July 18, 2014
As a college student studying anthropology, Michael Barton wanted to understand how people as a society impact the environment, and how the environment impacts society. He imagined some type of mathematical and visual technology was needed to study this interaction. The only problem was that this type of technology didn’t exist yet.
Today, technology has finally caught up to Barton’s ideas. Now a professor in the School of Human Evolution and Social Change at Arizona State University, Barton uses computer-run computational models to study the questions that he dreamed up at the beginning of his career. There’s even a name for the field of study that initially intrigued him: coupled natural-human systems.
Download Full Image
“I tell people that what I’m doing now is what I always wanted to do but it wasn’t possible,” says Barton, who is also director of ASU’s Center for Social Dynamics and Complexity. “Now I get to do what I wanted to do as a sophomore.”
Computational models help scientists make predictions about complex questions that would be hard to test otherwise. In addition, models can be set to simulate very long periods of time, such as thousands of years, which would be impossible to test in the real world. Models are also useful because they can test potential solutions to problems before we invest a lot of time and money to implement changes.
“Models are a really important complement to collecting empirical data, especially for socio-ecological systems where there are many, many complex interactions,” Barton says. “These are things that aren’t easily represented by simple equations and are frankly beyond the ability of our brains to comprehend.”
Barton uses computational algorithms to represent different human and natural components of complex systems. For example, in one project, he studies the profound effect that agriculture has on the landscape, and how that in turn influences human behavior.
Small farms, big impact
Farming is a prime example of a coupled natural-human relationship. Archaeological evidence indicates that farming moved into the western Mediterranean between seven and eight thousand years ago, making the region one of the oldest places in the world where coupled natural-human landscapes emerged. Using archaeological and paleoecological data from this region, Barton’s team has created a modeling laboratory that simulates farming conditions thousands of years ago. They use the laboratory to conduct digital experiments to understand the reciprocal interactions of farmers and the landscape over a period of hundreds of years.
The study, supported by the Coupled Natural and Human Systems Program at the National Science Foundation, will shed light on the processes that created the landscape we see today in the western Mediterranean and throughout the world. It will also help us understand the long-term effects of small-scale farming.
Barton notes that around 90 percent of the world’s farms today are still less than one hectare (10,000 square meters) in size, and they produce half of the world’s food. The more we understand the interactions between small-scale farmers and the landscape, the more we may understand how farming decisions today will affect the landscape and farming potential in the future.
The modeling laboratory uses variables such as crop type, livestock type, soil quality, landscape vegetation and the size of the land being cultivated by each farmer. By adjusting variables in the model and running simulations, scientists can observe different scenarios and outcomes.
The farming model simulates outcomes of the decisions of farmers, their impact on the land and the resulting influence on farmers’ decisions year after year. It can be set to run for a given time and then present results such as population size, soil quality, land use across the region and distribution of vegetation. These results are then compared to archaeological and paleoecological data to see which scenarios might have produced western Mediterranean landscapes.
While he was conducting fieldwork in Spain for this project, Barton had an important “eureka” moment about coupled natural-human systems. He was standing on a hill overlooking a valley in eastern Spain. He noted that the species of trees and grasses, and even the animals that he observed were the same species of trees, grasses and animals that were present in the area thousands of years ago.
The difference was that today’s trees and grasses appeared in different densities and distributions across the landscape. He realized then that humans don’t simply replace the natural landscape with a human one. Instead, people modify and rearrange components of the landscape to create an environment that is both human and natural.
“I was looking out thinking, ‘You know, these were all the same things that were here before, we just shifted them around, encouraged some and moved others to other places.’” Barton recalls. “So I think that was an important epiphany moment for me.”
Storm warnings and social media
Sometimes, of course, nature is the one moving things around, as in the case of storms and natural disasters. Fortunately, today we have advanced technologies that help us communicate about these risks. But are they as effective as they could be?
In collaboration with the National Center for Atmospheric Research (NCAR), Barton and his colleagues are working to understand – and hopefully improve – how information about natural hazards and disasters gets transmitted in the digital age. Currently, for severe weather such as a tornado or hurricane, the National Weather Service issues a bulletin. The bulletins are often very technical and difficult for a layperson to understand.
Media outlets such as television stations, radio stations and websites like weather.com issue forecasts based on their interpretation of the severe weather bulletin. Local governments also receive, interpret and share severe storm warnings through various channels. These forecasts, in turn, are shared among friends and acquaintances on social media platforms like Facebook and Twitter.
The result of this system of sharing information is numerous forecasts of severe weather, each with a different interpretation of the original weather bulletin. This can be confusing to people faced with the decision of whether to evacuate their homes. In some situations, individuals may delay necessary evacuation, or they may choose to evacuate when they do not need to. The former can lead to the loss of lives, while the latter can lead to unnecessary expenses.
To improve this type of communication, Barton and his colleagues are building computational models that simulate the information networks used to share severe storm warnings. The model simulates a series of information outlets that can influence a person’s decisions. These outlets include social media, mass media and information from local governments and from other citizens, such as friends and neighbors.
“We’re trying to create a complex model of many, many agents that can represent a real world full of these organizations and citizens and households,” Barton says. “Then we create different kinds of hurricane and storm surge scenarios and the types of forecasts to go along with them.”
In the model, people’s decisions about what action to take when faced with a severe storm warning are based on the information they receive, past experiences and how risk-averse they are. To build this model, the researchers are studying real-world outcomes of actual severe weather warnings along with the associated Twitter data, which is being shared by the University of Colorado at Boulder. This will result in a model that mirrors the way information actually travels.
By using a model to test different scenarios, new ways of sharing information can be tested and observed before they are implemented, saving time and money.
“It’s unusual but satisfying to be applying this knowledge [about models] to situations where real lives and property are at stake,” says Barton. “We hope that we save lives.”
Linking models to mirror the real world
Part of the problem with studying the relationships between people and the environment is that earth systems models – such as those that model climate, vegetation, carbon cycles, etc. – often don’t include human behavior as a variable. Composite models that simulate the interaction between earth systems and socioeconomic systems don’t exist, even though the systems directly affect one another in the real world.
The problem with this is that the actions of humans are not static. We react to changes in our environment and atmosphere. These actions create even more changes to the landscape and climate, which in turn affect our actions.
“We may not know exactly what the climate is going to be like 100 years from now,” Barton says. “But we have some idea about where it’s going to go if people do different things. So what surprised me is that these very large, complex models don’t include people in them. But, in fact, as climate changes it changes the likelihood that you can farm in different places, or have animals in different places, or whether you want to live in a city or not, or whether you’ll get fish from the ocean. And that will change what people do, which will change how they affect the landscape and the climate, which will change what people do. So all these feedbacks exist, but they aren’t modeled.”
To address this problem, Barton is heading an ASU team that is contributing to a large, multi-university collaboration led by NCAR. The collaborators are building a computational tool kit that links earth systems models with global-scale socioeconomic models. This tool kit can help people better evaluate not just the environmental, but the social impacts of actions.
“My hope is that this project will give us more informed decisions and a better idea about the longer-term consequences of what we do,” says Barton. “It’s not a matter of ‘this predicts what the future is going to be.’ The future can be many things. But it would be nice to know that if we do ‘x’ then there’s a good chance it would turn out this way, or there’s a good chance it will turn out that way. And then we can decide which of those outcomes we want.”
Studying coupled human-natural systems requires multiple academic disciplines, such as archaeology, economics and earth science. Questions that span different fields always interested Barton, but he says it used to be considered “weird” to be interdisciplinary and not focus on a single field of study.
“I see a lot of hope for the work we do,” he says. “And ASU right now is a good place to do this with its emphasis on interdisciplinary science. I’ve been very interdisciplinary all my life, and it wasn’t so widely accepted 15 to 20 years ago.”
Open source modeling
In addition to developing models, Barton is helping to share them among the scientific community through an online web portal called CoMSES Net (Network for Computational Modeling in Social and Ecological Sciences), supported by the National Science Foundation.
A central aspect of CoMSES Net is a computational model library, where scientists and researchers in the social and life sciences can publish and share the models they have developed and give feedback on models submitted by others. The site also features an active job board, education resources and discussion forums. It has become an important place for researchers in this niche to connect.
Each model submitted to the library receives a unique identification number in the same way that published scholarly articles do. Because of this, models can be used by other researchers and cited in scholarly articles. CoMSES Net is supporting increased transparency in scientific computation by making computational models for the social and life sciences available in this way. The portal was launched as a pilot in 2006 and today has over 220 models in the model library, with 200 full members and 1,000 affiliate members using the site.
Barton says, “We’re trying to increase communication among people who are doing this kind of research and to come up with ways to make computational models embedded in normal science in the 21st century.”
Written by Kelsey Wharton, Office of Knowledge Enterprise Development