Day in and day out, Bob, Alfred and Earl commute to work in Hamburg’s Hoheluft district, bring their children to the daycare, or go shopping. But they don’t actually exist. They are “agents,” who move through a virtual Hamburg in a computer model. My team and I use this model to explore how agents’ typical living situations and attitudes influence e.g. their choice of commute: Bob doesn’t have much time, Earl doesn’t have much money, and Alfred is very environmentally aware. In addition, factors like the weather, gasoline and bus prices shape their decisions.
Thanks to the model, we can also estimate how the agents are affected by “environmental stresses,” i.e., by environmental factors that are potentially harmful to their health, like heat, noise, air pollution or the impacts of climate change. In cities characterized by a high density of people, buildings and traffic, these stresses are especially dangerous: around the globe, air pollution alone is responsible for an estimated two million deaths every year. Further, because buildings store heat, extreme heat waves — which are likely to become more frequent in the future — can make cities far warmer than the surrounding countryside.
More than half of the global populace already live in cities, and that number is rising. For urban planners and politicians, making cities healthy and worth living in is a key priority. And the insights that we glean from agent-based modeling can help them achieve that goal. On the computer we can experiment to see how annoying construction sites, rising costs for public transportation, or additional bike paths affect the choices of individual citizens – and what that means for their health, and for the health of the city as a whole.
The method was made possible by the rise of computers. I first took advantage of it for my doctoral dissertation in 1989, where I used a self-programmed model to simulate the outcomes of various scenarios in the East-West conflict. Given the two main possibilities — escalation and de-escalation — my model predicted that growing trust between the two superpowers would likely produce a chaotic transitional period. And, just a few weeks after the simulation, the cold war ended with the collapse of the Eastern Bloc. Even I was amazed to see how quickly the reality caught up with my forecast.
Today, agent-based modeling has become indispensable for researchers. If our goal is to understand how a specific group will behave in a given setting, the method can offer valuable insights. It’s also well suited to researching the effects of urban environmental stresses — as the test with the representative agents Bob, Alfred and Earl shows. The results produced to date have laid the groundwork for expanding the model using real-world behavioral data. We could then e.g. simulate the consequences of extreme weather events, to determine whether or not urban evacuation and supply routes actually work as they should in a crisis. Another possibility would be to apply the method to other metropolitan areas – after all, cities around the world have to adapt to climate change.
This content was first published as a guest article in the newspaper Hamburger Abendblatt in June 2018.
Jürgen Scheffran is a Professor of Integrative Geography and a member of Universität Hamburg’s Center for Earth System Research and Sustainability (CEN), where he leads the Climate Change and Security (CLISEC) research group and presents joint modeling outcomes from the URBMOD project.
Liang Emlyn Yang, Peter Hoffmann, Jürgen Scheffran, Sven Rühe, Jana Fischereit, Ingenuin Gasser (2018) An Agent-Based Modeling Framework for Simulating Human Exposure to Environmental Stresses in Urban Areas. Urban Science 2, 36. Read Online.
Todd BenDor and Jürgen Scheffran: “Agent-based Modeling of Environmental Conflict and Cooperation” will be released in September 2018 by Taylor & Francis.