CMA-ChemRA Heavy Pollution Evaluation: Fidelity of PM_2.5 in 18 Events from a Weakly Coupled Chemical–Weather Reanalysis

Current global reanalyses (e.g., CAMSRA, MERRA-2) operate at coarse resolutions (> 50 km) and primarily assimilate satellite-based AOD rather than in-situ surface PM_2.5 observations, severely limiting their ability to capture the fine-scale evolution of heavy pollution over China. CMA-ChemRA, the nation’s first weakly coupled chemical-weather reanalysis with a horizontal resolution of 15 km and an hourly temporal resolution, overcomes this by simultaneously assimilating routine meteorological and ground-based composition data. Rigorously validating its fidelity is critical to establishing its credibility for operational early warning systems and supporting effective regional air quality mitigation policies. We systematically evaluate CMA-ChemRA against multiple independent observations, including CNEMC networks, U.S. Embassy monitors, AERONET sunphotometer, CALIPSO lidar, and the CAQRA reanalysis. The assessment spans 18 heavy pollution events (daily PM_2.5 > 150μg m⁻³, affecting≥3 adjacent provinces for≥2 days) from 2013 to 2023. CMA-ChemRA shows robust overall agreement to observations, with a mean correlation coefficient of 0.80. Biases exhibit concentration dependence: +3.27% (< 150μg m⁻³), −1.0% (150–300μg m⁻³), and increasingly negative beyond 300μg m⁻³. Spatially, positive biases dominate the Beijing–Tianjin–Hebei and Yangtze River Delta, while negative biases prevail in the Pearl River Delta, alongside nighttime overestimation. Vertically, aerosol layer heights are reasonably reproduced, but extinction coefficients are systematically underestimated (median CMA/CALIPSO ratio≈0.65) due to the absence of AOD assimilation. A December 2015 case study confirms its ability to capture PM_2.5 evolution and circulation, despite notable model overestimation of nitrate and underestimation of sulfate. Despite remaining deficiencies in chemical speciation and vertical intensity, CMA-ChemRA reliably reconstructs the three-dimensional spatiotemporal evolution and circulation drivers of major heavy pollution episodes. This dataset provides valuable support for PM_2.5 transport flux analysis and the development of targeted emission reduction policies.