Toward a Global View of the Atmospheric Chemistry of Aerosols

PARTICIPANTS
Marco A. Rodriguez, UC Irvine

ar exhaust, volcanoes, industrial plants, forest fires, and even ocean evaporation fill the Earth's atmosphere with not only gaseous pollutants, but also very small particles. These particles, which include dust, ash, ammonia, organic material, sea salt, and sulfur and nitrogen compounds, clump together with gasses and water in the atmosphere to form aerosols. The aerosols in the atmosphere have down-to-Earth effects on smog levels, human health, and regional and global climate. As part of NPACI's Earth Systems Science thrust area, Donald Dabdub of UC Irvine is linking his work on the chemistry of aerosols to global models of the Earth's atmosphere.

COMPLICATED MATTER

FROM REGIONAL TO GLOBAL

INTERFACE TO EDUCATION

COMPLICATED MATTER

Aerosols are small, complex clumps of matter, composed of anything in the atmosphere that's not a gas, and even in very small concentrations--parts per million or parts per billion--they can have major impacts on the overall dynamics of the Earth's climate. In addition to health effects and overall air quality, aerosols affect climate directly, by absorbing and scattering the sun's radiation, and indirectly, by changing the optical properties, lifetimes, and amounts of clouds. And just as importantly for atmospheric simulations, aerosols act as a small aqueous-phase chemical reactor.

Dabdub's group has pioneered the computational modeling of atmospheric aerosols, and their developments for modeling both the total mass and size distribution of the particles was published in the February 20, 1998, Journal of Geophysical Research and the July 4, 1997, Science. The main advance in Dabdub's model is that it does not make an assumption commonly used to simplify the computation in other models.

"Unlike earlier models, we have not assumed instantaneous gas-aerosol equilibrium for volatile inorganics," said Dabdub, a professor in UC Irvine's Mechanical and Aerospace Engineering Department and leader of the Earth Systems Science thrust area Atmospheric Chemistry project. "This assumption is usually made because otherwise it becomes a major computational sink and one of the most demanding parts of the model." In fact, the aerosol module consumes more than 90 percent of the CPU cycles used in an atmospheric simulation.

The equilibrium assumption works well in predicting the behavior of gasses in the atmosphere, such as carbon dioxide and organic vapors, but experimental studies have shown that it does not always hold for aerosols. For some compounds, mass transfer equilibrium between the gas and aerosol phase is established slowly relative to the time scale over which other changes are occurring. Thus, to be more accurate, the most general form of an aerosol model should not rely on the assumption of instantaneous, local equilibrium of volatile aerosol types.

Figure 2. Aerosol Distribution

The results of simulating a 24-hour period show the air quality across the Los Angeles Basin. The UC Irvine aerosol module provides additional information on the size distribution of particulate matter, here ranging from about 120 micrograms per cubic meter (red) to about 20 micrograms per cubic meter (light purple).

FROM REGIONAL TO GLOBAL

As part of NPACI's Earth Systems Science thrust area, Dabdub is planning to take advantage of the NPACI infrastructure to advance atmospheric chemistry modeling. Dabdub's aerosol code is a module that can be plugged into a host model--a chemical transport simulation that models the chemistry of gasses in the atmosphere. Currently, the aerosol module has been added to the Caltech Air Quality Model (AQM) which simulates the urban atmospheric chemistry of the Los Angeles basin (Figure 1).

The AQM alone models only the gas-phase chemistry, and simulating a 24-hour time period of the Los Angeles area requires a half hour on a fast workstation. Plugging in the aerosol module and running for the same 24-hour period, however, takes a week on the same workstation. Dabdub's group has parallelized the code and on a cluster of 16 Pentium II-based workstations has gotten the time down to four hours (Figure 2).

"We have been using parallel machines as a necessity, not a luxury," Dabdub said. "With NPACI resources, as a next step, we want to take our aerosol module and, instead of looking just at the Los Angeles area, look at the world." As a new NPACI project, Dabdub's group is beginning the process of porting the module to larger parallel machines and adding the code to complete the feedback loop between aerosol chemistry and the global model. And because of the modular design, once the aerosol code can be plugged into one global model, it will be possible to incorporate it into other host models.

At present, the group is working with the IMAGES global model developed at the National Center for Atmospheric Research in Boulder, Colorado. A two-year simulation of only the gas phase takes three months on a workstation, or two days on a CRAY C90 at the Environmental Protection Agency facility in Michigan. Within NPACI, Dabdub will look at incorporating the aerosol module into host models being developed as part of the Multi-scale, Multi-resolution Modeling project led by C. Roberto Mechoso in UCLA's Department of Atmospheric Sciences.

"Adding the aerosol module will be new science," Dabdub said. "And because of the computational demands, it's also an ideal application for a metasystem, such as Legion or Globus." Dabdub is currently examining the possibilities of both architectures, which are part of NPACI's Metasystems thrust area.

Figure 3. Problem-solving Environment

Donald Dabdub's group at UC Irvine created a graphical user interface to the Caltech Air Quality Model that can be used to enhance undergraduate education. Dabdub is working with NPACI's Ed Center at San Diego State University to make this tool available for use in the undergraduate curriculum.

INTERFACE TO EDUCATION

Dabdub's group is also making it possible to use computational models in the university curriculum. As part of their work with the Caltech AQM, Dabdub's group created a front end for the model in Visual Basic with a point and click interface (Figure 3). Without the computationally intensive aerosol module, the model can then be used for undergraduate or graduate studies.

For example, students might be presented with a question such as, "What emissions would you reduce in Los Angeles to decrease ozone concentrations?" After working out a potential scenario and the chemical reactions, the students can run the model to support their hypotheses. Dabdub is working with Kris Stewart, director of NPACI's Educational Center on Computational Science and Engineering, to make the modeling tool available to a wider audience of university instructors.

Access to an easy-to-use air quality model could also benefit fields outside of atmospheric science, such as urban planning, government policy, and health impact analyses. The model might be used for studies to establish emission control legislation, evaluate different emission control strategies, plan the locations of future pollution sources such as power plants and industrial areas, and assess responsibilities for existing levels of pollution.

"Our participation in NPACI benefits both our modeling and educational efforts," Dabdub said. "It is allowing us to expand the number of host models into which we can incorporate our aerosol module, providing us access to high-performance resources, as well as expand the reach of our educational use of atmospheric models. We are working with NPACI on many levels." --DH