A New Idea in Air Quality Forecasting
 

Air quality scientist Jim Szykman begins his story with a tone that suggests “You’re not going to believe this, but…” He describes how in late summer of 2002, an area of high atmospheric pressure settled in over the industrial Midwest south of Lake Michigan. The soot, smoke, and gases from regional power plants that consistently have some of the country’s highest emission rates began to pile up in the hot, humid, stagnant air. The Environmental Protection Agency’s air quality index values began to exceed 100—a level considered unhealthy for people with lung or heart conditions.

By itself that kind of aerosol pollution event isn’t unusual for that area in the summer. In a normal week one of the regular weather fronts that move from west to east across the United States would have passed through, breaking up the high-pressure system and sweeping the dirty air eastward toward the Atlantic Ocean. As it turned out, however, September 8-14 was not a normal week.

  Page 2 Photograph of factory smokestacks emitting smoke

Power plants, factories, and automobiles create air pollution that can spread far from its source. The U.S. Environmental Protection Agency (EPA) monitors air quality at ground stations across the country, most of which are located in urban areas. To predict where and when a local pollution problem will became a regional one, air quality forecasters need observations over a wider area than ground stations can provide. EPA and NASA scientists are combining satellite measurements of air pollution with data collected on the ground to improve forecasts of air quality. (Image copyright Hemera)

  Satellite image of pollution over the Midwest
 

A weather front did arrive from the northwest. But where it might normally have just tracked its way across the Mid-Atlantic and eastern United States, the front instead found its normal exit blocked by Hurricane Gustav, churning off the coast of North Carolina. “The dirty air got squeezed between the oncoming weather front and Gustav to the east,” says Szykman, “and flowed southward down the course of the Mississippi River toward the Gulf of Mexico.”

Just as the pollution-filled air was trying to clear the Gulf Coast and make its way out to sea, it hit yet another roadblock: Tropical Storm Hanna plowing northward through the Gulf. In the face of Hanna, the Midwest air mass hung over Texas like a confused funhouse visitor searching for an exit. Pollution concentrations remained at unhealthy levels for several days.

 

Humid, hot, stagnant air combined with industrial emissions in the upper Midwest to create a thick blanket of haze over the area in September 2002. Various weather sytems pushed the pollution around the eastern half of the country for a week, triggering widespread air quality alerts in places as far away as Texas and South Carolina. This image from the Moderate Resolution Imaging Spectroradiometer (MODIS) shows the polluted air like a gray veil over the region southwest of Chicago on September 9, 2002. (NASA Image by Jesse Allen based on data provided by the MODIS science team)

  Satellite Image of Smog over the Southeast United States
 

While Hanna blocked the southward evacuation route, Gustav began to head for the northern Atlantic, loosening its grip on the Southeast. Some of the dirty air snaked across the Deep South and out to the Atlantic, but all of the pollution didn’t get a chance to escape before Hanna came ashore like a snowplow, pushing what was left back toward the north. After a week on the road, the well-traveled and only partially diluted wave of polluted Midwest air was pretty much back where it started!

Aside from its being an interesting example of how complex our weather is, what’s so important about this story? To Szykman, who works for the Environmental Protection Agency (EPA), what’s important is that most local air quality forecasters in the locations where the pollution ended up didn’t even know it was coming. That’s a situation Szykman is determined to change. He thinks he can give local air quality forecasters better advance warning of significant regional pollution events by fusing the EPA’s local, ground-based air quality measurements with NASA’s satellite perspective of pollution. By combining this information with weather forecasts from the National Oceanic and Atmospheric Administration’s (NOAA) National Weather Service, Szykman and his colleagues will give forecasters a new view of how pollution moves over a region, like meteorologists track weather systems moving across the country.

 

Two days later, a weather front arrived from the West. With Hurricane Gustav off North Carolina blocking the eastward flow of air to the Atlantic, the air pollution spread southward to Texas. There, Tropical Storm Hanna prevented it from moving out over the Gulf of Mexico. As Hanna came ashore and Gustav moved northward, some of the polluted air escaped to the Atlantic across the Deep South. This image from September 11, 2002, shows the haze stretching from Louisiana (left) to South Carolina (top right). (NASA Image by Jesse Allen based on data provided by the MODIS science team)

 

EPA and NASA scientists have an IDEA

 

Combining the assets of NASA and the EPA with NOAA’s weather information is at the heart of a new NASA Earth Science Applications project called IDEA: Infusing Satellite Data into Environmental Air Quality Applications. The goal of IDEA is to improve the EPA’s decision-making tools with NASA satellite observations for better air quality forecasts. The project was conceived and planned by Szykman, along with Jack Fishman and Doreen Neil, researchers at NASA’s Langley Research Center. Szykman stumbled into research on air quality from a graduate school background in water resources. While his wife was getting a Ph.D. in business at the University of North Carolina, Szykman’s scientific expertise was broad enough to get him a position at the EPA’s office in Research Triangle Park, just outside Raleigh. The office devotes most of its activities to air quality research and national programs.

  page 1Page 3
  EPA air quality index ozone and particulate pollution maps
 

“In part, my wife Lisa also prompted my work with NASA,” says Szykman. When she accepted a faculty position at the College of William and Mary, the move brought them within 20 miles of NASA’s Langley Research Center in Virginia. “Jack had a long history of working with the EPA office in Research Triangle Park,” says Szykman, “and the timing seemed to be right to increase the collaboration between NASA and EPA offices there. While there are many others involved with this work, Jack, Doreen, and I had an ‘IDEA’.”

First, the team wanted to demonstrate how NASA’s satellite observations could improve computer models—the decision-making tools—used by EPA forecasters to predict pollution caused by aerosols. In the future, they will expand their work into improving computer predictions of other major air pollutants, like ozone or carbon monoxide.

What’s in the Air We Share

According to Szykman, the collaboration is just a step ahead of a shift in the nation’s data needs for air quality management. Agencies are beginning to realize that air pollution must be tackled on a regional basis: Denver’s “brown cloud” winds up in Rocky Mountain National Park, while the Great Smoky Mountains are hidden beneath the haze of pollution from cities and power plants as far away as Chicago, Detroit, Indianapolis, and St. Louis. In 1997, when the EPA began to enforce stricter rules for controlling ground-level ozone pollution (the cornerstone of smog), 11 states in the Northeast appealed to the EPA for waivers on the grounds that their failure to curb ozone pollution to acceptable levels was because of pollution wafting in from other states.

 

The EPA routinely measures ground-level ozone (left) and particle pollution with a diameter of 2.5 micrometers or less (right), the two most significant contributors to regional haze. Measurement locations (colored circles) for air pollutants are widely scattered, and only provide information about current conditions in a localized area (gray areas on maps represent places with no data). To solve regional air quality problems, EPA scientists need to know where pollution comes from and where it ends up. (Maps courtesy EPA AirNow)

  Comparison of clear and hazy conditions in Rocky Mountain National Park
 

While the regions the EPA needs to monitor grow larger, the size of the particles they need to track is growing smaller. In 1997, under the Clean Air Act, the EPA issued a new set of National Ambient Air Quality Standards for fine particulate matter, or PM2.5, which are particles with a diameter less than 2.5 microns. These particles are so tiny—a micron is only one millionth of a meter—that an individual particle can only be seen with a microscope. Spewed by the trillions out of power plants, automobile exhaust pipes, fireplaces, and forest fires, these particles mix with each other and with larger particles to create a hazy blanket that often spreads far from the original source. The EPA calls pollution caused by aerosols “particle pollution.”

 

Air pollution often spreads far from its source. These photographs of Rocky Mountain National Park in Colorado show the difference between a clear day (left) and an extremely hazy one (right). The haze often blows into the park from nearby Denver. (Photographs courtesy Interagency Monitoring of Protected Visual Environments)

  Micrograph of 2.5 micro meter particles

The PM2.5 particles’ tiny size makes them particularly treacherous to public health because the smaller the size, the easier it is for the particles to slip through the lung’s defenses, causing or worsening breathing problems in many people. According to an EPA report called “Latest Findings on National Air Quality: 2002 Status and Trends,” about 59 million people in the United States—that’s just over 20 percent of us—live in counties where the PM2.5 concentrations regularly exceed the safe limit set by the EPA.

In December 2002, the EPA administrator signed a proposed suite of actions to reduce current levels of power plant emissions. One of the actions is the Interstate Air Quality Rule, which targets sulfur dioxide and nitric oxide emissions in 29 Eastern states and Washington, D.C. Those two types of emissions contribute significantly to the fine particle and ozone pollution that creates regional haze, such as the Midwest event Szykman described.

The Future of Particle Pollution Forecasting

The need to monitor tiny particles over a wide region is a major change from previous air quality standards, according to Szykman. The EPA’s existing ground-based monitoring networks within North America are located primarily in urban areas. They provide detailed measurements on the kinds and amounts of pollution in the air. “These networks already cost the agency’s Office of Air Quality Planning and Standards about $200 million a year to operate,” says Szykman. “Adapting these existing networks for tracking regional haze would require substantial additions or reconfigurations of existing networks.”

When scientists make air quality forecasts these days, they focus on atmospheric processes in the lower atmosphere (near Earth’s surface), urban-scale models of pollution transport, and ground-based measurement networks. This approach works for predicting whether the pollution present in a city today will lift out by tomorrow, or whether joggers should consider taking their daily run on an indoor treadmill, but it wouldn’t have predicted, for example, that smoke from devastating forest fires in southern California in late October 2003 would wind up on the doorsteps of Maine in early November.

That is where NASA comes in. Not only do NASA scientists have experience in creating and running regional-and global-scale models of how the atmosphere moves aerosols around, but they also have satellites that observe aerosols on regional and even global scales. While EPA’s measurements are more detailed but localized, NASA’s satellite measurements are less specific but comprehensive. According to Szykman, fusing these complementary approaches is the future of particle pollution forecasting. Using this technique for other major pollutants, like ozone or carbon monoxide, could revolutionize how forecasters determine the overall air quality for a region.

 

This image shows a magnified view of aerosol particles collected in the industrial city of Port Talbot, England. Many of the particles measure roughly 2.5 microns across, small enough to easily enter and damage human lungs. (Micrograph adapted from Sixth Annual UK Review Meeting on Outdoor and Indoor Air Pollution Research 15th–16th April 2002 (Web Report W12), Leicester, UK, MRC Institute for Environment and Health)

 

Data Fusion

 

The EPA’s ground-based, air quality sensors collect actual particles; for example, they capture particles on air filters. They record measurements of air quality in terms of the amount (mass) of pollution per cubic meter of air, and the EPA sets ranges of pollution based on health effects. Satellites see pollution differently. A satellite will produce an image much like a digital photograph, showing an area of grayish haze.

To turn the haze they see in satellite pictures into real numbers, scientists make maps of something they call “aerosol optical depth,” which is a way to estimate how much aerosol is in the air by observing how much sunlight passes through. An aerosol optical depth of zero means few aerosols are present, and most of the light is getting through the atmosphere. Progressively higher values mean aerosols are increasing and less light is getting through. The IDEA project is using aerosol optical depth measurements from NASA’s Moderate Resolution Imaging Spectroradiometer on the Terra satellite.

  page 2
  Map of ground-based and satellite aerosol measurements

Szykman and the other scientists involved with the particle-pollution forecasting part of the IDEA project want to track and make forecasts for all aerosol events that reach a pollution concentration of 40.5 to 60.5 micrograms per cubic meter of air, referred to as “Code Yellow: Unhealthy for Sensitive Groups.” Szykman and his colleagues have found that this amount of pollution is equivalent to an aerosol optical depth of about 0.6. When they see a value close to 0.6 on a satellite image, they begin to keep an eye on it for a day or two, comparing what they see from satellite to the ground-based network measurements. If both sources show a trend toward increasing aerosol pollution, the scientists begin combining the aerosol observations with wind speed and air mass movement models using data from the National Weather Service to produce 48-hour forecasts of where the pollution will go. The end products are available to local forecasters via the Web and the Cooperative Institute for Meteorological Satellite Studies (CIMSS) at the University of Wisconsin.

 

EPA and NASA scientists fuse ground-based measurements with satellite data, meteorological measurements, and computer models to get a big-picture view of air pollution. This data fusion produces a map with several layers of information. The background is satellite data from MODIS, with aerosol optical depth in color (aerosol concentrations increasing from blue to red) and cloud thickness in shades of gray (thickness increasing from gray to white). Wind speed and direction are shown with white arrows, and ground-based measurements of air quality appear as colored dots (air pollution increasing from green to red). Fire locations from the GOES-12 satellite are marked with pink or violet diamonds. The image shows conditions on July 19, 2004. The large area of dense aerosols across the Midwest was caused by widespread fires in Alaska and Canada (see animation). (Map courtesy NASA/EPA/NOAA/CIMSS IDEA program)

  Map of predicted aerosol trajectories

This approach to particle-pollution forecasting has been through a couple of test runs this year, and Szykman is optimistic. “We didn’t have the best luck as far as the timing of our pilot study and the timing of aerosol events, but we got at least one good test during the period,” he says. He expects the particle-pollution forecasting part of the IDEA project to be fully operational by the summer of 2004, a prospect that is both personally and professionally rewarding.

“Each year the federal government spends a significant amount of public funds on scientific research, and it is an obligation of the different federal agencies to work together on the use of such research for the public benefit. I believe IDEA is a good demonstration of that,” he says. In addition, he says, “the other people involved with this project at both the EPA and NASA are passionate about their work. It’s nice to work with a group of people who enjoy what they do, who like coming to work every day, and who feel they have something to contribute to either science or society in some small, but meaningful way.”

  • References:
  • Szykman, James; White, John; Pierce, Brad; Al-Saadi, Jassim; Neil, Doreen; Kittaka, Chieko; Chu, Allen; Remer, Lorraine; Gumley, Liam, and Prins, Elaine. Utilizing MODIS satellite observations in near-real-time to improve AIRNow next day forecast of fine particulate matter, PM2.5. Conference on Atmospheric Chemistry, 6th: Air Quality in Megacities, Seattle, WA, 11–15 January 2004 (preprints). Boston, MA, American Meteorological Society, 2004, Paper 1.2. Reprint #3630.
  • Kittaka, Chieko; Szykman, James; Pierce, Brad; Al-Saadi, Jassim; Neil, Doreen; Chu, Allen; Remer, Lorraine; Prins, Elaine, and Holdzkom, John. Utilizing MODIS satellite observations to monitor and analyze fine particulate matter, PM2.5, transport event. Conference on Atmospheric Chemistry, 6th: Air Quality in Megacities, Seattle, WA, 11-15 January 2004 (preprints). Boston, MA, American Meteorological Society, 2004, Paper 1.3. Reprint #3631.
 

The combined satellite and ground observations are integrated into a computer model to predict the future location of pollution. This image shows the trajectories of aerosols (thick pink and white lines) from July 19, when the data were acquired (see top image), to July 21. Colored areas are measured aerosols, thick lines show trajectories (darker pink indicates air flow toward the surface, white indicates upward air flow), and the ends of the lines indicate the predicted position on the 21st. View the trajectory animation to see the predicted track of the aerosols. (Map courtesy NASA/EPA/NOAA/CIMSS IDEA program)

 

Composite of Surface and Satellite Measurements

 
Scales for composite animation
 

EPA and NASA scientists fuse ground-based measurements with satellite data, meteorological measurements, and computer models to get a big-picture view of air pollution. This data fusion produces a map with several layers of information. The background is satellite data from MODIS, with aerosol optical depth in color (aerosol concentrations increasing from blue to red) and cloud thickness in shades of gray (thickness increasing from gray to white). Wind speed and direction are shown with white arrows, and ground-based measurements of air quality appear as colored dots (air pollution increasing from green to red). Fire locations from the GOES-12 satellite are marked with pink or violet diamonds. The image shows conditions on July 19, 2004. The large area of dense aerosols across the Midwest was caused by widespread fires in Alaska and Canada. (Animation courtesy NASA/EPA/NOAA/CIMSS IDEA program)

 

Predicted Aerosol Trajectories

 
Scales for trajectory animation
 

The combined satellite and ground observations are integrated into a computer model to predict the future location of pollution. This image shows the trajectories of aerosols (thick pink and white lines) from July 19, when the data were acquired (see top image), to July 21. Colored areas are measured aerosols, thick lines show trajectories (darker pink indicates air flow toward the surface, white indicates upward air flow), and the ends of the lines indicate the predicted position on the 21st. (Animation courtesy NASA/EPA/NOAA/CIMSS IDEA program)