Investigation of sources of volatile organic carbon in the Baltimore area using highly time-resolved measurements

Abstract

A multivariate receptor model, UNMIX, was applied to hourly volatile organic hydrocarbon concentrations measured at a Photochemical Assessment Monitoring Station, located in the Baltimore, MD area, during the summer months from 1996 to 1999. Six possible volatile organic carbon source categories were identified, and gasoline-related sources such as vehicle exhaust, gasoline vapor, and liquid gasoline contributed more than half of the total VOC concentration. Natural gas, surface coatings, and biogenic source categories accounted for 13%, 12% and 11% of the total VOC, respectively, when all hourly measurements are considered. In order to investigate possible relationships between each source category and episodes of elevated ozone concentrations, NO_x, meteorological variables such as surface wind speed, and temperature as well as the hourly contribution of each source category, obtained from UNMIX, were split into high ozone days and low ozone days. Results of a comparison of diurnal patterns observed on high-ozone days with those of low-ozone days for each source category suggested that biogenic emissions may have contributed significantly to local ozone production during episodes in which the 8 h ozone standard was violated in this area.

Introduction

Ground level ozone is a major problem in rural areas as well as urban areas because of its adverse impacts on human health and on crops and forest ecosystems (Sillman, 1999). Since nitrogen oxides (NO_x) and volatile organic compounds (VOCs) were uncovered as key precursors of ground level ozone formation by Haagen-Smit, and co-worker in the1950 s (Haagen-Smit and Fox, 1954), many studies have been focused on determining effective strategies for ozone reduction. The complex relationship between ozone and its precursors, and the strong dependence of ozone formation on meteorological conditions are the major difficulties associated with the study of ground-level ozone. Though meteorological conditions and transport are important variables for ozone formation and accumulation, anthropogenic VOC and NO_x emissions are the primary focus of ozone reduction programs in that only they are controllable.

Most NO_x emissions are from combustion-related sources such as motor vehicles and fossil-fueled power plants in urban and suburban areas. On the other hand, VOCs are emitted from wide variety of sources, both anthropogenic and biogenic. In the United States, it is estimated that the amount of VOC emissions from biogenic sources is of the same order of magnitude as the total emissions of anthropogenic VOCs (Watson et al., 2001; Atkinson, 2000; Trainer et al., 2000). Guenther et al. (2000) concluded that over 98% of total biogenic VOCs in North America are from vegetation, with the remainder from soils and biomass burning. Biogenic VOCs are composed primarily of isoprene (35%), terpenoid compounds (25%) and non-terpenoid compounds (40%).

Lagrangian and Eulerian models have been used extensively to predict the change of ozone in response to controls of NO_x and VOC. Several modeling studies have shown that, in general, VOC controls may be effective in reducing the ozone levels in urban and suburban areas, which are most strongly impacted by anthropogenic emissions (Hanna et al., 1996; Possiel and Cox, 1993; McKeen et al., 1991). In addition, a simple rule, based on a modeling study, that morning VOC/NO_x ratios less than 10 indicate VOC-sensitive peak ozone, and morning VOC/NO_x ratios more than 20 indicate NO_x-sensitive peak ozone has been used to justify NO_x-VOC sensitivity prediction and policies (Sillman, 1999). Therefore, a reduction of VOC emissions in VOC-sensitive regions may be effective in reducing ozone, while a reduction of NO_x emissions in NO_x-sensitive regions may lead to ozone reduction. However, the emissions from biogenic sources have been reported to be most likely underestimated in modeling studies, and several analyses of observed measurements have suggested a significant role of biogenic hydrocarbon emissions in many urban and suburban locations in ozone formation (Chameides et al (1992), Chameides et al. (1988); Cardelino and Chameides (2000), Cardelino and Chameides (1995)). Thus, it cannot easily be determined whether reducing VOC emissions or NO_x emissions or both is the most effective strategy for ozone reduction for a given area, using a simple rule based on modeling studies conducted for only a few specific areas (Sillman, 1999).

Hence, the development of an effective ozone reduction strategy in a given urban or suburban area requires an accurate understanding of three key ozone precursor relationships: the relative concentrations of NO_x and VOC in the area, the importance of natural VOC relative to anthropogenic VOC in the atmosphere (Piety et al., 2003; Trainer et al., 2000; Cardelino and Chameides, 1995), and the significance of long range transport as compared to local emissions of ozone precursors. While the importance of long range transport versus local emissions can only be resolved using photochemical and meteorological modeling, source apportionment of VOC can be expected to give insight into the importance of natural VOC relative to anthrogenic VOC in the atmosphere for a given area.

In this work, hourly ambient surface measurements of the concentrations of 55 VOC species, taken from a Photochemical Assessment Monitoring Stations (PAMS) site in Baltimore County in the state of Maryland during the summer months of 1996–1999, are used to investigate the relationships between ozone and its natural and anthropogenic precursors. To identify and apportion VOC sources, the UNMIX receptor model is used. Specific focus is placed on analysis of the observed diurnal variation in concentrations as well as source contributions. These phenomena have been investigated for VOC in other geographical regions, for example, Seoul, Korea (Na et al., 2003) and the Paso del Norte region of the Texas–Mexico border (Fujita, 2001). VOC emissions can vary greatly from region to region, and in contrast to the above-mentioned regions, biogenic emissions at the location under study here are significant. With the receptor modeling results, a study of the contribution of the identified source categories to episodes of high ozone concentrations is conducted. The results of this study will be useful in the development of effective ozone control strategies in this and similar regions.