“Clearly, the idea of a systems approach is to capture the full impact of climate change; right now, we are capturing snippets.”—Georges Benjamin
“Modeling to understand systems at any level is essentially the same as asking whether we understand the systems well enough to make projections, predictions, or forecasts.”—Anthony Janetos
Systems thinking is not yet widespread in modeling efforts although the climate-change community is increasingly acknowledging its importance. Workshop speakers discussed definitions of systems thinking and how it might be used to advance models of health risks posed by climate change, drawing lessons from modeling efforts in agriculture and other sectors.
Why is systems thinking important for developing models of the health risks posed by climate change? Before answering that question, Georges Benjamin, of the American Public Health Association, offered definitions of systems thinking and health system (Box 4-1).7 “Systems thinking is an approach to thinking about how things interact with one another,” he said. The systems view of how we study health, climate, or anything intermediary is that “the components are interdependent,” noted Gary Geernaert, of the US Department of Energy (DOE). The interdependent components of a health system include medical infrastructure and resources, medical and public-health staff, and mechanisms to facilitate access to care, Benjamin explained. The challenge is to determine the extent of interdependence, Geernaert said; “if you are building models, some components are more interdependent than others, and our challenge is to figure out how to reflect the interdependence appropriately and come up with robust conclusions.”
Systems thinking is particularly relevant in the context of extreme weather-related events, Benjamin said. “We build our health systems for just-in-time management for the mean; we cannot afford to build for extremes.” But robust medical infrastructure is critical for response in the aftermath of a disaster. To illustrate that point, Benjamin described the medical and public-health challenges posed by the EF5 tornado that hit Moore, Oklahoma, in 2013. The
Definitions: Systems Thinking and Health System
Georges Benjamin offered the following definitions of systems thinking and health system to workshop attendees:
Systems thinking: An approach that gives one an understanding of a system by examining the linkages and interactions between the various components that make up the entirety of the system.
Health system: The total of all the organizations and resources whose primary purpose is to improve health. A health system needs staff, funds, information, supplies, transport, communication, and overall guidance and direction. And it needs to provide services that are responsive and financially fair while treating people decently.
tornado leveled the town and heavily damaged the Moore Medical Center. Effects on the health system included infrastructure loss, relocation of medical staff away from Moore, and reduction in access to medical care. He emphasized that what happens to employees or access to medical care after such disasters is often not accounted for in models. “And we often do not include human behavior in our models,” Benjamin said. For example, the Hurricane Katrina disaster in 2005 showed that people often will not leave their pets. In addition, there is often an enormous mental-health effect on both the people who have suffered losses and the officials who are trying to manage the disaster, Benjamin said. “We talk about direct effects, but we do not talk about these tail events.”
We need to build models that address cascading system failures and secondary and tertiary effects, said Molly Brown, of the National Aeronautics and Space Administration. An analysis conducted in the wake of Superstorm Sandy showed that what most affected “human welfare and public health were all the aspects that we never thought of and that are not in the models, such as generators below sea level and unexpected cascading power outages,” Brown said.
Systems thinking can enable us to prepare for climate change, Brown suggested. In the last 5 years, a massive transformation related to food crops and infrastructure has been going on in some African countries. What citizens want is for new crops and new infrastructure to be tolerant to climate change “so that they don’t have to retrofit everything 20 years from now,” she said.
When thinking about climate-change effects, we need to include both regional change and local change and their interactions with trade, food movement, effects on water systems, and how systems will adapt, said Robert Vallario, of DOE. “Understanding such complex systems and their potential evolution is not deterministic modeling. We are exploring thresholds, tipping points, and sensitivities in the context of science-based and quantitative approaches,” he said.
The kinds of models that apply systems thinking involve what Anthony Janetos, director of Boston University’s Frederick S. Pardee Center for the Study of the Longer-Range Future, referred to as two-way models (see Box 3-2). Such models can include adaptation and institutional response and therefore come closest to identifying health outcomes. Two-way
Figure 4-1 Health outcomes from a systems perspective
Janetos told workshop attendees that even the simplest diagram that he could think of to reflect health outcomes of climate change is complex. He emphasized that no study has yet modeled the entire system and that “any component could be expanded into its own system diagram.” Source: Janetos Slide 21.
models can account for feedback loops in health outcomes, adaptation of the economy, and later consequences for both mitigation and demographics (Figure 4-1).
“The complexity of the system is partly a function of what we choose to model and what level we want to try to grapple with,” Janetos said. “One big question that we have to confront is the degree to which we want to think about health effects in a systems context. Do we want to think of them as part of an overall climate–demography system, or do we want to think of them as a sector that has its own forcings, behavioral responses, and interventions that we could model?”
“Modeling to understand systems at any level is essentially asking whether we understand the systems well enough to make projections, predictions, or forecasts,” Janetos said. Building models so that they have those capabilities is difficult. Efforts to factor in such components as human adaptation and mitigation efforts are complicated by the reality that although much evidence suggests that adaptation will occur, little clarifies when or at what level it will occur, said Sari Kovats, of the Faculty of Public Health and Policy of the London School of Hygiene and Tropical Medicine.
The models may not necessarily be best created via the conventional “top-down” approach, said Anne Grambsch, of the US Environmental Protection Agency. “We tend to think of systems as things that we design from the top down, but systems also can emerge from data and observations,” she said. She added that systems thinking can reveal values that may
emerge from comparing tradeoffs. That raises the question of how to collect and structure such data so that they can tell us something useful, she said; “It is not so much how we use systems to identify cascading failures as it is how cascading failures tell us how distantly things are connected.”
Gregory Glass, of the University of Florida, noted that incorporating systems thinking into models of health risks posed by climate change will ideally not only improve our understanding of health outcomes but help us to identify “where our weaknesses and gaps are.”
Context matters, Brown emphasized. “The world has a distribution of the haves and the have nots, the resilient and the less resilient, and the least vulnerable and the most vulnerable.” More modeling outcomes need to be put into that context “so that we can ensure that safety nets are appropriate,” she said.
Joshua Elliott, of the University of Chicago’s Center for Robust Decision Making on Climate and Energy Policy (RDCEP), cautioned that developing a “supermodel” of the whole Earth may be unproductive. He suggested that a better approach may be to think of a sector, such as agriculture, as a “system within a system” and try to understand it and its interactions with other systems.
One way to find out how to create pathways to desirable end points is the approach used in economic models, Brown said. If you define the end point that you want to reach, you can run simulations to find ways to reach it. An example shared by Geernaert was “no blackouts by 2050”. He urged attendees to think of examples that were more health-focused in an appropriate timeframe. To identify those kinds of end points, it is important to get people to think about the different outcomes that they are willing to accept, Glass said.
Janetos suggested that the health goals in the Millennium Development Goals8 could serve as end points in models for assessing the health effects of climate change. He also noted that “forward” modeling may not always be needed to understand how to target interventions. “We might do well to turn the problem around, ask about the degree to which we understand parts of the system that determine vulnerability and sensitivity, and then model the degree to which an intervention could have an effect.” Only then, he suggested, does it make sense to ask whether a change in climate would make the intervention beneficial or not.
Kovats suggested there is value in thinking about health as both a result and a driver in the context of poverty eradication and sustainable development. She and Stéphane Hallegatte, of the World Bank, suggested that it will be important to make sure that investments take into consideration adaptation. To improve disaster-risk mapping, more attention is needed on the social side, Kovats added; for example, projections of poverty and of where people will live are critical.
Elliott discussed the results of a recent model-intercomparison project to illustrate the complexity involved in comparing results from one or two sectors. The study included data from more than 40 models. Groups in four sectors collaborated to compare the results of their models with regard to climate effects, including health. The comparison suggested a strong
8Global, time-bound, quantified targets for addressing extreme poverty in its many dimensions—income, hunger, disease, lack of adequate shelter, and exclusion—while promoting sex equality, education, and environmental sustainability. For more information, visit http://www.un.org/millenniumgoals/.
likelihood that a tropical region’s ability to produce maize and wheat would be adversely affected by rising CO2 concentrations, Elliott said; there is less model agreement with respect to rice and soy, which may be favorably affected.
The project included an ensemble of 11 global hydrologic models run with the same set of climate scenarios as the crop models. The researchers combined the results of the global crop and hydrologic models to analyze the potential of irrigation as both an adaptation mechanism and a measure of climate change. They found that water limitations caused by reduced runoff and hydrologic limits resulted in a decrease in food production of 8–24% compared with today’s agricultural production even when increases in plant productivity associated with increased atmospheric concentrations of CO2 were included. The decreases were much larger in scenarios that did not include increased plant productivity in the presence of increased CO2.
When the modelers included freshwater limitations implied by the hydrologic models and the reduction in runoff, their simulation suggested that 20–60 million hectares of irrigated cropland would have to be converted to being only rain-fed by the end of the century. That implies a further decrease in food production, Elliott said.
The US Centers for Disease Control and Prevention (CDC) Climate and Health Program created a model to help cities and states to prepare for health effects of climate change. Building Resilience Against Climate Effects (BRACE) uses the principles of adaptive management, which encompasses loop learning and feeding what is learned back into the system, said George Luber, CDC’s associate director for climate-change programs. “We forecast the climate effects for a region and couple them with the existing vulnerabilities of the location,” he said. On the basis of population dynamics and projected increases or decreases in heat, precipitation, drought, and so on, the model enables researchers to assess vulnerabilities.
The National Science Foundation is funding a project to investigate how an approach known as agent-based modeling can help researchers to understand how the experience of living through multiple disaster scenarios involving multiple system failures may affect human behavior on both the institutional and individual levels, said Benjamin Zaitchik, of Johns Hopkins University. The models are designed to be capable of investigating both engineering interventions to harden infrastructure and effects on human health. They are intended to investigate such questions as how many intensive heat waves are required to prompt institutions to change urban infrastructure to reduce the effects of the heat waves.
A Web site run by RDCEP includes access to programs that allow anyone to run simulations that show the effects of global climate change, Elliott said. The WebDice model runs simulations that project how such actions as carbon taxes, climate treaties, and “optimized policies” may affect global warming. It is based on a widely used integrated-assessment model of the economics of climate change called DICE, which was invented at Yale University in 2007. “We have added many features and assumptions and have enabled people to tweak assumptions to see what kinds of effects they have,” Elliott explained.
This page intentionally left blank.