Insights into the positive artifact correction method for ambient organic carbon in the IMPROVE network

Ambient particulate organic carbon (POC) is susceptible to positive and negative sampling artifacts when it is collected with quartz filter media. The Interagency Monitoring of Protected Visual Environments (IMPROVE) network corrects for gas-phase organics that adsorb onto quartz filters (positive artifacts) in its reporting of particulate carbon concentrations for visibility monitoring in pristine environments. In the IMPROVE method for positive artifact correction, the monthly median value of in-contact, quartz-behind-quartz (QQ) back filter organic carbon (OC) concentrations from selected sites is subtracted from the front filter OC concentrations to estimate POC. An evaluation of some of the assumptions made with this correction method was conducted by analyzing the data from two 17-day field campaigns that took place on the campus of the University of California, Davis in 2008 and 2009 using IMPROVE samplers. Three modules were used to collect 24-hour integrated front, back, and field blank samples for the study. OC concentrations for in-contact and separated QQ filter configurations were compared and the results indicated that no transfer of carbonaceous material occurred between in-contact QQ filters. A linear regression of the data from these configurations and paired t-test results indicated that OC from in-contact filters is no different from OC from separate filters when examining concentrations, partitioning between filters, and artifact-corrected concentrations. The field blank filter OC concentrations were below the minimum detection limit, did not vary with the front filter OC concentrations, and were determined to be statistically different from back filter concentrations in the same module. A third module housed an upstream Teflon filter with a back quartz filter (TQ); the back filter OC was always higher than the back filter OC for the QQ configuration. Although the TQ method produced consistently lower POC concentrations, they were mostly within measurement uncertainty of the QQ method. Therefore, the TQ setup may provide an upper bound estimate and the field blanks a lower bound estimate for the positive artifact on the front quartz filter. However, it was not possible to determine which method was the most accurate estimate of sampling artifacts in IMPROVE network samplers without quantification of the negative artifact.