Summary: Pursuant to a congressional request, GAO reviewed the status and results of the Environmental Protection Agency's (EPA): (1) testing of its air quality dispersion models to determine their reliability; and (2) efforts to develop additional models to meet identified needs.
GAO found that: (1) EPA evaluated 29 of its 48 air quality dispersion models using the recommended uniform statistical approach; (2) EPA evaluated two of the models before it adopted the uniform approach; (3) other organizations evaluated nine of the models, but did not use the uniform approach; (4) the models' reliability ranged from underestimating pollution concentrations by 50 percent to overestimating concentrations by 1,720 percent; (5) some models gave substantially different results when tested at different sites; (6) EPA is studying the use of simulated conditions to obtain more reliable estimates of model reliability; and (7) EPA believes that the model estimates are still useful as regulatory and research tools as long as users recognize their limitations. GAO also found that EPA: (1) in 1985, identified 65 of 241 modelling needs as high priority; (2) has not identified the actual number of models or amount of funding it requires to meet all of its modelling needs; and (3) estimated that it would cost $97 million to develop and evaluate models to address 6 urgent and 70 high-priority modelling needs.
