Emulation of complex air quality models - Results#
An interactive plot visualising how emission changes impact air pollution exposure and the associated disease burden in China. This work highlights the value of machine learning emulators in air quality research.
For more information, see here.
Highlights
We created accurate and fast machine learning emulators to predict long−term air quality and health impacts from emission changes in China.
The emulators (Gaussian process) predicted 99.9% of the variance in PM2.5 (fine particulate matter, annual-mean) and O3 (ozone, maximum 6−monthly−mean daily−maximum 8−hour) concentrations.
Reducing emissions linearly reduces PM2.5 exposure and the associated disease burden (MORT, premature mortality per year), with larger fractional reductions in exposure.
In 2015, we estimate that PM2.5 exposure was 47.4 μg m−3 and O3 exposure was 43.8 ppb, associated with 2,189,700 (95% uncertainty interval, 95UI: 1,948,000−2,427,300) premature deaths per year, primarily from PM2.5 exposure (98%).
Removing emissions from one sector reduces PM2.5 exposure by up to 28% (largest for industry) reaching the National Air Quality Target (NAQT, 35 μg m−3).
Removing emissions from two sectors reduces PM2.5 exposure by up to 48% (largest for industry and residential) reaching the World Health Organization (WHO) Interim Target 2 (25 μg m−3).
Removing emissions from five key sectors in China does not attain the WHO Annual Guideline due to remaining pollution from other sources.
Removing emissions from the five sectors in China reduces national PM2.5 exposure by 57% (to 19.8 μg m−3), avoiding 858,800 (95UI: 774,900−945,400) premature deaths.
This does not enable the attainment of the WHO Annual Guideline (5 μg m−3) due to remaining air pollution from other anthropogenic emissions inside China, anthropogenic emissions outside China, and natural emission sources.
Move the sliders to explore the results.