Emissions Inventory Estimation Improvements using a Four-Dimensional Data Assimilation Method for Photochemical Air Quality Modeling

Conclusions

Emissions inventories used in typical three-dimensional air quality model applications have significant uncertainties, and in many cases suffer major deficiencies in correctly approximating real source strengths and compositions. A four-dimensional data assimilation method for air quality models was used to suggest emission correction factors to the base case inventory used in modeling Atlanta’s August 9–10, 1992 ozone episode. Total VOC mass emissions from anthropogenic sources in the base case inventory are suspect of being underestimated by about a factor of two, while area source (stationary and mobile source combined) anthropogenic NO_x emissions appear to be well estimated in the base case inventory. Point source NO_x was estimated to be underpredicted by about 50 to 60%. Finally, biogenic VOC emission corrections were less that 10%. Limitations in the procedure due to a poorly resolved observations field were evident while trying to compute correction factors for NO_x from stationary and mobile sources (separately). Corrections for mobile-source NO_x were more certain than for area-source due to the concentration of nitrogen-containing species measurements around the city’s downtown area, precisely where the mobile sources are more abundant. Further analysis of the VOC emissions indicates that the speciation profiles used to partition the total anthropogenic VOC mass need to be revised. The AQM-FDDA approach provides not only information of the strength, uncertainty and bias of the base case emissions inventory, but also provides revised information for source contribution analyses and control strategies design. Here it was demonstrated that the ozone response toa similar mass reduction of emissions is different if the base case or corrected inventory is used. However, on a relative scale, VOCs reactivities are very similar no matter what inventory is used.