AERMOD and CALPUFF Dispersion Model Evaluation of Mass Conservation

摘要

Air dispersion model performance is typically evaluated using statistical measures and plots comparing predicted and observed concentrations. Many of these statistical measures focus on the highest ranked concentrations, for example Robust Highest Concentration and Fractional Bias, which is reasonable given that the air dispersion models are generally used to evaluate compliance with ambient air quality standards. However, air dispersion models, including the United States Environmental Protection Agency's AERMOD and CALPUFF dispersion modelling systems, are now being used in other applications such as an input to multimedia fate and transport models and inverse modelling to predict the magnitude of emissions from sources. A fundamental requirement for model success in the two above-noted applications is conservation of mass and yet, there are no model evaluation studies demonstrating that mass is conserved within the modelling system. This study evaluated the performance of both AERMOD, a steady-state Gaussian plume model, and CALPUFF, a non-steady-state Lagrangian puff model, in terms of their ability to conserve mass. A large three-dimensional receptor grid was created around a point and area source that emitted a unit quantity of a pollutant. These sources were modelled independently using both AERMOD and CALPUFF for a number of possible meteorological scenarios, such as winter and summer, and calm and windy conditions. In each scenario, AERMOD was run for a single hour since concentrations are calculated discretely, whereas CALPUFF was run for several hours after the unit release to ensure complete removal of puffs from the system since puffs can remain in the domain for multiple hours. For both models, the predicted concentrations and deposition were integrated over the entire receptor grid to give an accurate estimate of total mass within the domain. The mass emitted from the sources was compared to the mass predicted within the puffs and plume to evaluate mass conservation performance for these plume and puff models. The results showed that the use of dispersion models may not be suitable for certain applications, such as inverse dispersion modelling.