Analysis of fine particle pollution data measured at 29 US diplomatic posts worldwide.

Fine particulate matter with an aerodynamic diameter of 2.5 μm or less (PM 2.5) is the most commonly measured ambient air pollutant worldwide on account of its adverse health effects. In 2008 the US Embassy in Beijing started monitoring PM 2.5 on the embassy premises to provide US citizens in the area with actionable health information related to ambient air pollution. By May 2019 the United States Department of State ambient air quality monitoring network had expanded to 43 cities in 27 countries. This provided a reliable source of near-real-time PM 2.5 data to fill data gaps in previously under-served countries. This manuscript describes data analyses of seasonal, diurnal and meteorological trends which can be used for air quality planning and awareness in 29 of the embassies host cities. People living, working, traveling or considering relocating to those areas can gain a better understanding of the air quality they are exposed to. Potential explanations are offered for sites that deviate from expected trends: Addis Ababa experiences high PM 2.5 during the rainy season probably due to widespread biomass burning and light winds; Manama has high PM 2.5 throughout the day on summer weekends because widespread air conditioner usage causes coal-fired power plants to operate at high capacity to meet the electricity demand; Pristina experienced very high PM 2.5 during winter weekend nights, which turned out to be very cold and stagnant periods and likely increased wood and coal combustion for home heating. Although Chinese sites consistently exceed the country's annual PM 2.5 standards, analysis of long term trends confirm steady improvements in air quality. These improvements cannot be attributed to favorable meteorological conditions alone and are likely due to reductions of emissions. Since random forest-based machine learning methods explained >90% of the variability in PM 2.5 concentrations at 21 locations, they can be used in conjunction with meteorological models for air quality forecasting. • PM 2.5 data from 29 US diplomatic missions across 16 countries have been analyzed. • High seasons, low pollution hours and weekday/weekend relationships are identified. • Air quality forecasting based on machine learning was successfully tested at 21 sites. • These relationships can be used for air quality planning and awareness. • Emissions in Chinese cities appear to have reduced over time. [ABSTRACT FROM AUTHOR]