Leave the text-only version and return to the original page view.
The previous section demonstrated how climate change impacts could be incorporated into the EA analysis. This section examines how the resulting climate change uncertainties could be effectively communicated to the disparate stakeholders of the EA process. The involvement of various stakeholders is one of the most important aspects of EA processes in Canada. Wide-ranging views provided by stakeholders can contribute considerably to an EA. For this involvement to be both meaningful and effective, the appropriate information derived from the impact studies needs to be clearly communicated to, and understood by, a wide range of audiences. Stakeholders in EAs require an open forum for airing their own opinions and their management of environmental uncertainties; and their criteria, regarding the quality of information, can be quite different from those of experts (while also differing considerably among themselves). In such situations, where opinions, knowledge, and ignorance can become so intertwined that traditional categorizations prove scarcely applicable, intellectual discipline and practical guidance in the communication of information about uncertainties (and corresponding levels of ignorance) should help considerably in the clarification of the debate.
Depending on the data, models, and assumptions used to estimate the impacts, stakeholders will possess different levels of "trust" in the results. In policy processes for decisions on environmental problems, where scientific results have frequently been obtained from computer models, these results can be especially controversial. Given the contentious debates within and outside scientific circles about the degree of changes that could be caused by GHG emissions, considerable differences among stakeholders may exist in their acceptance of any results from a scenario analysis. For example, a study based on a certain set of scenarios may conclude that the annual risk of downstream flooding as a result of climate change would increase from 0.001 to 0.002. However, downstream residents may believe that this range is far too conservative since more extreme or "surprise" scenarios had not been considered in the analysis. This type of apprehension and uncertainty regarding climate change uncertainties and data reliability must be appropriately conveyed to the users of the EA analysis.
What seems imperative is the ability to examine and communicate information that will help stakeholders and decision-makers assess their beliefs in the estimates and to be able to readily discuss these as an integral part of the EA process. A well-designed EA process could increase consensus, and also enhance awareness, on the uncertainty and quality of information, without embroiling its users in contentious debates over peripheral issues.
Irrespective of the degree of stakeholder consensus, the communication of climate change uncertainties can be particularly complicated. Various studies have made different recommendations on what issues and factors should be considered for stakeholders to attain a good understanding of the uncertainties and their implications, and how this information can be effectively communicated to them.
Most environmental systems are representative of highly complex systems that are complicated by numerous inherent uncertainties and by a plurality of stakeholder perspectives (Chociolko, 1995; Funtowicz and Ravetz, 1999). Although science has long been viewed as providing an objective source for definitive answers, science now is recognized as seldom being absolute when applied to the complex systems addressed in public policy (Funtowicz and Ravetz, 1984, 1990, 1999). Even when scientific methods are sufficient to establish ranges and likelihoods of uncertainties in environmental systems, these expressions neither communicate any indication as to the quality of these estimates, nor do they permit ready comparisons to be made between disparate estimates and methods. When uncertainty and variation are understated or suppressed in the communication of public policy, a false impression that "everything is known" becomes pervasive (Hammitt and Shlyakhter, 1999).
To bridge this deficiency requires the development of a framework to convey the gaps in existing knowledge by capturing and expressing the inexactness, unreliability, and borders-of-ignorance that are present in the scientific estimates of uncertainty. Researchers and practitioners have examined numerous different approaches for processing and communicating uncertainties inherent in complex systems. These practices have ranged from the very quantitative to the very qualitative, and from the mathematical to the entirely graphical. In this section, several possible approaches are briefly summarized and their suitability for the communication of climate change uncertainties in project EAs is considered.
In preparation for the IPCC's Third Assessment Report (TAR), Moss and Schneider (2000) assessed several means for characterizing climate change uncertainties and prepared a guidance paper for use by all TAR authors. Noting the need for a consistent approach, Moss and Schneider (2000) proposed not only a general process for assessing uncertainties, but also several specific tools that could be used to communicate them. Six general steps were recommended for assessing the uncertainties in TAR. These steps can be summarized as follows:
One important recommendation was that extreme care has to be taken to avoid vague and/or overly broad statements that prove difficult to either support or refute. It was further recognized that because words used as descriptors can hold very different meanings to different stakeholders, verbal descriptions of scientific information must be calibrated consistently. Therefore, for communicating uncertainties in the TAR report, verbal confidence descriptors should be translated according to the quantification system shown in table 16.
| VERBAL DESCRIPTOR | LIKELIHOOD RANGES | |
|---|---|---|
| FROM | TO | |
| Very High Confidence | 0.95 | 1.00 |
| High Confidence | 0.67 | 0.95 |
| Medium Confidence | 0.33 | 0.67 |
| Low Confidence | 0.05 | 0.33 |
| Very Low Confidence | 0.00 | 0.05 |
Moss and Schneider (2000) recognized that achieving this translational precision in the adoption of such a quantification scheme is seldom possible. Therefore, to permit a more consistent qualitative communication of the state of knowledge, they proposed that the more general verbal descriptions shown in table 17 be applied when more quantifiable details were not readily available.
| LEVEL OF AGREEMENT &/OR CONSENSUS | AMOUNT OF EVIDENCE | ||
|---|---|---|---|
| Low | High | ||
| Low | Speculative | Competing Explanations | |
| High | Established but Incomplete | Well Established | |
This qualitative classification scheme could provide a valuable basis for communication in the TAR by using a vocabulary that is straightforward for users to actually understand and implement. Unfortunately, the classification scheme necessitates that some highly subjective judgment be applied to partition specific categories into their "high" and "low" levels (i.e., where does the boundary/threshold between these two categories actually lie?), and there is no universally available process to accomplish this task.
Finally, Moss and Schneider (2000) suggested that ancillary approaches for characterizing the uncertainties associated with key findings can increase the clarity of the conclusions achieved in the TAR. They concluded that communication of these uncertainties can be significantly improved using graphical representations of the results, and provided two explicit suggestions for this expression using:
However, the two methods provided are not useful for representing situations involving multiple attributes. The authors advise that any selection of graphical presentation for communicating uncertain quantitative information should be left to the specific user. While the information for communicating uncertainties supplied by the IPCC is applicable and useful at the macro level, these cannot be directly transformed into a specific framework for clearly communicating climate change uncertainties to the different stakeholders of an EA.
Recognizing the inexactness, unreliability, and borders-of-ignorance prevalent in scientific estimates of uncertainty, Funtowicz and Ravetz (1990) propose the NUSAP formalism for communicating both quantitative and qualitative evaluations of the prevailing uncertainty in environmental and social research. NUSAP is an acronym for Numeral,Unit, Spread, Assessment, andPedigree. The Numeral refers either to a number or to a set of elements and relations that expresses magnitude, intervals, rankings or even non-numerical descriptions such as "small", "medium", and "large". Numeral is the most quantitative element within NUSAP. Unit refers to the scale of what is being measured, such as "grams", "watts", "speed", and "number of different species". The Spread expresses the range of uncertainty or inexactness in the numeral estimate, and could be expressed as a variance or standard deviation; a specific confidence interval; a percentile estimate; lying within a factor of " n" of the estimate; lying within a logarithmic range of the estimate; or any other form of entropy measure from information theory. Assessment provides an estimate of the reliability associated with the information. Assessmentcan be expressed more quantitatively through confidence limits and significance levels, or more qualitatively using less formal descriptors for expressing degrees of optimism in the estimates through the use of such terms as "high", "medium", and "low" quality.
The Assessment component provides an evaluation of the pragmatic quality of the overall information and enables communication of where problem areas may exist through an expression of the (un)reliability associated with the quantitative information conveyed in the Numeral, Unit, and Spread categories. Finally,Pedigree is the most qualitative and complex of all categories and provides an overall quality measure by communicating the scientific basis underlying the methods that were used to actually produce the quantitative estimates. This basis conveys the comparative border with ignorance by displaying what more powerful means were not deployed in the production of the information (Funtowicz and Ravetz, 1990). Together, Assessment and Pedigree express conceptual operations of criticisms and evaluation on the entries in numeral, unit, and spread categories. Therefore, these categories tend to be very reflective in nature and can foster a degree of quality "self-awareness" among users.
NUSAP provides the first framework for explicitly incorporating issues of information quality into communication of scientific estimates. While the different components of NUSAP have been individually considered previously in scientific documentation (included generally through accompanying text, footnotes, and appendices), these earlier expressions of information quality tended to provide an ineffective mechanism for communicating the impact from these components and, therefore, often were ignored in policy formulation. The NUSAP formalism places the qualitative components into as prominent a position as the specific numerical estimates, thereby communicating both the quantitative estimates and the quality of these estimates concurrently. Thus, NUSAP can be used to indicate important trade-offs between uncertainty and reliability and quality. For instance, while decreasing the precision of an estimate may increase its reliability, the actual quality of the data may not have been sufficient to justify the higher precision in the first place.
Although NUSAP provides a general methodology for communicating both information and uncertainty, the system provides no explicit guidelines for actually developing specific qualitative categories related to uncertainty and reliability (i.e., the Assessment and Pedigree categories). Furthermore, the subjective operations related to the assessment and pedigree categories can neither be performed by "automatic" means nor can they be accomplished in isolation from the whole body of relevant scientific knowledge. Therefore, to communicate effectively using NUSAP, the various stakeholders would need to become familiar with and to develop extensive craft skills in NUSAP methodologies, in addition to those skills required for specific application under study. The convolution of these supplementary technical-skill requirements would add unnecessary extra layers of complexity to the communication of climate change uncertainties.
The World Resources Institute (WRI) presented a report that sought to identify gaps in data and information on global ecosystems in which the data quality was described using qualitative narratives and data descriptors such aslacking, not reported, not available,good-quality, anecdotal, and expert opinion (WRI, 2000). An example of a narrative statement illustrating a lack of information might be that the "data are limited by a difficulty in identifying species and assessing their impact" (Inch, 2001). This narrative approach may provide the only possible way to communicate the extent of situations covered in an analysis devoid of sufficient quantitative data (such as that undertaken by the WRI). However, broader applicability of the method is effectively hindered by its inability to provide any sound basis for comparison across different issues, as may be required in an EA.
An analysis by a Commission of the Environmental Protection Agency (EPA) of methods that could be used to reflect uncertainties in measurement and estimation techniques concluded that qualitative descriptions were needed for most risk assessments, but determined that a quantitative uncertainty analysis of risk estimates was seldom necessary (EPA, 1997a, 1997b; Inch, 2001). The EPA report placed considerable emphasis on distinguishing variance from uncertainty, where variance was considered to be the natural diversity/variability that could be known, while uncertainty represented unknown or only partially known information. Although strongly supporting the use of mathematical descriptions of variability, the Commission remained very "...doubtful that much value would be added… by formal mathematical analyses of uncertainty". However, it was recognized that words used as descriptors could hold very different meanings to different stakeholders, and therefore these descriptors should becalibrated according to a quantification translation (similar to that in the IPCC study above). Unfortunately, as with the IPCC recommendations, the details required for such quantification are insufficient to characterize the multiple climate change uncertainties.
The National Institute of Public Health and the Environment in the Netherlands (RIVM) created a "universally encompassing" taxonomy of all types of uncertainty in any subject area by identifying two meta-level sources of uncertainty;variability and limited knowledge (van Asseltet al. 2001). Under this scheme, the sources of variability uncertainty are categorized into the inherent randomness of nature; value diversity among people; the unpredictable, macro-level societal processes of social, economic, and cultural dynamics; non-rational human behaviour; and unexpected developments and/or consequences arising from technological surprises. The sources of limited knowledge uncertainty are categorized into inexactness arising from lack of precision, metrical uncertainty, and/or measurement error; a lack of observations and measurements; things that are immeasurable in practice; conflicting evidence; processes currently unknown but likely to become known; processes that are understood in principle but can never be fully predicted; and processes that cannot be determined. Although many of these sources of uncertainty are valid, not all are particularly useful (i.e., processes that cannot be determined) for decision-making (Inch, 2001).
RIVM developed a comprehensive framework for structuring an assessment of these uncertainties that combined a technical analysis with processes derived from social theory (van Asseltet al. 2001). The five primary steps in this framework involve:
A full application of the framework necessitates that lengthy consultations be undertaken to identify the perspectives of the various stakeholders. When multiple perspectives exist (as would be the case in almost any EA), the magnitude of work required to conduct all of the steps in the framework can quickly prove intractable. While this framework is valuable because it integrates technical and value-based human aspects into complex system principles, the approach does not propose any of the tools requisite for performing risk comparisons. By only supplying a general framework, the approach provides a limited tool for assessing uncertainty. Therefore, its usefulness (if any) for communicating climate change uncertainties in EAs appears very restricted.
Similar to the RIVM approach, Stirling (1998) produced a structured framework method for categorizing uncertainties, but with guidance as to methods that could be used to address the uncertainty type. This framework is summarized in table 18 (Stirling, 1998).
LIKELIHOODS | OUTCOMES | |
|---|---|---|
| Known | Poorly Defined | |
| Firm basis for probability | Risk Apply Frequentist Methods | Ambiguity Apply Sensitivity Analysis |
| Weak basis for probability | Risk Apply Bayesian Methods | |
Unknown | Uncertainty Apply Scenario Analysis | Ignorance Apply Precaution |
The Stirling framework characterizes situations as risk, ambiguity, uncertainty, or ignorance. These framework distinctions can be more broadly communicated as referring to those types of situations:
The Stirling framework is robust in content but weak as a communications device, as it contains terms that are both easily confused and expressed in a language that has "common usage" connotations that are different from the specific intent of the framework (Inch, 2001). To actually apply the Stirling framework, the target audience would need to both learn the classification system and apply new definitions to commonly used words. Such user-learning requirements could be problematic in EAs given the wide diversity of stakeholder skill-sets.
Richards and William (1999) introduced a "meta level" descriptive approach that partitioned uncertainties into four fundamental classification schemes. These schemes describe the nature of uncertainty as being either:
Inch (2001) describes how this scheme classification would be very difficult to apply to any complex system that has not already achieved steady-state. Given that the impacts from climate change are expected to occur on an escalating basis over the course of several centuries, and that a steady-state could only occur if GHG emissions are effectively controlled (and all feedbacks are inconsequential, see section 7), it would appear that the conditions needed for the application of this method would not be satisfied. Therefore, such a meta-level, qualitative style of analysis does not seem to hold any direct promise for effectively addressing how the assessment and communication of climate change uncertainties could be incorporated into EAs.
Under specific circumstances, each approach reviewed in the previous section possesses beneficial features that could be useful for communicating the uncertainties of climate change to stakeholders in project EAs. While the methods provide numerous alternatives for uncertainty communication, there is an underlying requirement for stakeholders to understand each approach that is used, which introduces a major hurdle for their implementation. Consequently, the overarching observation must be that none of these individual methods supplies a universally adaptable framework that could become the "gold standard" for communication.
Given the needs of decision-makers to weigh potential responses to climate change before all uncertainties can be resolved, the available information (imperfect as it may be) must be synthesized, evaluated, and communicated in a responsible and informative manner (Moss and Schneider, 2000; Ravetz, 1986). Therefore, the potential impacts from uncertainties should be effectively communicated somehow to the different stakeholders, irrespective of the intrinsically obvious difficulties in this communication. With considerable uncertainties and disagreements about climate change, this communication of uncertainties should accomplish two fundamental tasks:
Since the stakeholders will almost always be drawn from a wide range of constituencies possessing dissimilar technical skill-sets, adoption of the following recommendations (that have integrated ideas from the methods in the previous section) is suggested as the minimally prudent level of communication within EAs.
To accommodate disparate backgrounds and levels of knowledge among stakeholders, presentation of information should appear as a comprehensive verbal description written in an accessible, non-technical, clear, and concise format. Since such explanations would necessarily require a significant amount of work by the proponent, detailed verbal descriptions are warranted only for the most significant elements and for communication of their resulting uncertainties (Moss and Schneider, 2000). The proponent should identify the key components in the EA and articulate the nature of the major uncertainties inherent within each of these components. Specifically, the proponent, with input from stakeholders, should determine which models, data sets, assumptions, and results constitute key components. So stakeholders will be able to assess their degrees of beliefs and confidence in both the resulting estimates and underlying uncertainties, verbal explanations should be provided concerning the acceptability of each of the key components identified for the EA, namely:
Providing information about these factors can be challenging. But the inclusion of clear and detailed summaries of each key component is particularly important in the verbal documentation. Sufficient care must enter into this process, to ensure that technical meanings behind data presented in the description make the underlying implication readily accessible to the layperson.
Therefore, the following interrelated types of information should be provided within the verbal description to help assess the degrees of acceptability of each identified key component. For eachmodel used, the following should be identified:
For each set of data employed, the following should be noted:
For each key assumption made, the following should be stated:
For each set of resulting estimates, the following should be stated:
A summary assessment regarding the general level of overall confidence in the component should also be provided.
For the hydro example in the previous section, the scenarios used as the data source can be considered the key components of that study. If a verbal assessment were applied to these scenarios, the resulting description may be as follows. The identifiedsources for the scenario models used in the example were the IPCC and the CICS. These scenarios could be thought of as reflecting a particular school of thought as to future climate change outcomes, yet their actual representation of reality at this time is unknown. The scenarios have been subjected to extensive peer review due to ongoing worldwide use, but the general acceptance of them can only be consideredvariable, since numerous climate change "dissenters" exist. Thus, the degree of acceptance depends inherently on the specific viewpoints held by the user. There arevarioussources for the data in the scenarios but, overall, the figures provided should be consideredprimary data. The key assumptions used in the scenarios can be considered only a mediumrepresentation of reality, since the scenarios have been constructed from consensus opinions (i.e., highly "extreme" beliefs will have been averaged out) and, as such, the acceptance of these assumptions can be quite variable. Theresulting estimates of the scenarios have been independently reviewed and their overall acceptance can be considered as being at the medium level. This implies that while many people currently accept these scenarios as the best expert judgement available at this time, numerous dissenters and/or supporters of alternative interpretations exist. Therefore, the overall confidence in the scenarios used in the hydro example can be considered as being only at amedium level of acceptance.
This type of assessment of the key components can be summarized in a convenient tabular form as shown in table 19. The table shows an assessment of each of the three key interrelated components (the scenarios, the streamflow model, and the energy model) used in the Hydro example. Each column corresponds to a summary evaluation of a specific component. For example, the scenarios column summarizes the assessment given in the previous paragraph. When estimated impacts result from the product of a series or concatenation of components, the overall quality and acceptability of these results will depend on the combination of the quality of each of the individual models, data, and assumptions involved. For example, the overall confidence in the estimates of energy production in the hydro example depends on the acceptability of the scenarios, streamflow, and energy components. To communicate the acceptability of these combined results, an evaluation of the acceptability of each key constituent component would be needed, together with anoverall assessment of their combination.
While proponents should carefully explain the purpose and usefulness of a summary table as shown in table 19, the primary communication vehicle should be the verbal descriptions of the uncertainties in the components and related issues.
Since proponents, or their consultants who do this work, may not be able or comfortable providing an assessment of their own work (e.g., state that they have only "low confidence" in their results), requiring the proponent to have an independent reviewer provide this assessment would be an alternative approach.
| Stage 1: Scenarios | Stage 2: Streamflow | Results: Energy | ||
|---|---|---|---|---|
| Model: | Source | IPCC/CICS | Consultant | Consultant |
| Rep. of reality | Unknown | Medium | High | |
| Theory/Sch.of thought | School | Est. Theory | Est. Theory | |
| Peer review | Yes | No | No | |
| Acceptance | Variable | - | - | |
| Data: | Source | Various | MNR | - |
| Primary/Sec. | Primary | Primary | - | |
| Theory/Sch.of thought | - | - | - | |
| Key Assumptions: | ||||
| Rep. of reality | Medium | High | High | |
| Acceptance | Variable | High | High | |
| Resulting Estimates: | ||||
| Indep. review | Yes | No | No | |
| Acceptance by review | Medium | - | - | |
| Overall confidence | Medium | Low- Medium | Low- Medium | |
In addition to presenting information about the key components, information about the uncertainties in the quantitative and qualitative results (e.g., impacts on energy production) should be presented in a way that is useful for stakeholders and decision makers. To convey the uncertainties in the quantitativeinformation, the following types of summary information (also expressed in an accessible verbal fashion) should be presented (with the possible inclusion of supporting figures and tables) in the EA:
For impacts that are measured qualitatively, uncertainties only can be described and presented with considerably less precision. The following types of summary descriptions should be provided for the uncertainties about qualitative estimates (again with possible supporting tables and ancillary devices):
Furthermore, when imprecise, qualitative terms and descriptors (such as "low", "high", or "significant") are used, the basis underlying their particular application needs to be clearly explained.
This section has reviewed a number of possible methods that could be considered for communication of climate change uncertainties to stakeholders in project EAs. Each proponent will need to decide which methods are best to use in their EAs based on the needs of the stakeholders. The chosen methods should clearly communicate information about uncertainties in the results, as well as information about the degree of belief in the key components that led to the results. The information should be presented in a format that will be understandable to the range of stakeholders.
