Publication

A Three-Year Action Plan was implemented in 2018 to improve air quality in the Fenwei Plain. Yet, the effectiveness of this action plan is not clear, which is of great concern to the public and policymakers. Meteorological differences between years may lead us to misestimate the effect of pollutant reduction policies. Here we quantitatively assess the effectiveness of the Three-Year Action Plan in the Fenwei Plain using a meteorological normalization technique based on a machine learning model. To reduce the uncertainty of the model itself, we performed the optimization of hyperparameters and the selection of the optimal model. Ultimately, our study results showed that after removing the impact of meteorology on air quality, the true concentrations of PM2.5, PM10, O-3, SO2, NO2, and CO in the Fenwei Plain decreased by 29, 29, 2, 61, 20, and 42% from 2017 to 2020, respectively. The normalized concentrations of all pollutants decreased less than the observed concentrations, implying that if we focus only on the observed data, we could overestimate the effectiveness of the Three-Year Action Plan in the Fenwei Plain. Besides, combining the different decreases in concentrations of six pollutants and the emission characteristics of different pollution sources, we can conclude the improvement in air quality in the Fenwei Plain during 2017-2020 was mainly driven by the residential and industrial sectors. Our work could help the public and policymakers review the effectiveness of the Three Action Plan in the Fenwei Plain and provided a useful reference for the evaluation and formulation of pollutant emission reduction policies.