Main content area

Relatively weak meteorological feedback effect on PM2.5 mass change in Winter 2017/18 in the Beijing area: Observational evidence and machine-learning estimations

Zhong, Junting, Zhang, Xiaoye, Wang, Yaqiang
The Science of the total environment 2019 v.664 pp. 140-147
aerosols, algorithms, artificial intelligence, particulates, pollution, pollution control, relative humidity, specific humidity, winter, China
Heavy aerosol pollution episodes (HPEs) in Beijing are worsened by the two-way feedback mechanism between unfavorable meteorological conditions and cumulative aerosols. In Winter 2017/18, mean PM2.5 mass concentration substantially decreased by 62% from 113 μg m−3 in Winter 2016/17 to 43 μg m−3. With reduced PM2.5 levels, the meteorological feedback on PM2.5 was relatively weak in Winter 2017/18. However, the weakening degree and its contributions to PM2.5 reduction are still uncertain. In this study, we investigated the change in the aerosol-induced modification of atmospheric stratification by combining PM2.5 data, radiosonde observations, and ERA-Interim reanalysis data, and then estimated the weakened meteorological feedback effect on PM2.5 change using machine learning. During polluted days, near-ground cooling bias, specific humidity (SH) increase, and relative humidity (RH) enhancement in Winter 2017/18 merely account for 38%, 65%, and 36% of the meteorological modification caused by aerosols in Winter 2016/17, respectively. Using machine learning algorithms with three most related variables, we found that during polluted days, the PM2.5 increase due to the meteorological feedback in Winter 2017/18 was merely 49% of that in Winter 2016/17. Effective pollution control and more favorable meteorological conditions have resulted in an additional benefit in PM2.5 reduction.