UNCERTAINTIES IN SOURCE EMISSION RATE ESTIMATES USING DISPERSION MODELS

The source emission rates during the Prairie Grass dispersion experiments were carefully observed and were adjusted by the experimentalists so that they were about twice as high during unstable conditions as during stable conditions. The question was asked whether observed concentrations and meteorological conditions could be used in dispersion models in order to predict source emission rates and verify this factor of two difference. Three types of simple dispersion models were applied to this problem, with the result that for the model based on Monin-Obukhov similarity theory, the uncertainties in predictions of source emission rates for individual runs were at best about +/- 10-20% when observed cross-wind integrated concentrations from the 50 m arc were used. Consequently this model could discern the factor of two difference in average source emission rates for the two sets of field trials which consisted of about 20 runs each. However, some models, such as the Gaussian plume model, exhibit uncertainties of about +/- 70% to a factor of two in predictions for individual runs, and hence could not discern the difference in average source emission rates when concentration observations at downwind distances of 100-800 m are used. It is found that the use of observed cross-wind integrated concentrations produces more accurate conclusions than the use of observed point concentrations, for the uncertainties in predictions of source emission rates are about a factor of two larger when the observed point concentrations are used.