Comparison of particulate sulfate and nitrate at collocated CASTNET and IMPROVE sites in the eastern US

Airborne concentration measurements of particulate SO(4)(2-) and NO(3)(-) from two networks, the Clean Air Status and Trends Network (CASTNET) and Interagency Monitoring of PROtected Visual Environments (IMPROVE) Network, are compared at 10 collocated sites in the eastern US for the period 1990 through fall of 2006. CASTNET samples are comprised one nominal 168-h sample per week versus 24-h IMPROVE samples collected twice a week or every third day. To permit comparison, CASTNET and IMPROVE concentration data are matched by week and composed into paired seasonal mean concentration values. CASTNET and IMPROVE seasonal mean SO(4)(2-) and NO(3)(-) concentrations are highly correlated, with respective correlation coefficients of 0.97 and 0.91. Two sites with marine influence display large median biases (MRBs) for NO(3)(-) (-42% and -102%) and smaller MRBs for SO(4)(2-) (-14% and -17%), indicating seasonal mean CASTNET concentrations to be biased higher than corresponding collocated values from IMPROVE at these two sites. Statistically significant median biases (MBs) and MRBs are present for comparisons of seasonal mean CASTNET and IMPROVE SO(4)(2-) concentrations, indicating seasonal mean CASTNET SO(4)(2-) concentrations to be biased 4-7% higher than collocated values from IMPROVE. Larger and more variable MRBs are present for comparisons of seasonal mean CASTNET and IMPROVE NO(3)(-) concentrations. For the 10 collocated sites considered in the current study, MRBs between site-specific seasonal mean CASTNET and IMPROVE NO(3)(-) concentrations range between 39% and -102%, and between 39% and -25% when marine sites are excluded. For the ensemble of all paired sites, the summertime MRB of 44% indicates that CASTNET NO(3)(-) concentrations are biased low, consistent with volatilization losses of particulate NO(3)(-) in the CASTNET sampler at high temperatures and low ambient NO(3)(-) concentrations that are typical during summer. These results should be considered prior to combining data from the CASTNET and IMPROVE (and possibly Chemical Speciation Network, CSN) networks for obtaining more aerially comprehensive representations of the atmosphere or for comparison with air quality model predictions. Biases in particulate SO(4)(2-) and NO(3)(-) concentrations from collocated CASTNET and IMPROVE sites are likely due in part to differences in the configurations of the samplers employed by the two networks (e.g., presence or absence of size-selective inlets). Published by Elsevier Ltd.