Quantitative estimation of meteorological impacts and the COVID-19 lockdown reductions on NO2 and PM2.5 over the Beijing area using Generalized Additive Models (GAM)

Unprecedented travel restrictions due to the COVID-19 pandemic caused remarkable reductions in anthropogenic emissions, however, the Beijing area still experienced extreme haze pollution even under the strict COVID-19 controls. Generalized Additive Models (GAM) were developed with respect to inter-annual variations, seasonal cycles, holiday effects, diurnal profile, and the non-linear influences of meteorological factors to quantitatively differentiate the lockdown effects and meteorology impacts on concentrations of nitrogen dioxide (NO2) and fine particulate matters (PM2.5) at 34 sites in the Beijing area. The results revealed that lockdown measures caused large reductions while meteorology offset a large fraction of the decrease in surface concentrations. GAM estimates showed that in February, the control measures led to average NO2 reductions of 19 μg/m3 and average PM2.5 reductions of 12 μg/m3. At the same time, meteorology was estimated to contribute about 12 μg/m3 increase in NO2, thereby offsetting most of the reductions as well as an increase of 30 μg/m3 in PM2.5, thereby resulting in concentrations higher than the average PM2.5 concentrations during the lockdown. At the beginning of the lockdown period, the boundary layer height was the dominant factor contributing to a 17% increase in NO2 while humid condition was the dominant factor for PM2.5 concentrations leading to an increase of 65% relative to the baseline level. Estimated NO2 emissions declined by 42% at the start of the lockdown, after which the emissions gradually increased with the increase of traffic volumes. The diurnal patterns from the models showed that the peak of vehicular traffic occurred from about 12pm to 5pm daily during the strictest control periods. This study provides insights for quantifying the changes in air quality due to the lockdowns by accounting for meteorological variability and providing a reference in evaluating the effectiveness of control measures, thereby contributing to air quality mitigation policies. [Display omitted] •Generalized Additive Models quantified lockdown and meteorology impacts on PM2.5 and NO2.•NO2 decreased by 19 μg/m3 due to the lockdown which was offset by a 12 μg/m3 increase due to meteorology.•PM2.5 decreased by 12 μg/m3 due to the lockdown which was offset by a 30 μg/m3 increase due to meteorology.•Changes in vertical mixing had the largest impact on NO2, and changes in humidity for PM2.5.•Downtown Beijing experienced larger reductions