frequently or irrigating a reduced
area. Under current climate, the annual farm cap is exceeded in about 5% of years.
The analysis aimed to show how the frequency of annual exceedance may change
under climate change, and to calculate the probability of exceedance for a critical
threshold between 2000 and 2100.
The key climatic variables identified for the model were rainfall and potential
evaporation (Step 1). However, lack of climate change scenarios for potential evap-
oration required the use of a regression relationship between current temperature
and potential evaporation to estimate changes to potential evaporation. Projected
regional ranges of change for rainfall and temperature were calculated for rainfall
and temperature from CSIRO (1996) for ten-year intervals from 1990 to 2100 (Step
2). These were constructed from projected ranges of global warming multiplied by

214 ROGER N. JONES
scaled estimates of local change in temperature and rainfall based on a sample of
five GCMs. Using a Monte Carlo sampling technique that randomly sampled the
component ranges and multiplied them, a non-uniform distribution of projected
regional temperature and rainfall was produced. An example is shown in Figure 3
as part of a risk response surface. The techniques for calculating these changes are
described in detail in Pittock and Jones (2000) and Jones (1999, 2000b). Hulme and
Carter (1999) have produced a development of that technique incorporating decadal
variability and statistical dependence between regional changes in temperature and
rainfall based on individual GCMs.
The irrigation demand model utilised 100 years of weather-generated daily
meteorological data based on six years of record from the farm under investiga-
tion. The input data were scaled by the Monte Carlo sampled climate changes to
build up a relationship between irrigation demand and conditional probabilities of
climate change. In lieu of having information about realistic rates of adaptation
by farmers to capped water rights, a proxy critical threshold of the annual farm
cap being exceeded in 50% of years was chosen to illustrate the technique. This
critical threshold is considered to be a level where on-farm adaptation would be
insufficient, causing the activity (grazing perennial pasture under irrigation) to fail.
The results of that analysis are shown in a risk response surface for 2070 (Figure
3). The risk response surface shows the results of the sensitivity analysis of the
irrigation demand model, along with a two-dimensional plot of projected regional
changes for average temperature and rainfall. This is shown as a cumulative prob-
ability plot summed from the most likely to the least likely climates. The sensitivity
plot shows the likelihood of the farm cap being exceeded on an annual basis. The
straight lines show that with increases of temperature to 6.5 ◦C and rainfall of
±30%, the farm cap may be exceeded in up to 100% of years compared to about 5%
under the current climate. The climate projections show that the range of possible
temperature change exceeds <1to3.5◦C and the range of rainfall change +12
to −15%, but that the most likely 50% of possible future climates occupy only
about 20% of this uncertainty space. The proportion of possible climate changes
exceeding the critical threshold in 2070, i.e. the climate changes falling below the
critical threshold of 50% exceedance of the farm cap, is calculated separately as
23%.
The conditional probability that this threshold will be exceeded at decadal in-
tervals between 1990 and 2100 is depicted in Figure 4. Figure 3 shows the results
of the analysis framed in climatic terms where the outcomes are portrayed in terms
of change to the key climatic variables. Figure 4 shows the results in reference
to the impact, where the probability of exceeding any given degree of impact
can be calculated. Thus, the results can be portrayed in both ‘climate-centric’ and
‘impact-centric’ terms.
A simpler example is presented in Jones et al. (2000) who estimated the prob-
ability of exceedance for two arbitrary thresholds: a 50 cm sea-level rise and a 560
ppm atmospheric concentration of CO2, for 2050, 2075 and 2100. These thresholds

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 215
Figure 3. Risk response surface incorporating cumulative probability plots for regional cli-
mate change (shaded area expressed as most likely to least likely outcomes summed to 100%
probability) with the sensitivity response of the farm cap for annual irrigation demand (lines
expressed as percentage of years the farm cap is exceeded) for 2070 in northern Victoria. A
50% rate of exceedance of the farm cap is defined as a critical threshold. This is exceeded by
23% of projected climates (the total probability of the shaded area to the upper right of the
50% threshold).
Figure 4. Probability of exceeding the critical threshold (irrigation demand exceeding the
farm cap in 50% of all years) over time.

216 ROGER N. JONES
Figure 5. Exceedance of critical thresholds for sea-level rise (50 cm) and atmospheric CO2
(560 ppm) according to the IS92a–f scenarios and projections from IPCC (1996). The solid
lines describe the projected range of sea-level rise on the left vertical axis and the dashed line
is atmospheric CO2from the right vertical axis. The area with vertical lines shows exceedance
of the critical threshold by atmospheric CO2and the hatched area shows the exceedance of
both thresholds.
are nominated as critical thresholds for a hypothetical coastline in the tropical Pa-
cific with a fringing coral reef. Critical impacts are assumed when both thresholds
are exceeded. The CO2threshold is based on recent experimental data, suggesting
that atmospheric CO2concentration may affect the ability of corals to metabolise
carbonates, limiting their growth rate (Gattuso et al., 1999). A concentration of
560 ppm, twice that of pre-industrial levels, is estimated to reduce coral growth
rates by 17–35%, posing a risk to coral reefs (Buddemeier et al., 1998). Projected
ranges of global sea-level rise and atmospheric CO2based on the IS92a–f scenarios
were taken from (IPCC, 1996) and assumed to have a uniform probability distribu-
tion. These ranges were sampled using the Monte Carlo technique for 2050, 2075
and 2100 in a manner ensuring that anomalous combinations did not occur. The
probability of exceedance for each variable and for both occurring at once was
then calculated. Both ranges and their respected thresholds are shown in Figure 5,
which shows that the critical threshold for CO2is exceeded by 2050 but that the
critical threshold for both factors is not exceeded until about 2060. The conditional
probability of exceeding both thresholds is 16% in 2075 and 44% in 2100 (Jones
et al., 2000).
Other methods of risk analysis are also suitable for this step. A high degree
of uncertainty may only allow semi-quantitative methods to be used, e.g. rank-
ing of qualitative responses to the severity of a hazard. In such a case, a risk
assessment can apply a simple climate risk (low, moderate, high) times response
(negligible, significant, high, severe) relationship to determine whether risk should
be investigated in greater detail for certain impacts.
Another method of portraying risk under climate change does not utilise
thresholds but expresses the outcomes in terms of levels of probability. This
method integrates different ranges of uncertainty in a linked series of models that

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 217
is then sampled to create a probability distribution for the target outcomes. It is
then possible to attach an outcome to a particular percentile, e.g. the 90th, 95th
or 99th percentile. This method has been applied to sea-level rise by Titus and
Narayan (1996) and economic damage due to sea-level rise (Yohe and Schle-
singer, 1998). Titus and Narayan have argued that the 99th percentile outcome
should be determined, as this is the limit commonly associated with engineering
design specifications. However, if a precautionary approach is taken to hedge the
risk of climate change by considering the upper limit of a probability distribution
(cf. Shlyakhter et al., 1995; Beer, 1997), then the extremes of component ranges
become an important factor. The upper limits will be affected by the tails of com-
ponent ranges of uncertainty, which may have been decided on a subjective basis
and could be under- or over-estimated (cf. Morgan and Henrion, 1990).
Where a threshold is assessed, the consequences of that event occurring have
already been incorporated into the analysis, e.g. the exceedance of a critical
threshold means a particular activity can no longer take place. The next step is
to evaluate the risk analysis and assess options for risk treatment in the form of
planned adaptation.
3.6. EVALUATE RISK AND IDENTIFY FEEDBACKS LIKELY TO RESULT IN
AUTONOMOUS ADAPTATIONS
Identifying the likelihood that a critical threshold, or a lesser threshold identifying
a degree of harm, may be exceeded opens a window for adaptation (Jones, 2000b),
where options for risk treatment can be investigated. The results of the risk analysis
need to be evaluated by researchers with stakeholder groups. The following factors
will influence this and the following step:
•The planning horizon for a particular activity.
•The stakeholders’ ability to adapt to current climate, especially climate
variability.
•Knowledge of autonomous adaptations.
•Stakeholder experience in long-term planning.
•Stakeholder perception of uncertainties.

218 ROGER N. JONES
In climate change terminology, two types of adaptation can be identified:
planned and autonomous. Autonomous adaptation occurs unconsciously in bio-
logical, social and economic systems where they adjust in response to a changing
climate. Planned adaptation occurs in response to a formal adaptation assessment,
e.g. those described in Carter et al. (1994) and the risk assessment framework
outlined here. Adaptations to other driving forces such as land degradation, the
globalisation of commerce and trade and social and technological change also need
to be considered.
For irrigation demand under climate change (Jones, 1999; 2000b), autonomous
adaptations due to climate change include CO2fertilisation effects increasing pas-
ture productivity and changing commodity prices due to altered patterns of global
agricultural productivity. In addition land-use and water quality may deteriorate
faster due to the impact of climate on the widespread salinity problem already
affecting the region. Adaptations to other ongoing processes include economic
reform of the water industry (where the majority of investment is shifting from the
public to the private sector), changing dynamics between input costs (rising) and
commodity prices (flat), changing technology and the increasing globalisation of
agriculture. Even if these influences cannot be quantified due to high complexity or
large uncertainties, autonomous adaptations should be identified. This will identify
gaps in existing knowledge that can be assessed qualitatively.
The imposition of a farm cap, economic reform of the water industry and the
changing relationship between input costs and commodity prices are forcing on-
going adaptations in Australia independent of climate change. The identification
of a realistic critical threshold for irrigation demand on grazed pastures that ac-
knowledge these current adaptations would require at the very least, projections
of gross margins, CO2fertilisation effects and of competing activities for irrigation
water. This supports the conclusion of Kenny et al. (2000), that the determination of
criticality in many such cases requires a fully integrated assessment. However, such
assessments are resource intensive and complex, and exploratory assessments such
as those undertaken by Jones (1999, 2000b) and Kenny et al. (2000) can convey
a great deal of information. Calculating the risk of a critical threshold where a
fully integrated assessment cannot be undertaken may indicate whether a fuller
assessment is warranted in the future. The value of assessments of ‘rules of thumb’
cannot be discounted merely because they cannot be simulated in a fully integrated
system, and in some cases, may be all that is needed.
Autonomous adaptations need to be assessed whether they increase or decrease
the risk that is faced (Pittock and Jones, 2000). This can be undertaken by altering
critical thresholds in line with a particular rate of autonomous adaptation, allowing
for changes in the definition of criticality for a particular exposure unit over time.
This would require a repetition of Step 4 with new information about autonom-
ous adaptation. For example, Hoegh-Guldberg (1999) assessed the temperature
threshold for mass coral bleaching as remaining constant over time due to a lack of
evidence about rates of autonomous adaptation to increases in sea surface temper-

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 219
atures. However, the spatial distribution of different bleaching thresholds for coral
reefs suggests that adaptations to warming may be possible. Accordingly, both a
‘best case’ with high rates of adaptation and ‘worst case’ with no autonomous
adaptation can be assessed for risk. The principle is the same for assessing planned
adaptation.
As shown from the previous examples, new models may need to be added to
account for autonomous adaptation in a quantitative way (e.g. by adding a trade
model to account for commodity prices in agriculture). However, system com-
plexity may preclude the modelling of autonomous adaptations. In fact, a number
of very important autonomous adaptations, such as changing vegetation and at-
mospheric chemistry have not yet been incorporated into GCMs. Autonomous
adaptations should also be incorporated into the first iteration of a risk analysis
where the results would otherwise be unrealistic.
The use of Bayesian methods to test prior assumptions about sources of uncer-
tainty, such as autonomous adaptation, can also be used to test those assumptions
and determine how those uncertainties may influence the probability of various
outcomes. If prior assumptions of uncertainty indicate that risk is insensitive to
those assumptions, then new information will not significantly alter the outcome.
However, if the outcome is sensitive to a prior assumption, then the source of that
uncertainty may need to be better understood.
3.7. CONSULT WITH STAKEHOLDERS TO ANALYSE AND RECOMMEND
ADAPTATION OPTIONS
Two main questions concerning the results of a risk analysis are:
1. Is the level of risk sufficient to warrant treatment?
2. What forms of planned adaptation are acceptable to stakeholders?
To address the first question, stakeholders need to be presented with the results of
the risk analysis to determine whether planned adaptation, or risk treatment, should
take place. If the first question is answered positively, then options for planned
adaptation need to be described and tested. Of the few studies that have carried
out formal risk assessments, even fewer have assessed adaptation options with

220 ROGER N. JONES
stakeholders. For instance, both Jones (1999, 2000b) and Kennyet al. (2000) apply
thresholds defined by the researchers themselves to illustrate assessment tech-
niques and while they do not engage stakeholders in those processes, acknowledge
the importance of doing so.
The results of the risk analysis for irrigation demand show that the window of
adaptation is wide open compared with the planning horizons commonly applied
to on-farm irrigation. The planning horizon is generally from about one to three
decades, whereas the critical threshold is not attained until 2050 but is present in
over 50% of possible future climates by 2100. This suggests that farmers could
potentially adapt over several generations if aware of the problem. The adaptations
required by 2030 are modest (Jones, 2000b) and may well be accounted for in the
autonomous and planned adaptations already taking place, such as water industry
reform and increases in on-farm efficiency due to management adaptations to cope
with salinity, waterlogging and water pricing. This study did not present the results
to stakeholders, as it was intended to illustrate risk assessment techniques and
contains too many caveats to be used in forward planning for irrigation (Jones,
2000b).
In the real world, distinctions between planned and autonomous adaptations are
rarely clear-cut. but there are a number of examples of planned adaptation where
gradual changes have been imposed through treaty or legislation to meet a target
some years into the future, and include the Kyoto Protocol, vehicle emissions in
California, and the vehicle import tariff scheme in Australia.
In a joint study of the impact of climate on regional transport infrastructure in
Queensland Australia involving stakeholders, Queensland Transport et al. (1999)
identified key climatic variables affecting transport infrastructure in Queensland,
and produced projections for those variables. A qualitative level of risk was as-
sessed, based on the upper extreme of projected ranges of key climatic variables
affecting each type of transport infrastructure. From this information levels of
low, moderate, significant and high risk were assessed for a number of different
types of transport infrastructure (road, rail, airports). Stakeholders were unable to
define impact thresholds in terms of current financial exposure to damage under the
present climate so the study recommended improving the monitoring of responses
to infrastructure damage under climate variability.
Therefore, the transport study could only address adaptation in the most gen-
eral of terms. Recommendations for adaptation took into account the windows for
adaptation based on the severity and timing of the climate-related risks compared
with the engineering design life of infrastructure (the planning horizon). The ma-
jor recommendations identified the need to carry out more detailed assessments
for vulnerable types of transport infrastructure in specifically identified locations
(e.g. road and rail in flood prone areas, bridges and coastal airports) so that locally
defined thresholds directly related to damage could be identified. This information
is needed to assess the relative costs and benefits of different levels of adaptation.

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 221
The risk analyses presented in Section 3.5 describe methods for calculating con-
ditional probabilities of threshold exceedance opening windows of adaptation but,
to the author’s knowledge, these methods have not yet been applied in partnership
with stakeholders. This applies two steps, the first where stakeholders involved in
the construction of thresholds, the second where results of the analysis are presen-
ted as probabilities of exceedance to stakeholders, and they are to assess those in
terms of risk. If risk is perceived as sufficiently high, then planned adaptation may
be proposed and tested. The perception of risk is very important in this process.
Stakeholders are not a homogenous group and will weigh hazards differently to
each other and to researchers. This framework incorporates stakeholder interaction
in order to encourage all parties to speak each other’s “language” with regards to
risk perception and adaptation.
The acceptability of adaptations to stakeholders largely depends on a combin-
ation of economic, regulatory and cultural factors that will be case-specific. For
example, integrated coastal zone management in the South Pacific aiming to protect
local communities from storm surge and sea-level rise needs to take traditional pat-
terns of land ownership into account (Nunn and Mimura, 1997). A current project
identifying the risk of flooding and sea-level rise in the Gold Coast City Council
in Queensland Australia is aiming to regulate long-term flood zones in order to
prevent inappropriate development that could expose the municipality to financial
and legal penalties in the future (Walsh et al., 1998).
Where stakeholder interests are represented, the testing of adaptation options
is likely to be most valuable where a degree of vertical integration is present,
moving assessments beyond biophysical outcomes towards social and economic
outcomes (Kenny et al., 2000). Stakeholders participate in the listing of possible
adaptations, but where there is high uncertainty, will be most likely to opt for those
that keep them in their own ‘comfort zone’, rejecting potential benefits in more
radical alternatives. To overcome such conservatism, formal decision analysis can
be used to compare adaptation options within the impact-modelling framework
where outcomes are measured in terms of a reduction in the probability of a critical
threshold being reached by a particular date. The aim of introducing alternatives
is not to become prescriptive, as the overall intention of an assessment is to ac-
commodate the views of stakeholders, but is rather to encourage lateral thinking to
explore alternatives.
The following two factors will affect the development of climate change
adaptation assessments:
1. Adaptation to climate change is unlikely to be implemented in isolation but
integrated with adaptation to other drivers of change, such as economic change,
greenhouse gas mitigation, sustainable use of resources, and cultural and
political change.
2. There is a broad convergence of techniques under the subject of environmental
risk assessment/management that forms a body of knowledge encompassing
scientific, economic and social theory that is broadly applicable for use in

222 ROGER N. JONES
Figure 6. Risk assessment framework for assessing climate change impacts.
climate change impact assessments. Although some of the terminology is
different many of the principles are the same.
4. Discussion
This paper presents a framework for the risk assessment of climate change impacts
(Figure 6) that is compatible with existing environmental risk assessment method-
ologies. Risk assessment is a more appropriate methodology for impact assessment
than the ‘prediction’ of unique outcomes based on projecting a sequence of events
beginning with greenhouse gas emissions, the projection of climate change using
climate models, and the projection of impacts using impact models. While such
methods were the mainstay of assessments in the IPCC Working Group II Second
Assessment Report (Watson et al., 1996) and have proved very useful, new devel-
opments in the understanding of climate change and its attendant uncertainty mean
that significant improvements can now be made.
These developments are as follows:
•The improved capacity of coupled global climate models to simulate most
large-scale interactions within the climate system, producing more real-
istic representations of current climate and broadly consistent projections of
temperature change at the regional scale and rainfall change at the contin-
ental/oceanic scale.

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 223
•New IPCC sanctioned emissions scenarios (the SRES scenarios) encompassing
a comprehensive range of possible futures (Nakicenovic and Swart, 2000).
•Tools for managing and analysing major greenhouse-related climate uncer-
tainties, such as the DIALOGUE model of Visser et al. (2000), the ICAM
model (Morgan and Dowlatabadi, 1996; Dowlatabadi and Morgan, 1998) and
for sea-level rise (Titus and Narayan, 1996).
•Techniques allowing regional projections to be constructed using approaches
suggested by Wigley and Santer (1990) and Pittock (1993) and illustrated by
CSIRO (1996), allow the upper and lower limits of quantifiable regional cli-
mate change to be estimated. Jones (1999, 2000a and b) and Hulme and Carter
(1999) show how probability distribution functions for these projections can
be calculated using simple statistical techniques and assumptions.
•The development of impact thresholds as a concept against which climate un-
certainty can be measured by Swart and Vellinga (1994), Parry et al. (1996),
Jones and Pittock (1997), Pittock and Jones (2000) and Kenny et al. (2000), as
an alternative to projecting outcomes using “prescriptive” methodologies.
This paper describes a methodology for risk assessment that utilises the above
developments. Although interactions between stakeholders and researchers are an
integral part of the framework, they are less well developed than the risk analysis
methods. The communication of scenarios, key climatic variables and user-defined
thresholds is described in some detail but the communication of risk, and the prior-
itisation and analysis of adaptation options are less clear. These tasks will become
better understood by undertaking projects that incorporate scientific assessment
with stakeholder interaction, to learn by experience and to communicate that ex-
perience. Experience from other areas of environmental risk assessment where
stakeholder participation and risk communication are better developed can also
be utilised in climate change risk assessments. Examples from water resources,
the sector dealt with in this article, are Russell (1993) and Syme and Sadler
(1994). Hennessy and Jones (1999) also report on a stakeholder workshop explor-
ing thresholds for a scoping study for a regionally integrated assessment, that led
to the development of a risk assessment on heat stress in dairy cattle (Jones and
Hennessy, 2000).
Two inter-related difficulties facing the communication of risk under climate
change are:
1. The large uncertainties that accompany climate change projections and the
other major drivers of change that will manifest over long time scales.
2. Risks that unfold over long time scales are often discounted in favour of
commercial, political and event-based risks that may manifest over the short
term.
Therefore, impacts with long-term planning horizons are likely to be more
attractive for risk assessment.
Risks associated with climate change impacts can also be incorporated into
other forms of environmental risk that manifest over similar time scales, such as

224 ROGER N. JONES
those associated with desertification, land-use change, biodiversity and salinity. For
instance, projections of salt loads in the Murray-Darling Basin, Australia’s most
important river system over the next century (MDBMC, 1999) have not taken cli-
mate change into account, even though climate and salinity are inextricably linked.
The exploration of critical system thresholds using a risk assessment framework
offers the opportunity to explore the combined affects of climate change and salin-
ity and to treat the risk jointly. This framework also has the potential to be applied
to any system where numeric analysis in the form of modelling and thresholds
are utilised. Where they are not, more generic systems, such as the Australian and
New Zealand Risk Management Standard (Standards Australia and Standards New
Zealand, 1999) will be appropriate.
The integration of adaptation measures with short-term benefits is the so-called
‘win-win’ scenario that is the target of a great deal of climate change policy but
which requires more research to test existing adaptations under climate change.
Adaptations that reduce the risks associated with a wide variety of hazards, both
short- and long-term, are more preferable than adaptations treating a single source
of risk.
The risk assessment framework presented here is designed for assessing the
risk to individual exposure units or a set of closely related activities on a sectoral
or regional basis rather than defining or assessing dangerous climate change. Both
Swart and Vellinga (1994) and Parry et al. (1996) recognise that thresholds are
location-, activity- and value-dependent, and that a synthesis of multiple assess-
ments is needed to diagnose levels of dangerous climate change. This type of
approach is a ‘bottom up’ methodology where the risks to individual exposure
units are integrated to obtain a global outcome. Top down assessments are those
that set global thresholds or rates of change, as in the “safe corridor” type analysis
(Petschel-Held and Schellnhuber, 1997). To contribute to the assessment of danger-
ous climate change as in the FCCC, bottom up methods such as that proposed here
need to be further tested and refined before the results can be integrated. Only then
can they contribute an assessment of what degree of climate change constitutes
‘dangerous anthropogenic influence’ within the terms of the FCCC and contribute
to risk treatment in the form of mitigation.
The monitoring and assessment of risk treatment is an important part of en-
vironmental risk assessment methods, where treatment options are monitored over
time, assessed and revised if necessary. These tasks cannot be easily represented
under climate change because the time lag between greenhouse gas emissions and
the equivalent climate response is measured in decades to centuries, subjecting ad-
aptation responses to similar delays. A monitoring program that cannot take effect
for decades is clearly impractical. However, both the risk itself, and risk treatment
in the form of adaptation can be monitored and updated. The risk framework itself
is iterative and can be re-applied in response to new information. Given that the be-
nefits of adaptation to climate change may not be apparent for some time, there is a
need to ensure that other short-term benefits are available e.g. adaptation to climate

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 225
variability, increased production, amelioration of land degradation, institutional
capacity to deal with unexpected and extreme events.
Pittock and Jones (2000) recommend that an entire risk assessment framework
should be assembled as soon as practicable rather than to build it sequentially from
the ground up, getting each component ‘right’ before moving on, as scientists are
so tempted to do. Instead, creating the communication links between the various
stakeholders and researchers to share language and concepts and to set thresholds
should take precedence over building new and better mathematical models. Build-
ing the conceptual framework with the cooperation of stakeholders may demand
new models, or modifications to models that focus on the desired outcome, rather
than building a model of every interaction that is perceived by scientists to have
some significance.
5. Conclusion
The risk assessment framework outlined in this paper offers a formal methodology
for assessing the risk of climate change to individual exposure units identified as
potentially vulnerable to climate change. This framework is consistent with other
forms of environmental risk assessment/management through its stakeholder focus,
strong communication links between parties throughout the process and a flexible
framework for analysis that embeds scientific methods within abroad framework of
social decision-making. The analysis phase of the framework involves the linking
of key climatic variables with impact thresholds. The conditional probabilities of
exceeding those thresholds are then assessed within the context of projected ranges
for key climatic variables under climate change.
An example based on an irrigation demand model is used to show how condi-
tional probabilities for the exceedance of a critical threshold can be used to assess
the need for adaptation. The time between the identification of an acceptable level
of risk and its exceedance is identified as a window of adaptation. The risk treat-
ment phase can consist of two complementary actions, adaptation to anticipated
changes in climate change or the mitigation of climate change through reductions
in greenhouse gas emissions. If properly assessed and implemented, both of these
actions will reduce the risk of critical thresholds being exceeded.
Many decision-making uncertainties still need to be addressed, particularly
those that affect the communication and perception of risk. In an environment
dominated by short-term planning, the long time lags between cause (emissions)
and effect (climate change) will influence risk perception in a different way to
problems with short lead times. It is also difficult to define acceptable levels of
risk where the uncertainty is high. Will stakeholder groups be motivated to adapt
to climate change when conditional probabilities for the exceedance of critical
thresholds that they have helped to define exceed a certain level? And what are
those levels? These questions need to be investigated through experience, where the
framework is applied and adjusted to incorporate new knowledge, as the science of

226 ROGER N. JONES
climate change improves and risk is applied more widely to determine ecologically,
socially and economically based risks. This experience means contributing to, and
taking guidance from, a broader public process.
Climate change is only one of the drivers grouped under the banner of global
change so, in many situations, adaptation to climate change will have to satisfy
multiple drivers of change, and adaptations to other drivers will have to incor-
porate climate change. An assessment framework that has the capacity to include
multiple drivers, or that is compatible with similar frameworks, is more likely to
gain acceptance that one that is developed for a single purpose. A further goal
of climate change risk assessment is to identify levels of stabilisation for green-
house gases to avoid dangerous anthropogenic climate change. The risk assessment
framework presented here, moves impact assessment beyond simplesensitivity and
vulnerability assessments towards that goal.
Acknowledgments
This paper is dedicated to the memory of Gyn Jones, my father, who demonstrated
the importance of researchers interacting with stakeholders long before it became
acceptable, and who provided the data used to build the irrigation demand model
used in this paper. Professor Tom Beer and two anonymous reviewers made a
number of suggestions that improved the manuscript. A number of researchers,
especially Barrie Pittock, Peter Whetton and Kevin Hennessy, have contributed to
this framework through their collaborative efforts. Funding from the Governments
of NSW, Queensland and Victoria, Environment Australia, the South Pacific Re-
gional Environment Programme and the Rural Industries Research Development
Corporation is also acknowledged. The paper is a product of the Climate Risk and
Integrated Assessment Project of CSIRO Atmospheric Research.
References
Bass, B. and Brook, J. R.: 1997, Downscaling procedures as a tool for integration of multiple air
issues, Envir. Model. Assess. 46, 151–174.
Bass, B., Huang, G. H. and Russo, J.: 1997, Incorporating climate change into risk assessment using
grey mathematical programming, J. Envir. Man. 49, 107-123.
Beer, T. and Ziolkowski, F.: 1996, Environmental risk assessment – an Australian perspective, in
TW.Norton,T.BeerandS.R.Dovers,Risk and Uncertainty in Environmental Management,
Centre for Resource and Environmental Studies, Canberra, pp. 3–13.
Beer, T.: 1997, Strategic risk management: a case study of climate change, World. Res. Rev. 9, 113–
126.
Buddemeier, R. W., Gattuso, J.-P., and Kleypas, J. A.: 1998, Rising CO2and marine calcification,
LOICZ Newsletter, No. 8, pp. 1–3.
Carter, T. R., Parry, M. L., Harasawa, H., and Nishioka, S.: 1994, IPCC Technical Guidelines for
Assessing Climate Change Impacts and Adaptations, University College, London and Centre for
Global Environmental Research, Japan, 59 pp.

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 227
Carter, T. R., Hulme, M., and Viner, D. (eds): 1999, Representing Uncertainty in Climate Change
Scenarios and Impact Studies, ECLAT-2 Workshop Report Number 1, Climatic Research Unit,
Norwich, UK (in press).
Chapman, A. D. and Milne, D. J.: 1998, The Impact of Global Warming on the Distribution of
Selected Australian Plant and Animal Species in Relation to Soils and Vegetation, Environmental
and Resource Information Network (ERIN),Environment Australia, Canberra, 146 pp.
CSIRO: 1996, Climate Change Scenarios for the Australian Region, Climate Impact Group, CSIRO
Division of Atmospheric Research, Melbourne, 8 pp.
Davis, W. P., Thornton, K. W., and Levinson, B.: 1994, Framework for assessing effects of global
climate change on mangrove ecosystems, Bull. Mar. Sci. 54, 1045–1058.
Dexter, E. M., Chapman, A. D., and Busby, J. R.: 1995, The impact of global warming on the dis-
tribution of threatened vertebrates, Environmental and Resource Information Network (ERIN),
Environment Australia, Canberra. URL:
http://www.erin.gov.au/life/end_vuln/animals/climate/climate_change/cc_exe.html
Dowlatabadi, H. and Morgan, M.G.: 1999, The Integrated Climate Assessment Model,Centerfor
Integrated Study of the Human Dimensions of Global Change, Carnegie Mellon University,
Pittsburgh.
Gattuso, J. P., Allemand, D., and Frankignoulle, M.: 1999, Photosynthesis and calcification at cel-
lular, organismal and community levels in coral reefs: A review on interactions and control by
carbonate chemistry, Am. Zool. 39, 160–183.
Hall, W. B., McKeon, G. M., Carter, J. O., Day, K. A., Howden, S. M., Scanlan, J. C., Johnston, P.
W., and Burrows, W. H.: 1998, Climate change in Queensland’s grazing lands: II. An assessment
of the impact on animal production from native pastures, Rangel. J. 20, 177–205.
Hennessy, K. J. and Clayton-Greene, K.: 1995, Greenhouse warming and vernalisation of high-chill
fruit in southern Australia, Clim. Change 30, 327–348.
Hoegh-Guldberg, O.: 1999, Climate Change, Coral Bleaching and the Future of the World’s Coral
Reefs, Greenpeace, Sydney, 29 pp.
Hulme, M. and Brown, O.: 1998, Portraying climate scenario uncertainties in relation to tolerable
regional climate change, Clim. Res. 10, 1–14.
Hulme, M. and Carter, T. R.: 1999, Representing uncertainty in climate change scenarios and impact
studies, in T. R. Carter, M. Hulme and D. Viner (eds), Representing Uncertainty in Climate
Change Scenarios and Impact Studies, ECLAT-2 Workshop Report Number 1, Climatic Research
Unit, Norwich, UK (in press).
Huppert, A. and Stone, L.: 1998, Chaos in the Pacific’s coral reef bleaching cycle, Am. Natural. 152,
447–459.
IPCC: 1993, IPCC procedures for preparation, review, acceptance, approval, and publication of its
reports, IPCC, Geneva.
IPCC: 1996a, Technical Summary, in J. T. Houghton, L. G. Meira Filho, B. A. Callander, N. Harris,
A. Kattenberg and K. Maskell (eds), Climate Change 1995: The Science of Climate Change,
Contribution of Working Group I to the second assessment report of the Intergovernmental Panel
on Climate Change, Cambridge University Press, Cambridge, pp. 9–49.
IPCC: 1996b, Technical Summary, in R. T. Watson, M. C. Zinyowera and R. H. Moss (eds),
Climate Change 1995: Impacts, Adaptations and Mitigation of Climate Change: Scientific–
Technical Analyses, Contribution of Working Group to the second assessment report of the
Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, pp. 1–53.
IPCC-TGCIA, 1999: Guidelines on the Use of Scenario Data for Climate Impact and Adaptation
Assessment Version 1. Prepared by T. R. Carter, M. Hulme and M. Lal, Intergovernmental Panel
on Climate Change, Task Group on Scenarios for Climate Impact Assessment, 69 pp.
Jetten, T. H. and Focks, D. A.: 1997, Potential changes in the distribution of dengue transmission
under climate warming, Am. J. Trop. Med. Hyg. 57, 285–297.

228 ROGER N. JONES
Johnson, B.: 1998, Consequences for the Macquarie Marshes, in Hassall and Associates (eds), Cli-
mate Change Scenarios and Managing the Scarce Water Resources of the Macquarie River,
Australian Greenhouse Office, Canberra, pp. 61–68.
Jones R. N. and Pittock, A. B.: 1997, Assessing the impacts of climate change: the challenge for
ecology, in N. Klomp and I. Lunt (eds), Frontiers in Ecology: Building the Links, Elsevier Science
Ltd., Amsterdam, pp. 311–322.
Jones, R. N.: 1999, Climate change scenarios, impact thresholds and risk, in S. M. Howden and J.
T. Gorman (eds), Proceedings of a Workshop on the Impacts of Global Change on Australian
Temperate Forests, Resource Futures Program Working Paper Series 99/08, CSIRO Wildlife and
Ecology, Canberra, pp. 40–52.
Jones, R. N.: 2000a, Managing uncertainty in climate change projections – issues for impact
assessment, Clim. Change 45, 403–419.
Jones, R. N.: 2000b, Analysing the risk of climate change using an irrigation demand model, Clim.
Res. 14, 89–100.
Jones, R. N., Hennessy, K. J., Page, C. M., Pittock, A. B., Suppiah, R., Walsh, K. J. E., and Whetton,
P. H.: 2000, An analysis on the effects of the Kyoto Protocol on Pacific Island Countries. Part
Two: Regional Climate Change Scenarios and Risk Assessment Methods, South Pacific Regional
Environment Programme, Apia Samoa, 75 pp.
Keinast, F., Brzeziecki, B., and Wildi, O.: 1996, Long-tern adaptation potential of Central European
mountain forests to climate change: a GIS-assisted sensitivity assessment, For. Ecol. Manage.
80, 133–153.
Keinast, F., Wildi, O., and Brzeziecki, B.: 1998, Potential impacts of climate change on species
richness in mountain forests – an ecological risk assessment, Biol. Cons. 83, 291–305.
Keinast, F., Fritschi, J., Bissegger, M., and Abderhalden, W.: 1999, Modeling successional patterns
of high-elevation forests under changing herbivore pressure – responses at the landscape level,
For. Ecol. Manage. 120, 35–46.
Kenny, G. J., Warrick, R. A., Campbell, B. D., Sims, G. C., Camilleri, M., Jamieson, P. D., Mitchell,
N. D., McPherson, H. G., and Salinger, M. J.: 2000, Investigating climate change impacts and
thresholds: An application of the CLIMPACTS integrated assessment model for New Zealand
agriculture, Clim. Change 46, 91–113.
Lane, M. E., Kirshen, P. H., and Vogel, R. M.: 1999, Indicators of impacts of global climate change
on US water resources, J. Water. Resour. Plan. M.-ASCE 125, 194–204.
Lavee, H. Imeson, A. C., and Sarah, P.: 1998, The impact of climate change on geomorphology and
desertification along a Mediterranean-arid transect, Land Degrad. Devel. 9, 407–422.
Luo, Y., TeBesst, D. O., Teng, P. S., and Fabellar, N. G.: 1995, Simulations studies on risk analysis
of rice leaf blast epidemics associated with global climate change in several Asian countries, J.
Biogeog. 22, 673–678.
Martens, W. J. M, Niessen, L. W., Rotmans, J., Jetten, T. H., and McMichael, A. J.: 1995, Potential
impact of global climate change on malaria risk, Environ. Health Persp. 103, 458–464.
McMichael, A. J.: 1996, Human population health, in R. T. Watson, M. C. Zinyowera and R. H.
Moss (eds), Climate Change 1995: Impacts, Adaptations and Mitigation of Climate Change:
Scientific-Technical Analyses, Contribution of Working Group II to the second assessment report
of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, pp.
561–584.
MDBMC: 1999, The Salinity Audit of the Murray-Darling Basin. Murray-Darling Basin Commis-
sion, Canberra, 39 pp.
http://www.mdbc.gov.au/Download/SalinityAudit/Final_Salt_Audit.pdf
Mimikou, M. A. and Baltas, E. A.: 1997, Climate change impacts on the reliability of hydroelectric
energy production, Hydrolog. Sci. J. 42, 661–678.
Morgan, M. G. and Dowlatabadi, H.: 1996, Learning from integrated assessment of climate change,
Clim. Change 34, 337–368.

AN ENVIRONMENTAL RISK ASSESSMENT/MANAGEMENT FRAMEWORK 229
Morgan, M. G. and Henrion, M: 1990, Uncertainty: A Guide to Dealing with Uncertainty in
Quantitative Risk and Policy Analysis. Cambridge University Press, Cambridge, 332 pp.
Nakicenovic, N and Swart, R. (eds): 2000, Emissions Scenarios: Special Report of the Inter-
governmental Panel on Climate Change. Cambridge University Press, Cambridge UK, 570
pp.
New, M. and Hulme, M.: 2000, Representing uncertainty in climate change scenarios: A Monte Carlo
approach, Integrated Assess. 1, 203–213.
Nunn, P. D. and Mimura, N.: 1997, Coastal vulnerability and integrated coastal zone management in
the Pacific Island Region, in Leatherman, S.P. (ed.) Island States at Risk: Global Climate Change,
Development and Population. Journal of Coastal Research, Special Issue #24, 133–151.
Palmer, W.: 1965, Meteorological drought, Tech. Rep. 45, US Weather Bureau, Washington DC, 58
pp.
Panagoulia, D. and Dimou, G.: 1997, Sensitivity of flood events to global climate change, J. Hydrol.
191, 208–222.
Parry, M. L., Carter, T. R., and Hulme, M.: 1996, What is a dangerous climate change? Global
Environ. Change 6,1–6.
Patz, J. A. and Balbus, J. M.: 1996, Methods for assessing public health vulnerability to global
climate change, Clim. Res. 6, 113–125.
Patz, J. A., Martens, W. J. M., Focks, D. A., and Jetten, T. H.: 1998, Dengue fever epidemic potential
as projected by general circulation models of global climate change, Environ. Health Persp. 106,
147–153.
Patt, A.: 1999, Extreme outcomes: The strategic treatment of low probability events in scientific
assessments, Risk, Decision, and Policy 4, 1–15.
Petschel-Held, G. and Schellnhuber, H. J.: 1997: The tolerable window approach to climate control:
optimization, risks, and perspectives, in F. L. Tóth, (ed.), Cost-Benefit Analysis – the broader
perspective. Birkh"user, Basel, pp. 121–139.
Pilgrim, D. H. (ed.): 1987, Australian Rainfall and Runoff (Volume 1), The Institution of Engineers
Australia, Canberra, 374 pp.
Pittock, A. B.: 1993, Climate scenario development, in: A. J. Jakeman, M. B. Beck and M. J. McAleer
(eds), Modelling Change in Environmental Systems, John Wiley & Sons, Chichester, pp. 481–
503.
Pittock A. B. and Jones R. N.: 2000, Adaptation to what, and why? Environ. Monit. Assess. (in press).
Power, M. and McCarty, L. S.: 1998, A comparative analysis of environmental risk assessment/risk
management frameworks, Environ. Sci. Technol. 32(9), 224A–231A.
Queensland Transport, CSIRO and PPK: 1999, The Effects of Climate Change on Transport In-
frastructure in Regional Queensland: Synthesis Report, Queensland Transport, Brisbane, 18
pp.
Reading, C. J.: 1998, The effect on winter temperatures on the timing of breeding activity in the
common toad Bufo bufo, Oecologia 117, 469–475.
Russell, C.: 1993, Guidebook for risk perception and communication in water resources planning,
Part I: Underpinnings and planning applications, IWR Report 93-R-13, US Army Corps of
Engineers and Institute for Water Resources, Fort Belvoir, Virginia.
Schimmelpfennig, D. and Yohe, G.: 1998, Vulnerability of crops to climate change, in G. Fris-
vold and B. Kuhn (eds), Global Environmental Change and Agriculture: Assessing the Impacts,
Edward Elgar, Cheltenham, pp. 193–217.
Schreider, S. Yu, Jakeman, A .J., Pittock, A. B., and Whetton, P. H.: 1996, Estimation of possible cli-
mate change impacts on water availability, extreme flow events and soil moisture in the Goulburn
and Ovens Basins, Victoria, Clim. Change 34, 513–546.
Shlyakhter, A., Valverde, L. J., and Wilson, R.: 1995, Integrated risk analysis of global climate
change, Chemosphere 30, 1585–1618.

230 ROGER N. JONES
Spano, D., Cesaraccio, C., Duce, P., and Snyder, R. L.: 1999, Phenological stages of natural species
and their use as climate indicators, Int. J. Biomet. 42, 124–133.
Standards Australia and Standards New Zealand: 1999, Risk Management-AS/NZS 4360:1999,
Standards Association of Australia, Strathfield NSW, 46 pp.
Swart, R. J. and Vellinga, P.: 1994, The ‘ultimate objective’ of the framework convention of climate
change requires a new approach for climate change research, Clim. Change 26, 343–349.
Syme, G. J. and Sadler, B. S.: 1994, Evaluation of public involvement in water resource planning: a
researcher-practitioner dialogue, Evaluation Rev. 18, 523–542.
Titus, J. G. and Narayan, V.: 1996, The risk of sea level rise, Clim. Change33, 151–212.
Tucker, G. E. and Slingerland, R.: 1997, Drainage basin responses to climate change, Water Resour.
Res. 33, 2031–2047.
USPCC RARM: 1997, Framework for Environmental Health Risk Management, Final Report
Volume 1, US Presidential/Congressional Commission on Risk Assessment and Risk Manage-
ment, Washington DC.
Visser H., Folkert R. J. M., Hoekstra J., and de Wolff J. J.: 2000, Identifying key sources of
uncertainty in climate change projections, Clim. Change 45, 421–457.
Walsh, K. J. E., Jackett, D. R., McDougall, T. J., and Pittock, A. B.: 1998, Global warming and
sea level rise on the Gold Coast: Report prepared for the Gold Coast City Council. CSIRO
Atmospheric Research, Melbourne, Victoria, 34 pp.
Walsh, K.J., Allan, R. J., Jones, R. N., Pittock, A. B., Suppiah, R., and Whetton, P. H.: 1999, Climate
Change in Queensland under Enhanced Greenhouse Conditions First Annual Report 1997–1998,
CSIRO Atmospheric Research report for the Queensland Government, Melbourne, 84 pp.
Watson, R. T., Zinyowera, M. C., and Moss, R. H. (eds): 1996, Climate Change 1995: Impacts,
Adaptations and Mitigation of Climate Change: Scientific–Technical Analyses. Contribution of
Working Group II to the second assessment report of the Intergovernmental Panel on Climate
Change, Cambridge University Press, Cambridge, 878 pp.
Watson, R. T., Zinyowera, M. C., and Moss, R. H. (eds): 1998, The Regional Impacts of Climate
Change: An Assessment of Vulnerability. A special report of the Intergovernmental Panel on
Climate Change Working Group II, Cambridge University Press, Cambridge, 517 pp.
White, D. H. and Karssies, L.: 1997, Australia’s National Drought Policy: aims, analyses and im-
plementation, Proceedings of the IXth World Water Congress, 1–6 September 1997, Montreal,
Canada, World Water Council and International Water Resources Association (in press).
Wigley, T. M. L. and Santer, B. D.: 1990, Statistical comparison of spatial fields in model validation,
perturbation and predictability experiments, J. Geophys. Res. 95, 851–865.
Woodbury, P. B., Smith, J. E., Weinstein, D. A., and Laurence, J. A.: 1998, Assessing potential
climate change effects on loblolly pine growth: a probabilistic regional modelling approach, For.
Ecol. Manage. 107, 99–116.
Yohe, G. W. and Schlesinger, M. E.: 1998, Sea-level change: the expected economic cost of
protection or abandonment in the United States, Clim. Change 38, 447–472.