 |
NARSTO Workshop on Innovative Methods for Emission Inventory Development and Evaluation University of Texas, Austin October 14-17, 2003 |
 |
Source Apportionment of VOCs in the Houston, Texas Area
Steven G. Brown, Hilary R. Hafner
Sonoma Technology, Inc., Petaluma, CA 94954
Hourly concentrations of over fifty volatile organic compounds (VOCs) were measured between 1998 and 2001 by automated gas chromatographs (auto-GCs) at a number of sites in Houston, Texas, as part of the Photochemical Assessment Monitoring Stations (PAMS) program. This extensive database was collected and analyzed in order to further understand emissions in the Houston area and the role of specific sources, source areas, compounds, and meteorology in ozone formation. The Clinton Drive site is well suited for detailed source apportionment analysis using Positive Matrix Factorization (PMF) because it is located in the heavily industrialized Houston Ship Channel, where extremely high concentrations of reactive ozone-forming VOCs are commonly found in abundance. One advantage of PMF is that it uses error estimates for each individual data point, which is different than other source apportionment tools. Using this corresponding error matrix allows PMF to appropriately scale and treat missing and below detection data differently than the “true” measured data. This effort allows specific factors and sources to be identified and analyzed, with the overall goal to determine their potential impact on ozone formation in the area.
Source apportionment using PMF was successfully completed on the dataset of over 21,000 hourly observations of 39 species from 1998 to 2001. Mass was well reconstructed by the model, and numerous sources were identified. Most factors showed significant dependence on wind direction, indicating specific source areas. When compounds in the source profiles were scaled by their ozone formation potential using the Maximum Incremental Reactivity (MIR) scale, no individual source appeared to dominate the total reactivity. Only six factors were significantly higher on mornings of ozone episodes, indicating that these sources may be more important in ozone exceedences. This paper will further discuss results and implications with regard to ozone formation and emission sources.