Spatial clustering of air pollutant trends and exposure burden across Romania (2013-2024).

Abstract

Air pollution remains a major environmental and public health challenge in Romania, where exceedances of European Union air quality standards persist. However, a national-scale assessment identifying regions of simultaneous multi-pollutant burden and coherent worsening trends has been lacking. Here we analyse hourly concentrations of PM2.5, PM10, NO2, and O3 from the Copernicus Atmosphere Monitoring Service (CAMS) reanalysis over 2013-2024, validated against 317 in situ stations-pollutant validation series. The Mann-Kendall test, Sen's slope estimator, and Local Indicators of Spatial Association (LISA) are used to quantify trends and detect spatially coherent clusters. CAMS demonstrates a solid correlation with the measured data for various pollutants: PM2.5 shows a strong correlation (coefficient of 0.83), followed closely by O3 with 0.82, and PM10 at 0.78. However, the performance for NO2 is noticeably weaker, with a correlation coefficient of only 0.48. Atmospheric pressure serves as the primary factor influencing pollutant levels, surpassing other variables. Following this, the proximity to urban centers and the height of the boundary layer also emerge as significant determinants in understanding air quality dynamics. Trend analysis reveals seasonally structured patterns: winter PM2.5 and PM10 increases of up to ＋10 μgm-3decade-1 in northern Transylvania with maxima recorded in the Cluj area of +10.01 μgm-3decade-1 (95% CI: 3.91-15.62 μgm-3decade-1) for PM2.5 and +10.22 μgm-3decade-1 (95% CI: 5.19-17.38 μgm-3decade-1) for PM10, and widespread summer O3 increases exceeding ＋15 μgm-3decade-1 in southeastern and western Romania. LISA clustering identifies High-High trend regions in northwestern Transylvania, the Bucharest-Giurgiu corridor, and northeastern Romania. Conversely, Low-Low clusters in southwestern Romania and the Galaţi area reflect lignite-fired power plant decommissioning and industrial restructuring, respectively. Combining a composite pollution-burden score with LISA clustering delineates priority areas where multi-pollutant exposure is persistently elevated and continues to increase, providing spatially explicit evidence for geographically targeted air-quality management.