Intercomparison of automated methodologies for determination of ambient isoprene during the PROPHET 1998 summer campaign

The Program for Research on Oxidants: PHotochemistry, Emissions, and Transp ort (PROPHET) 1998 summer campaign, conducted at the University of Michigan Biological Station, provided a unique opportunity to compare isoprene meas urement techniques that were automated, sampled and analyzed on-line, and p rovided relatively fast time resolution. Assessment of the data quality for fast isoprene measurements is important because isoprene dominates the sur face chemistry at many rural sites and even some urban environments. An inf ormal intercomparison was conducted by evaluating ambient isoprene mixing r atio data generated by five different instruments: quadrupole ion trap (QIT ) MS, the chemiluminescent-based fast isoprene sensor (FIS), and three gas chromatograph/mass spectrometry (GC/MS) techniques. The GC/MS methods were deployed and maintained by Purdue University (GC/MS-P), the National Center for Atmospheric Research (GC/MS-NCAR), and the Rosenstiel School of Marine and Atmospheric Science (GC/MS-RSMAS). The FIS was deployed and maintained by NCAR, Hills-Scientific.com and Washington State University, while the Q IT was implemented by Purdue University. The GC/MS-P was chosen as the refe rence method to evaluate the agreement of the data set. The data were evalu ated for time-matched samples through regression analysis, ratio analysis, and percent difference analysis relative to GC/MS-P. For measurement data i n the central 90th percentile relative to the median, the mean percent diff erence was 21% for GC/MS-NCAR, 41% for QIT, 42% for GC/MS-RSMAS, and 88% fo r the FIS. Potential sources of disagreement, especially for low-concentrat ion data, such as variations in sampling time, interferences, method precis ion and accuracy, and limited cross-calibration, are discussed.