An analysis of AERMOD sensitivity to input parameters in the San Francisco Bay Area

The U.S. Environmental Protection Agency (EPA) plans to replace its current air quality guideline model, ISC3, with a new guideline model, AERMOD. AERMOD has been specifically designed to include newly developed and current state-of-the-science modeling techniques. The incorporation of these techniques requires significantly more input parameters compared to ISC3, such as urban population, albedo, Bowen ratio, surface roughness, cloud cover, solar radiation, and heights at which ambient temperature and winds are measured.

In the San Francisco Bay Area, over six hundred different facilities are evaluated every year. Most facilities subject to modeling evaluation either maintain a meteorological station or obtain information from a nearby station for the preparation of meteorological inputs into ISC3. Because of the large number of facility evaluations, preparation of inputs for ISC3 is mostly automated. This study evaluates whether current measurements at existing meteorological stations are adequate for the application of AERMOD. Some of the required additional parameters are not directly measured at current monitoring stations, while others are only being measured at select meteorological sites. Because of the large number of facility modeling evaluations each year, this study also evaluates whether it is possible to automate the preparation of meteorological inputs for AERMOD and whether additional, costly measurements will be beneficial.

This study evaluates AERMOD’s sensitivity to input parameters and how accurately they should be prepared. Current modeling guidance does not completely address the type, accuracy and location of additional measurements needed to apply AERMOD successfully. For example, determining surface roughness, a parameter not directly measured at meteorological stations, is somewhat subjective and there are no clear guidelines for the process. Guidance does not clearly address the preferred type of solar radiation, net or total, or if solar measurements are to be taken at or near every site where AERMOD is applied or at locations deemed representative of larger domains. Guidance is also incomplete in addressing issues arising from meteorological sites not being at the pollutant(s) release location and how sensitive AERMOD is to differences in surface characteristics.

Because a number of the AERMOD input parameters are not currently measured at existing Bay Area meteorological monitoring stations, one objective of this study is to assess whether several input parameters affect modeling results significantly enough to warrant expanded monitoring. By knowing the relative percent change in AERMOD concentration predictions when an input parameter is changed, one can then determine the accuracy needed for specifying that parameter and better estimate which parameters should be measured at each location. Another objective of this study is to investigate how the results from the proposed AERMOD model compare with the existing ISC3 model.

For our analysis, we chose to model several hypothetical typical emission sources in an industrial section of San Francisco. Because the model sensitivity could vary for different source types, we chose to consider three different typical sources: a turbine source (elevated), a ground level point source, and a gas dispensing facility (volume source). A complete yearlong meteorological dataset was compiled for 1992, to serve as our meteorological base year. Sensitivity of AERMOD was investigated for the following parameters: urban population, albedo, Bowen ratio, surface roughness, cloud cover, solar radiation, ambient temperature, and ambient temperature measurement height. Because AERMOD sensitivity could vary by concentration averaging period, we examined the 1-hour, 3-hour, 8-hour, 24-hour, and annual average concentrations.

Over 400 runs were made to determine the sensitivity of AERMOD to these parameters. The modeling analysis showed that for all three typical source types, AERMOD was most sensitive to surface roughness. For the rest of the parameters, AERMOD showed inconsistent sensitivity rankings among the three source types. As an example, for the turbine source a factor of four in the input parameter showed that AERMOD was twice as sensitive to surface roughness as it was to solar radiation, and three times as sensitive to surface roughness as it was to albedo. In contrast, for the gas dispensing facility a factor of four in the input parameter showed that AERMOD was four times as sensitive to surface roughness as it was to both solar radiation and urban population. This illustrates the complex interaction of AERMOD with the many input parameters as a function of source type.