Climate Variability and Change: Embracing Complexity and Uncertainty

Julie WinklerLast spring, at a listening session I attended on climate variability and change in northwestern Michigan, a local fruit grower summarized his concerns with the analogy that his industry is the “canary in the coal mine” for the potential impacts of climate variability and change on agriculture. This statement was motivated by the sensitivity of fruit production to climate extremes, particularly changes in the frequency of devastating spring freezes, and the limited short-term adaptation options given the relatively long-term investment of orchard blocks.

Geographers have increasingly become involved in assessments of the vulnerability to, and potential impacts of, climate variability and change. These challenging interdisciplinary endeavors are providing many geographers with exciting opportunities to work collectively with scientists from a range of disciplines, interact directly with stakeholder groups, and engage in research that is not only stimulating but also has considerable applied significance. I am concerned, however, with what I see as a continuing tendency in assessment studies to downplay the complexity and uncertainty of the potential impacts of climate variability and change.

Several years ago, in an editorial in Environmental Science & Technology, Baruch Fischhoff, a well-known decision scientist, argued that scientists, although traditionally trained to consider uncertainty, multiple approaches and a range of data sources, often turn to an advocacy-based communication when they are highly concerned about the potential consequences of either action or inaction and/or when they believe that the “science will not receive a fair hearing.” In advocacy-based communication, a case is made for a specific viewpoint and uncertainty is introduced only through arguments with contrasting viewpoints. Although advocacy-based communication has its place, a potential consequence is the loss of confidence in, and appreciation for, science by the general public. As an alternative, Fischhoff argued for what he refers to as nonpersuasive communication, an approach that explicitly considers uncertainty and “allows science to speak for itself.” From Fischhoff’s perspective, communication of climate variability and change involves climate scientists, or more generally domain scientists, who develop the information to eventually be communicated and confirm that it is scientifically sound, decision scientists who help identify the information relevant to a particular decision, and social scientists who work to overcome communication barriers.

Personally, I have long been uncomfortable with communication regarding climate variability change that fails to convey the associated complexity and uncertainty, particularly the many limitations of climate observations and projections, with which I am all too familiar as a geographer/climate scientist. Thus, Fischhoff’s argument for nonpersuasive communication of climate variability and change resonates strongly with me, although I would expand Fischhoff’s model to include a broader range of experts as domain scientists and would blur the distinctions between the domain, decision, and social science experts, emphasizing instead the communication among experts and between experts and stakeholders.

Climate scientists are not the exclusive domain experts in the communication of the potential impacts of climate variability and change. In fact, few stakeholders can directly incorporate future projections of climate variables in their decision-making. Rather, stakeholders require information on changes in climate-influenced parameters of relevance to their activity or industry. Expertise from a range of disciplines is needed, including social science (e.g., human geography, economics, demography) whose involvement extends well beyond overcoming communication barriers to the development and evaluation of information required for decision-making. For example, while growers of commodity crops (e.g., maize, soybeans, and wheat) are cognizant that changes in temperature and precipitation during the growing season will affect their operations, projected changes in yield and farm income are much more relevant parameters for their decision-making.

Furthermore, inferring potential yield or income from simplified climate scenarios (e.g., change in growing-season mean temperature and precipitation) is suspect given the complex relationships between weather/climate and yield, and between yield and income. Consequently, climate scientists, agronomists, economists and others need to collaboratively explore, in a scientifically sound manner, the ways that a perturbed climate may influence yield and, subsequently, profitability and livelihood.

The concept of the “usability” of assessment outcomes also needs to be broadened. Although a number of previous authors have implored climate scientists to consider the “usability” of their observations and projections, even chiding them for the too often opaqueness of the metadata (when provided) of climate information, the usability of the outcomes of the different impact models employed in an assessment, such as yield models, is less often considered. In addition, one can argue that stakeholders should be part of, rather than separate from, the assessment team, working with decision scientists to identify the information relevant to the decisions that they will be making, and with domain scientists to facilitate the co-creation of that information.

As someone involved in the development and use of climate projections for local/regional assessments, I am often asked by scientists from other fields for advice on the availability and suitability of climate information for a particular assessment. Lately, I have been somewhat disheartened by the number of requests I receive for “simple” climate scenarios (often little more than a projected change in mean temperature and precipitation). To be sure, simple scenarios, even “what if” scenarios, are extremely useful, particularly for vulnerability assessments, and they complement more detailed projections which, in conjunction with suitable impact models, can illuminate potential “surprises” that fall outside stakeholder experience. I am more concerned that a reliance on simplistic projections, especially when paired with relatively unsophisticated impact models, will fail to fully illuminate the complexity and uncertainty associated with climate variability and change, and fail to provide the information needed for robust decision-making, in contrast to when a plurality of approaches — both simple and complex — are employed. I have also been rather dismayed by the disconnect between the very fine spatial resolution at which climate information frequently is requested versus the information content of the scenarios which often varies much more broadly in space.

Another concern is the lack of consideration of the assumptions of the impact models that will be employed in an assessment in the context of the nature and limitations of climate information, or of the contribution of the impact models themselves to the uncertainty of the assessment outcomes. That said, several recent publications represent initial steps in addressing these concerns. In particular, a recent analysis conducted at the University of California-Berkeley illustrated that the high degree of spatial autocorrelation in gridded climate observations can violate the independent assumption of empirical economic models that are often used in assessment studies and recommended that station observations may be the more appropriate choice of climate information for the development and application of these models. Also, members of the AgMIP (Agricultural Modeling Intercomparison and Improvement Project) team recently demonstrated that uncertainty introduced in future projections of wheat yield by the choice of yield model was as large or larger than the uncertainty introduced by an ensemble of climate projections. Both these studies point to the need for careful attention to the assumptions of impact models and to the necessity of evaluating the uncertainty surrounding all components of an assessment, rather than just the uncertainty of the climate information.

Geographers are in a unique position to develop enhanced approaches for climate assessments that improve the usability of assessment outcomes and to advocate for nonpersuasive communication in decision-making that embraces complexity and uncertainty. Geography is an “interdisciplinary discipline.” We regularly and effectively work across the many subfields of Geography and across disciplinary boundaries. We are also sensitive to disciplinary differences in research culture, methods and approaches, and, therefore, can help facilitate a more seamless integration across assessment components. Geographers are already actively involved in assessment efforts, but there is much more that we can do to advance new assessment approaches. The fruit grower in northwest Michigan, and the many others facing complex choices in an uncertain future, could use our help. Let’s step up to the task.

—Julie Winkler

DOI: 10.14433/2013.0023

For more information on articles referred to above see:

Nonpersuasive Communication about Matters of Greatest Urgency: Climate Change” by Baruch Fischhoff in Environmental Science and Technology, pages 7204-7208, November 1, 2007

“Uncertainty in Simulating Wheat Yields under Climate Change” by Asseng et al. in Nature Climate Change, Volume 3, pages 827–832, 2013, DOI:10.1038/nclimate1916

“Using Weather Data and Climate Model Output in Economic Analyses of Climate Change” by Auffhammer et al., NBER Working Paper No. w19087, 2013, DOI:10.3386/w19087

Photo credit: Dwight Burdette; Apple orchard on Wassem Fruit Farm, Augusta Township, Michigan