6/30/2014 Mike Koon
Written by Mike Koon
By Mike Koon, Engineering Communications Office
Engineers planning the layout and design of the island of Manhattan in the mid-19th century saw a need for a massive natural retreat in the middle of the metropolitan area and developed Central Park. It seemed the idea of green space in the city made sense for several reasons from providing an attractive setting for carriage rides to improving public health. While few argue the benefit trees and similar green spaces can have on cities, the desire for them has largely been based on theory and personal preference.
That has all changed in the 21st century with the advent of supercomputers where the ability to store and process data is lending proof to the positive impacts of green infrastructure, and cities across the globe are beginning to understand its certifiable and quantitative benefits from an ecological and economic perspective.
“The decisions we’re making as engineers have a huge impact on people, and we need to be aware of that and be able to design in a way that will benefit people more than it does today,” Minsker said.
Her work is another example of how Big Data is changing the landscape of nearly every industry. In addition to her project on green infrastructure, she is collaborating with a team at the University of Texas-Austin on a real-time water modeling and decision framework, which helps state agencies allocate water supply during drought conditions.
As part of that Microsoft Research funded project, Minsker is helping to create a Web application that will allow regulators in Texas to analyze current conditions and make educated predictions of the near future to determine water allocations.
“In the past we took theory and then tried to prove theory,” Minsker explained. “Now we’re starting from data and trying to see how the theory explains what we’re seeing, and if it doesn’t, determining what theory is needed to explain the data. It’s research driven by data rather than the reverse, which is more challenging, but more interesting,”
The GI projects are funded by the National Science Foundation and the National Socio-Environmental Synthesis Center and are focused on five cities in the United States – Baltimore, Chicago, Portland (Oregon), Phoenix, and Durham, North Carolina -- and two in Australia -- Melbourne and Adelaide. The cities were chosen because they already have a lot of data on the topic, but the challenge has been pulling the data together where each city has differences in collecting agencies, storage sites, and formats of the data.
“We are trying to create data agreements with each city and have the data archived in a central location,” Minsker said. The team is collaborating with the National Center for Supercomputing Applications (NCSA) at the University of Illinois and the Renaissance Computing Institute at the University of North Carolina Chapel Hill to incorporate the data into a user-friendly data system that can be used by other researchers interested in interdisciplinary urban challenges.
The group will use these data to create models and assess long-term benefits of GI designs. For instance, how much will green infrastructure benefit sewer overflow as compared to installing more pipes? Specifically, the tasks include: “1) create integrated models to predict hydrologic, human, and ecosystem impacts of green infrastructure designs from site to catchment scales, 2) develop methods for crowd-sourcing green infrastructure design, and 3) implement modeling and crowd-sourced design methods in a cyberinfrastructure framework.”
“In making the case for showing the tremendous benefits of GI, we want to put a dollar value on it because that’s often how these decisions are made,” Minsker said. “Droughts are going to come. Floods are going to come. If we make these changes, will we have fewer problems because of them in the future? We can look at these types of events in the past and see what would have happened had we done something different. Long term and short term impacts haven’t been linked very well in the past. These kinds of approaches allow us to do that much better.”
“In an age of climate change where there have been more extremes in the weather, this study will better plan for green infrastructure in ways that help conserve water and better deal with heat waves, droughts, or floods,” Minsker said. “We believe that the added green space can help reverse the pattern of weather because when we went the other way and paved everything, it got very hot.”
However, as she saw in her most recent trip to Australia, these types of decisions often are driven by organizational and political factors. That’s why the working group includes social scientists who understand how these topics play out and can help better determine how to foster change in that realm.
“Green infrastructure has effects on storm water, on health, and on climate,” Minsker said. “Having storm water engineers work with urban planners, social scientists, and landscape architects bridges all kinds of organizational barriers.”
The work is a collaboration with Illinois researchers Bill Sullivan, Professor in the Department of Landscape Architecture; Art Schmidt, Research Assistant Professor in the Department of Civil and Environmental Engineering; and Jong Lee, Senior Research Scientist at NCSA.
“We’re using machine learning to train a model to be able to predict how well people will like certain landscape designs,” Minsker explained. “It has proven to be about 80 percent accurate so far.”
Minsker is not only using her expertise and bevy of researchers, but also her proximity to the National Center for Super Computing Applications to become a leader in the field. The crowd-sourced green infrastructure design study is just one of the cases that is driving the development of a $10 million Data Infrastructure Building Blocks (DIBBS) initiative. led by Kenton McHenry at NCSA, through the National Science Foundation.
“My partnership with NCSA completely changed my career direction,” Minsker said. “I can be right here at the leading edge of computation, trying out things that my colleagues haven’t even heard of yet. I started learning about data mining and incorporating that into my research about 10 years ago. Now everything I do is machine learning of one kind or another, whether it’s working with data streams or modeling. There are projects that come my way that wouldn’t otherwise because of my connection with NCSA.
“I have always been driven by wanting to make a difference in the world,” Minsker concluded. “That’s why these real world problems are so appealing. I want to do something that is going to make a difference out there and not just have my research sit on a shelf somewhere and never be used. This is a very exciting time for me to be able to do this kind of work.”