Clearing the air

Clearing the air

BY PATRICK KURP
Special to the Rice News

Young engineers frequently delve into case studies as a way of applying their classroom learning to real-world problems. Last spring, though, the students in visiting lecturer Loren Raun’s STAT 685 course experienced a case that was not just real-world, but real-time. They examined air-quality monitoring systems put into place by the city of Houston and reported their findings to representatives from the mayor’s office.

Since its inception four years ago, the ”Quantitative Environmental Decision Making” course has used a project-based format to help students learn to apply statistical methods in answering difficult scientific questions associated with environmental problems. In the past, students have worked with data associated with Environmental Protection Agency cleanup projects and have been tasked with ”Mission: Impossible”-like mandates to determine the sites’ problems and how to handle them.

This year, Raun, a visiting lecturer in statistics, introduced a new dataset for the project, one with implications close to home. As a senior environmental analyst for the city of Houston, Raun has been working with the city’s air-quality database, which contains monitoring data for toxic emissions collected hourly at hundreds of locations around Houston. Air quality has long been a concern in Houston, and the Houston City Council addressed it by advocating two programs in spring 2007: a voluntary benzene reduction plan and a nuisance ordinance requiring reductions at point sources to bring ambient concentrations of pollutants below certain risk levels.

During their analysis, Raun’s students focused on two toxic chemicals in Houston’s air: benzene and 1,3-butadiene. The students applied statistical techniques to assess the placement of monitors in the city and track discrete trends across the airshed. This enabled them to identify locations with monitoring weaknesses and provide detailed air-quality report cards that looked not just at averages, but at the upper and lower values of the totals. ”Rather than looking at just one statistic, they developed a suite of statistics having to do with health levels,” said Raun. ”Is the number of hours with high concentrations going down? Are we having more hours per year within a safe and acceptable range? All this tells us more about how we are doing than a single average does.”

Because the course attracts students from a range of disciplines, students learn not just from Raun, but from each other.

”We end up with interdisciplinary teams very much like what you would find in industry,” Raun noted. ”The stat graduate students know statistics, but they may not have ever used that knowledge in the context of a complicated environmental problem. Conversely, we have environmental engineering students and even chemists who may understand air pollution or water contamination, but don’t know much about nonparametric statistics — the specific class of tools used to work with environmental data. In that way, this course proceeds much like it would in an actual industry setting.”

Sarah Thomas, a third-year student in statistics, concurred. ”I learned about aspects of my field which would not normally be a focus in theoretical probability and statistics classes, and I enjoyed the challenge of collaborating with students and professionals from other fields,” she said. ”The exchange of information traveled both ways: I learned from others’ environmental or engineering expertise, while in turn communicating statistical knowledge and providing guidance to my team members.”

The presentations, delivered in April to representatives from the Houston Department of Health and Human Services’ Bureau of Air Quality Control and Houston Mayor Bill White, were well-received. ”I was surprised to hear about a large source of emissions being indicated by the data,” said Karl Pepple, director of environmental planning for the city of Houston. ”It seemed to point toward a specific facility on the Ship Channel that had not previously been considered. Results pointed toward other large sources as well, but we knew about those.”

He added that he was excited the class took the project on. ”This was not a straight-forward analysis process,” he said. ”They had to process the data, assess it for accuracy and analyze it. As far as the data analysis goes, the results I saw looked very good.”