Elsevier

Atmospheric Environment

Volume 37, Issue 4, February 2003, Pages 545-562
Atmospheric Environment

Comparison of hybrid receptor models to locate PCB sources in Chicago

https://doi.org/10.1016/S1352-2310(02)00886-5Get rights and content

Abstract

Results of three hybrid receptor models, potential source contribution function (PSCF), concentration weighted trajectory (CWT), and residence time weighted concentration (RTWC), were compared for locating polychlorinated biphenyl (PCB) sources contributing to the atmospheric concentrations in Chicago. Variations of these models, including PSCF using mean and 75% criterion concentrations, joint probability PSCF (JP-PSCF), changes of point filters and grid cell sizes for RTWC, and PSCF using wind trajectories started at different altitudes, are also discussed.

Modeling results were relatively consistent between models. However, no single model provided as complete information as was obtained by using all of them. CWT and 75% PSCF appears to be able to distinguish between larger sources and moderate ones. RTWC resolved high potential source areas. RTWC and JP-PSCF pooling data from all sampling sites removed the trailing effect often seen in PSCF modeling. PSCF results using average concentration criteria, appears to identify both moderate and major sources. Each model has advantages and disadvantages. However, used in combination, they provide information that is not available if only one of them is used. For short-range atmospheric transport, PSCF results were consistent when using wind trajectories starting at different heights.

Based on the archived PCB data, the modeling results indicate there is a large potential source area between Joliet and Kankakee, IL, and two moderate sources to the northwest and south of Chicago. On the south side of Chicago in the neighborhood of Lake Calumet, several PCB sources were identified. Other unidentified potential source location(s) will require additional upwind/downwind field sampling to verify modeling results.

Introduction

To reduce pollutant concentrations in the atmosphere, pollution sources must be identified, emission estimates made, and effective management strategies developed. To help locate pollutant sources several receptor models have been developed that combine meteorology with the measured chemical compositions so that the probable locations of emission sources can be found. These methods combine estimates of the motion of the air backward in time with concentrations measured at a sampling location. Examples of these hybrid receptor models include residence time analysis (RTA) (Ashbaugh et al., 1985; Poirot and Wishinski, 1986), quantitative transport bias analysis (QTBA) (Keeler, 1987), potential source contribution function analysis (PSCF) (Zeng and Hopke, 1989; Cheng et al. (1993a), Cheng et al. (1993b); Gao et al., 1993; Hopke et al., 1993), concentration weighted field (CWT) (Seibert et al., 1994) and residence time weighted concentration (RTWC) (Stohl, 1996). These models have been used to identify pollutant sources hundreds of miles from the receptors. However, they have not been used to locate sources relatively near to the sampling sites, which was the purpose of this study.

Polychlorinated biphenyls (PCBs) emission sources are not well known and based on existing emission inventories are largely underestimated (Hsu et al., 2002). Since measurements of PCBs require sampling times of at least several hours, tracing pollutant sources is difficult and expensive. Without proper source inventories and known source locations, dispersion models cannot be used to estimate impacts on ambient air concentrations that can be used for air quality management.

A method that can be used to locate pollutant sources using long-term samples is receptor modeling. In this study to locate sources of PCBs to the atmosphere, a hybrid receptor model, PSCF, was initially used. Modeling results resolved three PCB source sectors in Chicago. They were (1) the northwest direction pointing toward Madison, WI, (2) a southwest trend toward Joliet, IL (Fig. 1) and (3) the south side of Chicago in the neighborhood of Lake Calumet (Hsu et al., 2002). After identifying potential source areas, modeling results were verified with upwind/downwind samples that pinpointed several PCB sources in Chicago (Hsu et al., 2002). PSCF modeling was essential in locating the unknown PCB sources. This paper will compare two other hybrid receptor models, RTWC and a new method, concentration weighted trajectory (CWT), to PSCF to determine their utility in locating unidentified pollutant sources.

Results from variations of these models including PSCF using average and 75% criterion concentrations, Joint Probability PSCF, and changing point filters and grid cell sizes for RTWC, and PSCF using wind trajectories started at different altitudes are also compared.

Section snippets

Potential source contribution function (PSCF)

Air parcel backward trajectories are related to the composition of collected material by matching the time of arrival of each trajectory at the receptor site to the sampling time interval. The PSCF value for a single grid cell is calculated by counting the trajectory segment endpoints that terminate within that grid cell. The number of endpoints that fall in the ijth cell is n(i,j). The number of endpoints for the same cell having times of arrival at the sampling site corresponding to pollutant

Trajectory calculation

The NOAA HYSPLIT trajectory model and meteorological data downloaded from NOAA were used to calculate air parcel back-trajectories for the hybrid receptor modeling (HYSPLIT4, 1997).

Ambient air PCB data

Four PCB data sets were used in this study (Table 1). These include the Lake Michigan Urban Air Toxics Study (LMUATS) (1991) (Keeler, 1994), Atmospheric Exchange Over Lakes and Oceans Study (AEOLOS) (1994) (Simcik et al., 1997), the Integrated Atmospheric Deposition Network (IADN) (1996–97) (Basu, 1999), and measurements made at IIT Rice campus in Wheaton IL west of downtown Chicago (Hsu et al., 2002). Since ambient air PCB concentrations increase with increasing temperature (Lee et al., 1998;

PSCF results using 75th percentile criteria (75% PSCF)

In PSCF modeling, data were divided into two groups, greater and less than a predetermined criterion. The PSCF results for IIT using the AEOLOS results generated using a mean concentration criterion (Fig. 2, top) were similar to the ones using a 75th percentile concentration (Fig. 2, bottom) with some potential source areas removed. In general, sources for Chicago are to the southwest. This direction is typically associated with elevated temperatures that increase the volatilization of

Conclusions

All three trajectories ensemble models consistently indicated similar PCB source locations. Each modeling approach was found to have some advantages compared to the others. Thus, when used together, they provide better information on source areas than could be obtained by using only one of them. Seventy-fifth percentile PSCF and CWT may be useful for distinguishing between large and moderate sources. JP-PSCF permits pooling data from multiple sampling sites and has the potential of removing

Acknowledgements

This work was funded in part by the Great Lakes National Program Office with Angela Bandemehr and Paul Horvatin Project Officers. We also thank Eleanor Hopke at Clarkson University, Dick Todd at IIT Rice Campus, Thomas Murphy at DePaul University and Jeff Chiarenzelli and Jim Pagano and their research group at SUNY Oswego for their assistance. The program for RTWC was supplied by Andreas Stohl. His assistance is also appreciated.

References (20)

There are more references available in the full text version of this article.

Cited by (0)

1

Current address. California Air Resources Board.

View full text