Refuge, by Paul Klee, 1930

A 2013 paper, Addressing uncertainty in adaptation planning for agriculture, deserves a wide readership among those finding themselves caught between the desire for certainty and the need for action. The paper was published in 2013 in the prestigious US science journal Proceedings of the National Academy of Sciences (PNAS). It tackles ‘uncertainty’ in climate change research for development. If you can manage your way through the technical jargon, you get to this:

The results demonstrate the potential for robust knowledge and actions in the face of uncertainty.

The paper was written by scientists of the CGIAR Research Program on Climate Change, Agriculture and Food Security and various CGIAR centres, including the International Livestock Research Institute (ILRI), based in Africa.

‘We present a framework for prioritizing adaptation approaches at a range of timeframes. The framework is illustrated by four case studies from developing countries, each with associated characterization of uncertainty. Two cases on near-term adaptation planning in Sri Lanka and on stakeholder scenario exercises in East Africa show how the relative utility of capacity vs. impact approaches to adaptation planning differ with level of uncertainty and associated lead time. An additional two cases demonstrate that it is possible to identify uncertainties that are relevant to decision making in specific timeframes and circumstances. The case on coffee in Latin America identifies altitudinal thresholds at which incremental vs. transformative adaptation pathways are robust options. The final case uses three crop-climate simulation studies to demonstrate how uncertainty can be characterized at different time horizons to discriminate where robust adaptation options are possible.

‘We find that impact approaches, which use predictive models, are increasingly useful over longer lead times and at higher levels of greenhouse gas emissions. We also find that extreme events are important in determining predictability across a broad range of timescales. The results demonstrate the potential for robust knowledge and actions in the face of uncertainty.

Achieving food security under climate change is a complex public policy issue, a so-called “wicked problem.” The magnitude of plausible impacts, and costs of inaction or delayed action, mean that individuals and societies must undertake adaptation actions despite uncertainty.

Policymakers are accustomed to making decisions under considerable uncertainty and do not necessarily need systematic reductions in uncertainty to act on climate change. Nonetheless, science can make a major contribution by elucidating or prioritizing uncertainties in ways that are helpful to the decision-making processes of national policymakers and other stakeholders. The purpose of this article is to demonstrate how science can provide practical approaches to addressing uncertainty that can assist adaptation planning for agriculture in developing countries over multiple lead times. We achieve this goal by presenting four case studies linked by a framework that combines a simple uncertainty analysis with a characterization of different approaches to adaptation planning.

‘Adaptation planning can incorporate scientific information both from projections of climatic impacts and assessments of adaptive capacity. Impact approaches use statistical or mechanistic models to attach probabilities to possible outcomes under a range of scenarios; they arrive at adaptation options for agriculture and food security via analyses that start with climate forcings and global circulation models, and from these project progressive impacts on local climates, crop physiology, crop yields, food prices, and, finally, outcomes for human welfare and nutrition. Capacity approaches start by assessing the existing capacities and vulnerabilities of socioeconomic groups such as communities, industries, or countries. From this base, they develop sets of “no regrets” options that are considered politically and economically feasible over a range of possible climatic futures. Overall, capacity approaches to analysis and planning are more compatible with stakeholder-driven processes.

‘The two approaches also have different implications for uncertainty. Impact approaches have been criticized on technical grounds for the accumulation of uncertainties along the cascade of impact, and exclusion of potentially important factors about which little is known. Global change researchers have put considerable emphasis on quantifying imprecision in projections—e.g, through the use of ensemble modeling techniques. A key concern is that models are more conducive to an emphasis on precision (measurable uncertainty, or known unknowns) than on ambiguity (nondescribed uncertainties, or unknown unknowns).

‘Some critics have gone further to argue that systematic reductions in uncertainty have little or no relevance to policy-making on climate change; worse still, the “uncertainty fallacy” hinders urgently needed action by providing a rationale for delay. Furthermore, complexities in economic and social systems may outweigh climatic uncertainties in determining possible and desirable suites of adaptation actions, thus favoring a capacity approach. However, capacity approaches, though increasingly used in national planning, have received less scrutiny than impact approaches. The treatment of uncertainty in vulnerability assessments is relatively immature, with little explicit treatment of either imprecision or ambiguity.

‘The need to integrate impact approaches with capacity approaches is increasingly recognized. For example, the Intergovernmental Panel on Climate Change’s (IPCC) method for vulnerability analysis integrates an impact assessment of exposure with assessments of sensitivity and adaptive capacity. Arguably, the main challenge is not the technical task of bridging impact and capacity analyses, but rather the effort needed to bridge science and policy. There is a considerable literature on this topic. Recommended strategies include specific go-between roles for boundary organizations or decision scientists. Emerging principles of “consensus beats reality” and “good enough is best” suggest that stakeholder trust and agreement may be more important to effective evidence-based decision making for wicked problems than high levels of scientific rigor and certainty.

‘The uncertainties pertinent to longer-term vs. nearer-term adaptation planning are likely to differ. At low levels of climate change, the climate signal may be indistinguishable from climate variability, and thus improving precision and managing known risks may be more important than identifying completely new (and ambiguous) possibilities. Farmers and societies can adjust by making incremental adaptations and innovations based on long experience in dealing with a highly variable climate. Thus, the key investment in incremental adaptation is likely to be institutional support to farmers to enlarge their portfolio of strategies, both old and new, to manage increasing climatic risks. There are rich historical antecedents for innovation systems for agricultural risk management that share learning among farmers, businesses, scientists, and other stakeholders.

‘At higher levels of climate change, systemic or even transformative adaptation may be needed: wholesale reconfigurations of livelihoods, diets, and the geography of farming and food systems, as has happened historically in response to market changes, for example. These adaptations require different understandings of uncertainty. For example, assessments of seasonal predictability have the potential to improve risk management, such as by informing the financial efficiency, and hence affordability, of index-based crop and livestock insurance.

‘However, at some particular magnitude of climate change, risk insurance will become significantly less effective, and a change in herd size or crop variety (systemic change), or a switch entirely from crop systems to livestock (transformative change) may be needed. Systemic and transformative adaptation will benefit from large-scale, anticipatory investments in infrastructure, livelihoods diversification, and possibly migration. These innovations carry major costs, and in some cases disruptive social changes that are not evenly distributed, and thus can lead to massive misallocations of resources if misjudged.

Shared learning among stakeholders is arguably even more critical to successful adaptation over decadal timeframes than to near-term innovation. The ability of stakeholders at all levels to make long-term no-regrets decisions will be limited by their capacity to envisage and prepare for unknown unknowns. Adaptation planning based purely on stakeholder consensus may lead to maladaptation, particularly where there is future likelihood of entirely novel climates or crossing thresholds in productivity of crops, rangelands, livestock, or fisheries.

‘Here we present a framework for prioritizing adaptation approaches at a range of uncertainty levels linked to lead times. The framework draws on methods to calculate when a climate change signal emerges from the “noise” of climate variability . . . .’

‘Taking our four case studies together, we offer the generalization that capacity analyses are most important for near-term adaptation planning, but impact predictive tools, though useful even in the near term, generally become increasingly important over longer-term planning horizons, which contain increasingly novel climates. Building analytic approaches into iterative stakeholder processes is crucial whatever the timeframe and whatever the combination of impact and capacity analyses. Stakeholder processes for near-term planning may emphasize consensus-building around current knowledge, but over longer time horizons this may shift toward scenarios-based dialogue and priority-setting, as climate change surpasses human experience and major transformations are required. . . .’

Read the whole paper, Addressing uncertainty in adaptation planning for agriculture, by Sonja Vermeulen (CCAFS), Andrew Challinor (CCAFS), Philip Thornton (CCAFS/ILRI), Bruce Campbell (CCAFS), Nishadi Eriyagama (CCAFS/IWMI), Joost Vervoort (CCAFS), James Kinyangi (CCAFS/ILRI), Andy Jarvis (CCAFS), Peter Läderach (CCAFS/CIAT), Julian Ramirez-Villegas (CCAFS/CIAT), Kathryn Nicklin (University of Leeds), Ed Hawkins (University of Reading) and Daniel Smith (University of Leeds),  Proceedings of the National Academy of Sciences (PNAS), vol. 110 no. 21, 8357–8362, doi: 10.1073/pnas.1219441110