Computer models may under-predict air pollution by 30 percent

The models are based in part on studies of how gases combine to form smog in clean laboratory environments. Research carried out by Jean Andino, an assistant professor of environmental engineering at the University of Florida, indicates some of these gases may react more quickly when mixed with small particles often found in urban and natural atmospheres.

Until now researchers have carried out their experiments in pristine atmospheres. The problem is that the real atmosphere contains both gaseous compounds and particulates. Andino’s research shows that some of these particulates can act as catalysts, increasing the reaction rates of some gases.

Andino’s results, which will appear in the journal Atmospheric Environment, indicate the particulates may speed up the reactions of some smog-forming gases by as much as 26 percent. Tests of simple models based on the sped-up reactions have resulted in increases as high as 30 percent in the models’ predicted levels of ozone.

“We just looked at the chemistry, so our models were very simple compared with the real models, which consider meteorology and many other factors,” she said. “Still, we think that 30 percent is a fairly significant increase, and further study is warranted.”

Andino’s experiments show that the introduction of different types of small particles sped up the reaction rates of several types of alcohols found in petrol. The particles in the tests included ammonium nitrate and ammonium sulphate aerosols, both common particles in urban areas.

Andino’s results are unlikely to affect the computer models currently in use by cities, said Roger Atkinson, director of the Air Pollution Research Center at the University of California at Riverside. “If it’s true for more alcohols, then it could be important,” he said. “I think the other line is that it probably needs to be confirmed by other labs.”

Future research will examine a much broader array of smog constituents with the hope of improving the accuracy of the air quality models. “The next step is really to look at a whole series of compounds to see if we can generate some kind of correction factor for the data used for these models,” Andino said.