Integrated approaches to managing uncertainty

By Brian Bevirt
11/06/2012 - 12:00am
David Budescu
David Budescu of Fordham University lectures on judging and communicating uncertainty. —Photo by Brian Bevirt, CISL

Workshop participants
Early-career scientists discussed real-world climate change problems during working group sessions. —Photo by Toni Rosati, CISL

"Doubt is not a pleasant condition, but certainty is an absurd one" —Voltaire

In broadest terms, uncertainty refers to a state of incomplete knowledge or lack of knowledge. It is usually viewed as having two major components: a random component reflecting random processes and an epistemic component, which refers to incomplete knowledge due to the complexity of the world. In the context of climate change, there are many different types of uncertainty in different parts of the problem. Many, but not all, are related to mathematical models used in climate change research.

Computer models are useful for planning because they allow researchers to project future events by simulating complex physical, economic, or ecological systems with mathematical representations. Since simulations involve simplifications, however, models cannot be 100% accurate. While models allow us to project trends for hundreds of years, the details of long-term projections by different models do not necessarily agree.

The sources of uncertainty include approximations (called parameterizations), imperfections in the input data used to initialize model runs, limited computing resources that prevent modeling all aspects of relevant systems, and the fact that — by definition — a model is a simplified representation of the system it simulates. Moreover, uncertainty exists because the natural internal variability of the climate system cannot be predicted. Another type of uncertainty arises when the output of one model is used as input for another, like when climate model output serves as input to crop models and hydrology models to assess societal impacts. This compounding of uncertainties may affect society's trust in the models' projections of future climate and its possible impacts. Therefore, modeling experiments must be carefully planned to reasonably represent the uncertainties in models used in climate change research.

From 6–17 August, CISL's Institute for Mathematics Applied to the Geosciences (IMAGe) presented its second 2012 Theme of the Year event, "Uncertainty in Climate Change Research: An integrated approach." Its three co-directors, Haley Fowler (University of Newcastle, UK), Chris Forest (Pennsylvania State University), and Linda Mearns (IMAGe) designed the experience to introduce young scientists to strategies for supporting good decision making despite multiple types and levels of uncertainty. Recognizing, assessing, and managing compounded uncertainties is daunting when researchers begin to appreciate the complexities. And uncertainty is present in all phases of climate change research and applications, from the physical science (e.g., projections of future climate), to the impacts on society, to the decision-making efforts about mitigation and adaptation problems.

Kristen Pitts
Kristen Pitts is an environmental biologist and ecologist at Southern Illinois University, Carbondale who works with aquatic species distribution models that are used in climate and agricultural land-use scenario analysis for the U.S. Midwest. Her work informs managers, stakeholders, and policy-makers about the ecological tradeoffs of different policy options. —Photo by Toni Rosati, CISL

Thirty-one graduate students, postdocs, and early-career scientists were selected from a worldwide pool of almost 160 applicants who included climate science modelers, statisticians, societal impacts modelers, societal vulnerability researchers, and interdisciplinary researchers doing integrative work. The next-generation scientists in this IMAGe workshop responded very positively to the challenges of the many types of uncertainty involved in the entire climate change problem. Their feedback comments included: "For me, truly understanding the extent of uncertainty in climate models and thinking about 'bottom-up' approaches for understanding impacts was most valuable. I plan to incorporate this line of thinking into my dissertation." And, "It was frustrating at the time, but in the end it was a good learning experience in seeing how different fields approach problems differently and in the challenges of communication in interdisciplinary groups." And, "[The workshop's greatest value was] having the time to think about the big picture and exploring (facing up to!) uncertainties which often seem too daunting to do in day-to-day work, e.g., high impact events. The group projects really gave us time to do this."

Participants praised the strategies and techniques for managing uncertainty that were discussed and practiced in small groups. These groups were tasked with solving real-world climate change problems or planning research programs that would help society cope with climate change. The workshop's approach to managing uncertainty begins with needed societal decision-making, then considers the uncertainties in the various parts of the climate problem that may be relevant to making these decisions. In addition to focusing climate modeling, this approach could improve projections of greenhouse gas concentrations, relevant vulnerabilities, and involvement of future governance structures.

Jonas Bhend
Jonas Bhend is a meteorologist and climate researcher at CSIRO Marine and Atmospheric Research. He has explored climate change impacts on the winemaking industry with interest in management. —Photo by Toni Rosati, CISL

Deciding how to quantify these uncertainties formed an important part of the topics covered. All of these topics were pursued with the ultimate goals of supporting efforts to cope with climate change and of reducing risks to people and infrastructure. If the research goal is to make good decisions for the benefit of society, then those decisions are the proper context for evaluating which uncertainties to consider. Rather than expending resources trying to reduce all uncertainties, researchers are far more effective when they use the decision-making context to focus on managing only the most relevant uncertainties.

The causes and consequences of climate change are not distributed equally among people, places, or moments in time. The issue of vulnerability fits into the overall climate change debate as a fundamental part of understanding why and how impacts are distributed across scales from individuals to economic sectors or geographic regions. It is therefore important to determine effects (where, who, or what will gain or lose) to better understand climate change impacts on the coupled human-natural system – as well as to inform policy and ethical decisions about adaptation priorities and response efforts.

Workshop participants
This photo shows almost all of the workshop participants at NCAR's Foothills Lab campus. —Photo by Brian Bevirt, CISL

The participants were organized into six working groups – each having one person from each of these five areas – to integrate multiple types and levels of uncertainty into mitigation and adaptation guidance for decision makers, stakeholders, and policy makers. One such project was: How can scientists help guide city planners who need to allocate scarce resources to water management projects over the next 10 years? The participants were introduced to their groups on the first day. They blended the insights from their diverse expertise to solve problems that required integrating different types of uncertainty. The projects took various forms, including developing adaptation plans, preparing letters of intent for an NSF DMUU call, and developing modules for an Integrated Assessment Model. Decision-making contexts tended to be those for adaptation, but the importance of mitigation decisions was also acknowledged.

Stephanie McAfee
Stephanie McAfee is an interdisciplinary climate researcher at the University of Alaska at Fairbanks. Her current research is focused on climate change, variability, and developing actionable research products for stakeholders concerned with a variety of impacts. Her interests include effective ways to communicate climate information to stakeholders. —Photo by Toni Rosati, CISL

Participants learned how to glean benefits from each other's areas of expertise to manage the uncertainties and improve outcomes for the people and infrastructure affected by possible climate change. Mentors from various disciplines circulated among the six groups to provide insight and direction. Each working group then presented the results of their projects at the end of the second week.

In addition to the working group projects, 24 researchers and practitioners presented 32 lectures about the issue of uncertainty in all aspects of climate change research: climate science, impacts, vulnerability, policy making, governance, and the structure of decision making. Discussion sessions and panels offered plentiful opportunities for interaction between the presenters and the participants.

This workshop was unique because it introduced so many types of uncertainties to early-career researchers who began to appreciate the big picture in climate change research that includes the nature and scope of the uncertainties, where they arise, how they are quantified, and how we can manage them. The young scientists were open to this interdisciplinary investigation and eager to see the whole problem: not to become generalists, but to gain a clearer view of how their research could fit in, how they can connect with researchers in other disciplines, and what the long-term effects on their research could be.

The Uncertainty Cake
The Uncertainty Cake was served during the workshop's first-week reception. —Photo by Chris Forest, Penn State University

The most fundamental outcome was an intensive learning experience for very bright young scholars who grappled with the concept of integrated uncertainty and what it could mean for them and their research. To summarize, organizer Linda Mearns said, "The workshop was designed to help researchers in all disciplines by helping them do their work with an understanding of the larger context, which can influence what they pursue and thereby produce results that can lead to better management of the climate change problem. Consciousness raising about uncertainty throughout climate change research gives people a broader, more appropriate frame of reference for their disciplinary work."

Another outcome of the workshop is a discussion article proposed to The Bulletin of the American Meteorological Society by Mearns, Forest, and Fowler. The article will describe the integrated approach to uncertainty used in this workshop and explain why such an approach is necessary both for seriously addressing climate change problems and educating future generations to help solve problems engendered by climate change.