Obtaining high temporal and spatial resolution estimates of lockdown impacts on air quality using a generalized additive model that accounts for the influence of meteorology and air pollution transport

Abstract:

Changing emission patterns in response to COVID-19 represent an unprecedented natural experiment in atmospheric chemistry. While the pandemic has been global, changes in emissions have been highly variable both spatially and temporally due to differences in urbanization, lifestyle, and culture. Air quality is the result of the combined effect of local emissions, meteorology, and regional transport. To quantify the impact of the lockdown caused by reductions in human activities, it is necessary to account for differences in meteorology between the period of interest and reference times from previous years. Estimates can be further improved by distinguishing between the impacts of changing local and regional emissions as well as by combining an analysis of air pollution transport and variations from multiple sites in the same region.

A Generalized Additive Model (GAM) was developed to obtain a high time resolution estimate of the impacts of the lockdown at multiple sites in South Korea, China and the United States. The model combines temporal factors at multiple scales, meteorological observations from local sites, boundary layer height estimates from the ERA5 reanalysis, and transport regimes from FLEXPART Lagrangian particle simulations. By accounting for the impact of the boundary layer height and diurnal profiles of wind speed and direction, it is possible to estimate the diurnal profile of the emissions of air pollutants and to characterize their differences by region and by time periods. To increase the statistical robustness of the results, the model analyzes hourly in-situ data from multiple years such that tens to hundreds of thousands of data points are included for each site.

The model is able to distinguish the impacts on different air pollutants as well as the impacts on speciated components of fine particulate matter in order to provide insights for both sector-specific changes in emissions and changes in atmospheric formation and processing of pollutants.