Air Pollution Statistics Often "Grossly Unreliable," Study Finds

Ronald Henry, Ph.D., led a team of University of Southern California and Texas A&M University scientists who carried out an air-resources study in the Houston Ship Channel area -- home to one of the world's thickest concentrations of petrochemical industries.

Taking wind direction into account, Dr. Henry, Texas A&M statistician Clifford H. Spiegelman, Ph.D., and two co- workers attempted to correlate the substances detected by the commission's monitoring instrument with the substances reported on emission reports filed with the EPA by nearby industrial plants.

The team consistently found major discrepancies between observed and expected levels of the substances monitored. "We found that about 65 percent of the organic gases measured at the site were coming from four nearby sources," Henry says. "Yet we could identify only one of those sources, accounting for merely 5 percent of the gases, with the emissions listed on reports the refineries and petrochemical plants had filed with the EPA."

In an article published in the June 24 issue of the Proceedings of the National Academy of Sciences, the authors wrote: "We were able to demonstrate severe inconsistencies between reported emissions and major sources.... The composition and location of the sources as derived from the data [from on-site instruments] are not consistent with the reported industrial emissions.... This paper provides strong empirical evidence that regulatory agencies ... are making predictions based on inaccurate industrial emissions."

Henry, an associate professor of civil and environmental engineering at the USC School of Engineering, says the study results don't necessarily indicate malfeasance on the part of industry.

The on-site readings showed that the refineries and chemical plants sometimes overestimated their emissions of "volatile organic compounds" (smog-producing products that evaporate into the air). One large petrochemical plant reported itself as a major emitter of such chemicals, but monitoring during the June-November testing period consistently failed to find those chemicals in the air in quantities anywhere near the quantities reported.

Monitoring did, however, detect large quantities of volatile organic compounds not on emissions reports of any petrochemical factory within a wide radius of the test site.

"The industrial emissions inventory failed to agree with our observation-based approach in two ways," Henry explains. "First, the reported inventory failed to account for the emissions we did detect. Second, large-scale emissions were reported but didn't show up in our observations."

Henry believes that monitoring instruments should be more widely used in air-quality data gathering. Monitoring equipment of the type used in Houston costs about $50,000 per unit, plus manpower costs. (The employee costs might be reduced by automation.)

In Southern California, a network of approximately 20 stationary monitoring sites could give regulators a far more accurate picture of the state of the air. A small number of mobile units could be deployed to provide more detailed information in special situations -- when a large, unexplained burst of pollution was detected, for example.

"It's like when you learn arithmetic," Henry says. "You check your work. If we had such a network, and were using it, we might be able to develop more efficient and cost-effective methods of pollution control."

In addition to supplying data, the Texas Natural Resource Conservation Commission supported the study financially. Other support came from the National Science Foundation and the U. S. Environmental Protection Agency.

Also working on the study were graduate assistant Eun Sug Park of the Texas A&M department of statistics and John F. Collins, a graduate student in the USC School of Engineering's department of civil and environmental engineering.