January 2014: Air Modeling and AERMOD

The United States Environmental Protection Agency (U.S. EPA) has determined that the appropriate air quality model to use for most purposes is a model called AERMOD. This model is used for two primary purposes. First, when an existing source has measured air quality levels above a national ambient air quality standard (NAAQS), the model is used to determine how much the source must reduce emissions to meet the standard. The second use of the model is when a new source wants to build, they must predict what the air quality impacts of their facility and those around it will be.

The accuracy of the model is extremely important. In the past, without one hour standards for sulfur dioxide (SO2) and nitrogen dioxide, the accuracy was less of a concern. Now it is very important how well the model does in predicting air quality levels. If it overpredicts, sources may be required to put in unnecessary controls. New sources may not be allowed to build because the model predicts that their impacts are too great. If the model underpredicts, then we will not be protecting the public adequately because measured values would be higher than those expected.

IDEM has been working for over two years on a project to determine how accurate the model is. We have a power plant in southwestern Indiana, Duke – Gibson, that has four ambient monitors that measure SO2. The stacks at this facility record hour by hour data, which includes the amount of sulfur dioxide emitted during each hour. By using these two datasets we were able to predict SO2 levels for each hour of an entire year (2010) and compare the predicted levels with those measured at the four monitors.

Our analysis focused on two areas. The first was hours when the model predicted SO2 levels at or above the standard of 75 parts per billion (ppb). Under these conditions, the model almost always overpredicted the measured SO2 level, normally by more than a factor of two. The second area that we looked at was the hours when the monitors recorded high SO2 levels (above 35 ppb). Under these cases, the model appears to underpredict SO2 levels.

We are working to propose some adjustments to the model that would make the measured and predicted SO2 levels agree better. Once we have these adjustments worked out we will submit them to U.S. EPA to see if they are interested in making corrections to AERMOD. In the long run we all would like to have an air quality model that does a good job of accurately estimating concentrations so that we can have confidence in the values that it predicts. Our single case for one utility may not be persuasive to U.S. EPA. If not, we hope that other states or utilities will conduct similar studies so that we will have the necessary information to improve AERMOD, if it is justified.