AGCIEOC96/ReillySSSII

AGCI Session II: Characterizing and Communicating Scientific Uncertainty

Session Chairs: Dr. Richard H. Moss and Dr. Stephen H. Schneider

July 31 to August 8, 1996

Communicating Estimated Climate Change Impacts on Agriculture

John Reilly

U. S. Department of Agriculture

Washington, D. C.

Different perspectives across scientific disciplines give rise to differences in how to communicate the risks of climate change. Both within and among disciplines, individual researchers bring different sets of experiences with regard to how information they generate will be used. Most of this background experience and expectation of how the information will be used is never addressed directly and instead may be reflected in debates about the accuracy of estimates and what types of information to report. Some of the key background perspectives scientists bring to the table of scientific assessment but rarely discuss directly are: who is the audience, what does this audience already know, and what will this audience do with the information?

Different perspectives across scientific disciplines give rise to differences in how to communicate the risks of climate change.

For the Intergovernmental Panel on Climate Change (IPCC), the audience is obviously the world community of policymakers, yet this is a diverse group. Further, there is a general recognition that the top level ministers will not read the entire report and likely will not even read the Summary for Policymakers themselves. So in addition to different perceptions scientists have about who the direct audience is for the report, they have different views about how information in the report will reach the ultimate audience. What "policy" is, and where, when and by whom it is "made," are also issues on which scientists bring many different perspectives.

Following are three caricatures of the views different scientists bring to the table and the implications for the type of information they see as important to represent in a document like the IPCC report.

View One

Global negotiators on climate change, representing their own country's interests, must make a decision about the best level of climate change control policy to undertake at this time. Once an efficient level of control is determined, negotiators can proceed to design a control strategy that addresses differential distributional consequences of damages and the costs of control across countries. From this perspective, getting an estimate of the global impact on agriculture is the first order of business for a chapter dealing with agricultural impacts. It is important that this estimate represent the best guess scientists can make at this time, even if it is highly uncertain. Moreover, from this perspective, it is useful to denominate agricultural impacts in units that can be added to or easily compared with the units being used in other sectors.

View Two

Policymakers at all levels and the public are ill-informed about the risks of climate change. Policymakers representing different interests and the general public within each country will or should have a voice in formulating a national view on the risks of climate change. The decision process is highly interactive and somewhat chaotic; thus it is not possible to identify or separate different types of policymakers who are responsible for international negotiations. Impact analysis helps determine those sectors/regions/people who are most vulnerable to climate change so that actions can be taken to reduce their vulnerability. Alerting people to the potential negative consequences of climate change is the first priority. It is important to relate impacts in terms that vulnerable groups or those responsible for them can understand. There is little benefit to seeking common units with which to report impacts.

These caricatures of background views on how IPCC information will be used illustrate the considerable potential for conflicting opinions on what to include and how to present material.

View Three

Policy will not be made until the scientific evidence is clear. The scientific staffs of governments will review the quality of the scientific evidence presented in the IPCC process and make recommendations to policymakers on the basis of this evidence. From this perspective, there is a high priority on presenting issues of methodology and background on data quality rather than simply summarizing findings. Presentation of material in such a way that it is credible to scientific peers is essential because this is the principal audience who will interpret the material for others. The focus is on presenting variables and information that science has investigated and on which there is strong evidence rather than on answering questions that arise from outside the community of scientists investigating the problem.

These caricatures of background views on how IPCC information will be used to illustrate the considerable potential for conflicting opinions on what to include and how to present material. This conflict intensifies when it comes to writing a 10- to 20-page policymakers' summary.

Points of Debate

These issues can further be illustrated with some of the specific debates that arose in preparation of the agricultural impacts chapter of the IPCC Working Group II report.

One major point of debate was how to address the issue that estimates of the net impacts of climate change on a global level were small but the regional effects could be large. Some believed that the net global effects were meaningless and should not be reported. There was skepticism about the global results from both sides; some felt they overestimated damages while others felt they underestimated them. In the end, the executive summary was worded very carefully to state that "global agricultural production can be maintained relative to baseline production under climate change as expressed by general circulation models (GCMs) under doubled CO2 equilibrium climate scenarios." The reader is left to judge whether or not this climate scenario is reasonable.

One major point of debate was how to address the issue that estimates of the net impacts of climate change on a global level were small but the regional effects could be large.

The chapters' authors made no attempt to make a statement about the dynamic path of impacts or to identify the period of time in which these GCM scenarios might occur. These studies were based on climate scenarios with global mean temperature changes of 4.0°, 4.2°, and 5.2°C. This range of temperature changes is on the high end of what Working Group I concluded to be the equilibrium climate sensitivity, and also represents global mean temperature changes that seem unlikely to occur before 2100.

If the chapter had been driven more by "View one" described above that policymakers needed scientists' best judgment of what might happen through 2050, which was the general guidance for the Working Group II effort, the author team might well have speculated on the transient damages. Instead, the author team chose not to speculate even though this information is critical for near-term decisions. The lack of real studies of adaptation to transient climate scenarios and the lack of realistic transient scenarios (e. g. , the limited scenarios available did not include the cooling effect of sulfate aerosols) was the basis for this unwillingness to speculate.

Adaptation potential and capability was also a significant topic of controversy. This was reported as an unresolved controversy, recognizing that historically, farming systems had adapted to a variety of stresses, but recognizing that specific research addressing the future rate of climate change had not been conducted. The chapter did go on to assert that adaptation was likely but the extent depended on a number of factors. The overall wording of this section was a compromise between "View one" and "View three."

General guidance for all chapters was to present information on thresholds, sensitivities and vulnerabilities. In early rounds of writing, there was skepticism of the usefulness of reporting information of this type.

As writing progressed, a fair amount of material was added with regard to broad temperature ranges for different crops, indicating that productivity would fall off sharply beyond these limits. Specific examples based on crop modeling studies were cited but no discussion of the extent of these thresholds or the likelihood that they would be encountered under specific scenarios of climate change were reported. In this regard, the report adopted "View two," warning readers of potential hazards without indicating their likelihood. Technical reviewers sometimes focused on this type of reporting as a bias toward reporting negative effects.

There was a debate about whether the chapter should be about the basic agronomic relationships of temperature, precipitation, and CO2 or about the results of models that simulated specific impacts. Fundamentally, it was a debate about what was admissible science.

There was a debate about whether the chapter should be about the basic agronomic relationships of temperature, precipitation, and carbon dioxide or about the results of models that simulated specific impacts. Various technical reviewers offered the idea that either (1) only the basic agronomic relationships should be reported because this was sound science whereas the models used to simulate impacts were not adequately validated under the simulated conditions or (2) only the modeling results should be reported because the basic agronomy results provided nothing new that could not be found in relatively basic texts on crop/climate interactions. In part this was a debate between those holding "View one" versus "View two," but more fundamentally, it was a debate about what was admissible science.

Ultimately, the chapter went relatively far in defining the concept of vulnerability. The chapter defined it as "the potential for negative consequences that were difficult to ameliorate through adaptive measures given the range of possible climate changes that might occur." Identifying a region or population as vulnerable is thus not a prediction of negative consequences of climate change. Rather it is an indication that across ranges of possible climate changes, there are some climatic outcomes that would lead to relatively more serious consequences for that region than for other regions.

It was also recognized that vulnerability could be defined for different aspects of agriculture: yield, farmer income (in terms of regional, national or global economic vulnerability), or in terms of vulnerability to hunger. Vulnerability also depends on the scale at which one examines the problem. Here the chapter attempted to provide a relatively precise definition for a term that is used vaguely in the literature. As defined, vulnerability combines the probability density function for climate and the damage function that relates impacts to varying levels of climate change to generate a loss function.

In practice, the chapter focused on identifying those populations that were subjectively judged to be vulnerable to hunger or famine. While in principle, the probable range of climate is important in this determination, the judgment that went into defining vulnerable populations largely viewed this probability function as reasonably flat across a wide temperature range and with roughly equal probability of precipitation increase or decrease. As such, specific climate model results played little role in the determination. As vulnerability was used in the chapter, it was roughly consistent with "View two" as described above.

Vulnerability combines the probability density function for climate and the damage function that relates impacts to varying levels of climate change to generate a loss function.

A general problem the chapter authors faced was recruiting contributors. About 200 separate quantitative studies were reviewed. Given the limited participation, there was not time to critically assess the merits of individual studies. In this regard, chapter authors mainly summarized the findings and focused less on critiques of the methods and approaches. Berk suggested that there are formalized meta analysis techniques for summarizing and aggregating studies that could be helpful in validating studies.

The chapter spanned a wide range of issues, from understanding crop/climate/insect interactions and the science of carbon dioxide fertilization to the socioeconomic conditions under which hunger occurs and the economic modeling of agricultural trade. This added to the difficulty in enlisting experts with deep knowledge in these many subject areas because in the end, an expert's contribution would likely represent only a few paragraphs.

Reilly stressed that future research into agricultural impacts should involve the development and broad application of integrated agricultural modeling efforts, development of the capability to readily simulate agricultural effects under transient climate scenarios so as to better address the cost of adjustment, and evaluation of the effects of variability and changes in extreme events.

Forward to Next Section//Back to Table of Contents//AGCI Homepage//Comments: agcimail@agci.org