Relocation-associated discontinuities in Korean air-quality monitoring: a Bayesian structural time-series analysis of PM10 and PM2.5

Air quality monitoring networks are essential for tracking pollution trends and informing public health policies. However, station relocations can break the continuity of measurement records and affect trend estimation, exposure assessments, and epidemiologic studies. This study examines the impact of particulate matter (PM 10 and PM 2 . 5 ) monitoring station relocations in South Korea from 2010 to 2023. International comparison of station relocation policies revealed that the data management protocols related relocation are different, raising concerns about potential bias in analyses that pool records across a relocation. Records of relocation cases in South Korea were investigated to rectify the undocumented relocation cases using public records and historical satellite and street-level imagery. Using quasi-experimental Bayesian interrupted time series models, we estimated average pointwise changes in daily concentrations over symmetric 90-day windows on the log scale. Significant discontinuities occurred in 40.0 % (28 of 70) of analyzable PM 10 events and 41.3 % (19 of 46) of PM 2.5 events. Increases and decreases were similarly common, with median effects of -16.2 % and +22.1 % for PM 10 and -18.3 % and +19.5 % for PM 2.5 . Horizontal relocation distance did not explain effect magnitude. These results underscore the need for stringent relocation protocols considering data continuities and highlight the implications for air quality monitoring management. We recommend versioned identifiers, calibrated overlap monitoring, complete site metadata, and criteria for continuity decisions.