For one thing, the mix of natural and manmade air pollution ingredients varies among regions, notes James Meagher, a lead NOAA investigator of the New England study.
"Understanding emissions should be first and foremost," he says. "We really don't know these as well as we could or should."
Without that understanding, he adds, forecasting models inevitably fall short.
In cities such as Atlanta and Nashville, for example, researchers found that on some hot days, smog levels exceeded those attributable to urban sources alone. The culprits turned out to be coal-fired power plants in the region around the cities and a hydrocarbon, isoprene, emitted from broad-leaved trees as a byproduct of photosynthesis. When winds blew power-plant emissions over the woods toward Nashville or Atlanta, each city could be enveloped in a haze of pollution.
If Southeastern cities demonstrated the critical role played by natural sources of ozone-forming ingredients in urban air pollution, Houston highlighted the need to "get it right" for human sources as well.
During the summer of 2000, a NOAA research team endured sweltering heat and long, turbulent runs in an aircraft in hopes of discovering why Houston's smog levels consistently exceeded model forecasts.
Scientists found the region's petrochemical plants were emitting vastly larger amounts of hydrocarbons, as measured by NOAA's airborne sensors, than they were reporting to air-quality officials using US EPA-approved estimation techniques.
"There's a lot of leeway in how the petrochemical industries can report their emissions," NOAA modeler Stuart McKeen says. Plants didn't really know "how much they were releasing in 'fugitive' emissions," he adds.
New, faster, and more sensitive sampling instruments hold out the promise of identifying the source of an emissions plume more precisely than in the past, Dr. McKeen says.
Not surprisingly, the discovery that emissions were higher than previously believed "changed the landscape in terms of how Houston approaches air quality," Dr. Meagher adds.