Applications and evaluation of receptor modeling methods for source contribution of volatile organic compounds

Receptor models are important complement to source modeling for estimating source contributions to ozone precursors at source and receptor locations. The comparisons of receptor and source models have served as the basis for developing alternate emission inventories to evaluate the sensitivity of modeling results to emission inventory uncertainties. Many Volatile Organic Compounds (VOC) sources have similar composition and this can present a challenge for the receptor models to resolve source contributions of emission sources. Using source profiles with specific tracers could minimize that problem.; Given the critical role that emission inventories play in development of air quality management strategies it is important to determine the uncertainty associated with inventory of various source categories in addition to advancements in VOC fuel technology. In the dissertation this was addressed with multiple source attribution methods under realistic conditions where the actual source contributions were known. The contributions are in effect an inventory of emissions that affect the VOC level at the receptor site therefore by comparing derived with a known inventory, uncertainty can be evaluated.; A comprehensive VOC dataset in regard to number of species, time resolution, and locations was used for receptor modeling. Source contribution results showed that fewer uncertainties were associated with the dominant mobile source emissions. For CMB and PMF models greater uncertainties were associated with minor source categories, which do not contain tracer species that have a strong influence on the apportionment. Photochemical aging of VOC emissions that occurred between source and receptor in the large populated areas of Central and Southern California presented challenges for both models. CMB predicted source contributions of emissions with minimal photochemical degradation whereas PMF gives greater contribution to biogenic sources compared to the source model. A emphasize was on using dataset where photochemistry can be avoided and to focus on fresh industrial emissions with the greatest source contribution uncertainty. The use of specific wildfire tracers, chloromethane and acetonitrile with long atmospheric lifetimes facilitated the attribution of wildfire contributions to VOC in Reno, Nevada. This dissertation enables policymakers/modelers to improve emission inventories and to support effective air quality management decision for the foreseeable future.