Evaluating consistency of PM_2.5 chemical composition measurement in the Chemical Speciation Network (CSN) over time

The U.S. Environmental Protection Agency Chemical Speciation Network (CSN) has been monitoring PM_2.5 chemical composition in urban areas since 2000. In 2015, the CSN national contract transition initiated a series of network-wide changes in instrumentation, laboratory techniques, and data processing, extending through 2018. This study assesses how these changes impacted CSN's chemical composition data and identifies species that show consistent measurements through the transition. This analysis uses PM_2.5 speciation data from 90 CSN sites with comprehensive data coverage from 2002 to 2022. We calculated network-wide statistical summaries of all reported species, including elements, ions and carbon, to evaluate the consistency of their statistical distributions and interannual variability. Major PM_2.5 components and crustal elements (e.g. S, Ca, Fe, K, Si, OC, EC, NH₄⁺, NO₃^-) are consistently measured well above the method detection limit (MDL) throughout CSN's history, with network medians showing stable trends across the transition. However, several trace elements are infrequently measured above their MDL and exhibit abrupt concentration changes at the time of transition. Comparisons between SO₄^2- and K⁺ against their paired elements S and K indicate that the ions were likely underestimated between 2016 and 2018. Instrument transitions also introduced biases in carbon measurements, including underestimation of OC and EC in 2016-2017 and a high bias in EC after 2018. Additionally, difference in handling data below zero over time has introduced artificial step changes in concentration time series, potentially affecting the visual interpretation and trend estimates for many species. Our findings emphasize the importance of considering differences in measurement techniques and detection limits when analyzing atmospheric data. By accounting for these factors, we achieve a more accurate assessment of atmospheric trends and changes in composition over time. This paper also provides guidance on effectively utilizing CSN speciation data to address specific research needs and interests.Implications: This study highlights the impact of changes in instrumentation, laboratory techniques, and data processing on the U.S. EPA's Chemical Speciation Network (CSN) PM2.5 data, particularly for trace elements and species that frequently measured below their detection limits. Our findings underscore the need for careful consideration of methodological shifts and detection limits when interpreting long-term atmospheric data. This research not only contributes to improving the accuracy of atmospheric composition assessments but also offers practical guidance for researchers in utilizing CSN data for future studies on air pollution and its environmental impacts.