Biases in Clean Air Status and Trends Network filter pack results associated with sampling protocol

Network filter pack sampling protocol changed in 1989 from requiring a week-long daytime sample and a week-long nighttime sample to requiring a single week-long sample per week at each monitoring site. In the current study, single-filter pack weekly results are compared with weekly results aggregated from separate daytime and nighttime weekly filter pack samples collected at the monitoring site located at Egbert, Ontario, Canada. Comparisons of the concentrations resulting from the two sampling protocols for all major chemical species (SO42−, NO3−, NH4+, HNO3, and SO2) show median biases of <5 nmol m−3 (0.1 ppb) and median relative biases of <10%. Median relative biases have the same sign for each species, suggesting biases in the same direction. Based on median differences, composite day–night weekly sampler results generally exceed the single-sampler weekly results (in all cases except for the summer nylon filter HNO3), and the magnitude depends on the constituent and on the season. Examination of seasonal results reveals large discrepancies in some cases, especially during summer. To use Clean Air Status and Trends Network results for trends analyses over time periods encompassing the 1989 protocol change, it may be useful to put all of the data on the same basis of sampler protocol. Algorithms derived from linear regression analyses using paired bootstrap sampling are offered to convert the recent results to the pre-1989 basis; however, they may only be appropriate for sites in the eastern US. Chemical and statistical reasoning suggests that the results of day–night weekly sampling are usually consistent with higher accuracy than single-sampler weekly results. Adjustments are indicated for summer Teflon NO3 and nylon HNO3, for summer and fall Teflon NH4, and for Total SO2 in each season. Nylon filters are also shown to have variable collection characteristics for SO2 that are consistent with a humidity effect. A network-wide change in the SO2 collection and/or retention characteristics of the nylon filters is found in April 1997.