Measurement Errors in the PM2.5 Chemical Speciation Network (CSN): Implications for Exposure Estimation

JAMES FLANAGAN (1), R.K.M. Jayanty (1), Ed E. Rickman (1), Max R. Peterson (1)

(1) RTI International, Research Triangle Park, NC

     Abstract Number: 319
     Last modified: November 9, 2009

Abstract
The PM2.5 Chemical Speciation Network (CSN) sponsored by the U.S. EPA is now ten years old, and over 300,000 filter samples from more than 250 locations in the U.S. have been collected. In that time, the network data have been used to:
- assess trends in species concentrations over time and assess the success of mitigation strategies
- support state implementation planning by supplying data for source identification
- supply data needed for human exposure and risk assessment and epidemiological studies

Sampling and analysis methods, and the suite of chemical species to be reported under the CSN, were chosen in the late 1990s based on a variety of sometimes conflicting considerations. This presentation will examine several significant measurement issues that should be understood when the CSN data are used in exposure assessment and other health-related applications.

The relatively low flow rate of the CSN sampler limits the detectibility of low-level analytes of health significance, including certain trace elements (e.g., Cu, Cr, Se, As), which are measured by X-ray fluorescence. Tables of detection limits and and data completeness will be presented and the potential impacts of analytical detection limits and completeness will be discussed in the contexts of health-related assessments and source apportionment.

The method used for measuring Elemental Carbon (EC), a proxy for diesel soot, and Organic Carbon (OC) has several known measurement issues. These include determining the split between OC and EC based on laser reflectance, and the "OC artifact" which complicates the accurate assessment of particle phase semivolatile organics. In addition, the multiplier used to convert from measurements made "as carbon" to total mass (by accounting for unmeasured components such as bound oxygen, hydrogen, and associated water) depends on the composition of the aerosol, which can vary by season or locale.

Finally, historical CSN data demonstrate that nearby CSN sites can differ signficantly from each other in the concentrations of key species, which could have implications for the accuracy and precision of exposure estimates. Being aware of issues such as these will help data users deal with some of the potential biases and uncertainties in the CSN and other PM speciation datasets.