iSGTW Feature - Powerful pollution model moves to the grid


Feature - Powerful pollution model moves to the grid


High ozone levels threaten Bulgaria's most important agricultural crop, wheat, with yields down by ten to twenty percent.
Image courtesy Tzvetan Ostromsky

In January 2008, Bulgarian authorities installed a public air pollution monitoring system in Sofia, the country's capital. The system allows passing citizens to check air quality and keeps authorities informed of current pollution levels.

Computer scientist Tzvetan Ostromsky is also keeping an eye on pollution levels. Ostromsky is working at the Bulgarian Academy of Sciences to gridify a large-scale air pollution model.

Called the Danish Eulerian model, it combines data on convection, diffusion, precipitation, emissions and more to predict future air pollution trends.

"For a small country like Bulgaria it's especially important to know what pollutants we'll receive and from where," says Ostromsky. "Emissions can travel thousands of kilometers-most of what Bulgaria gets actually originates from elsewhere in Europe."

Ozone-damaged wheat

Pollutants can also chemically react in the air, forming new pollutants as they travel.

For example, emissions from fuel, factories, gas-run vehicles, paints and solvents can all combine to form ozone, a chemical that impairs the growth of wheat. Annual yields of wheat, Bulgaria's most important agricultural crop, are down ten to twenty percent due to high ozone levels.

Sofia, a low-lying city surrounded by mountains, endures an ongoing struggle with air pollution.
Image courtesy of Cranker

Fortunately, the Danish Eulerian model is already making an impact: predictions from DEM contributed to restrictions in ozone emissions in the late 90s.

"We now seeing benefits of those restrictions and yields are increasing," says Ostromsky. "But it's very fragile."

The future of forecasting

"I believe grids are the future of complex and long-term forecasting," Ostromsky says. "We work with very large data sets, we require a lot of storage and we collect data that originates in many different physical locations. Now, using a grid, I don't know where the task is calculated, but I get the results and I'm surprised at how quickly I get them."

Tzevatan Ostromsky recently presented his results at the 3rd EGEE User Forum in France.

- Danielle Venton, EGEE

Authors