New Algorithm Increases Accuracy of Air Pollution Predictions

26 May 2008

This map shows the bias in the calculations of carbon monoxide concentration in the atmosphere after the application of Kotamarthi’s algorithms. Click to enlarge.

A new computer algorithm developed by environmental scientist Rao Kotamarthi of the US Department of Energy’s Argonne National Laboratory, in collaboration with Alexis Zubrow, now at the University of North Carolina, and Li Chen, now at Bristol University, UK, quickly assimilates observational data into climate models to generate more reliable forecasts of where carbon monoxide hot spots will occur.

Although Kotamarthi’s model looks expressly at carbon-monoxide emissions, researchers could use similar algorithms to examine the atmospheric concentrations of carbon dioxide and other greenhouse gases and aerosols. Kotamarthi and Argonne environmental scientist Paul Hovland have initiated a NASA-funded project to develop data assimilation methods for worldwide chemical transport models that can incorporate satellite measurements of several atmospheric gases.

Differences of one or two parts per million in the concentration of pollutants like ozone and carbon monoxide in certain cities can mean the difference between achieving a target and having to implement additional costly provisions to rectify non-compliance.

Because of the high stakes involved in meeting air-quality targets, scientists, city officials and regulators all desire an effective and accurate way not only to measure air quality but also to predict where pollution hot spots will occur and plan for additional control strategies.

By incorporating observation data into our models, we can refine our predictions. Meteorologists have been doing it for a while, but people in the chemical trace gas and aerosol modeling community have just started doing it.

—Rao Kotamarthi

When scientists include measurement data in their models, the uncertainties in those measurements compound the uncertainties already present in the model. Compensating for these new uncertainties requires a mathematically rigorous analysis, so Kotamarthi and his colleagues decided to launch many simulations with slightly different initial conditions. This ensemble-based approach creates a better method to correct for uncertainty, he said.

We need to generate better forecasts of ozone, carbon monoxide and other trace gases for air-quality applications. And the way to do that is by assimilating the data taken today into the forecast for tomorrow. But the data come with certain types of uncertainty that most models are unable to accommodate.

There’s very little merit in trying to decide a policy based on a single emissions scenario. We need to combine different measurements with a suite of new mathematical techniques in order to help reduce the uncertainty in our forecasts.

—Rao Kotamarthi

Data assimilation may also boost researchers’ ability to project likely climate scenarios for the “near-term decadal scale”—approximately 10 to 20 years—which would help public officials assess the consequences of their decisions that concern climate change.

The results of the study were published in the Journal of Geophysical Research.

Kotamarthi’s research was funded by the Office of Biological and Environmental Research in the US Department of Energy’s Office of Science.

Resources

• Zubrow, A., L. Chen, and V. R. Kotamarthi (2008), EAKF-CMAQ: Introduction and evaluation of a data assimilation for CMAQ based on the ensemble adjustment Kalman filter, J. Geophys. Res., 113, D09302, doi:10.1029/2007JD009267