Slower than expected reduction in annual PM_2.5 in Xi'an revealed by machine learning-based meteorological...

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Data Science

Background:

• Air pollution control policy evaluation relies on pollutant trend analysis.
• Meteorological variations complicate multi-year air pollutant trend analysis.
• Decoupling meteorological effects is crucial for accurate emission trend assessment.

Purpose of the Study:

• To perform a trend analysis of hourly fine particulate matter (PM2.5) in Xi'an from 2015-2019.
• To decouple the influence of meteorological parameters on PM2.5 levels.
• To provide insights into real changes in PM2.5 due to emission strength or atmospheric chemistry.

Main Methods:

• Utilized a machine learning algorithm for trend analysis of PM2.5.
• Applied a novel meteorological normalization technique by using constant meteorological inputs over 5 years.
• Analyzed hourly PM2.5 data from an urban background site in Xi'an.

Main Results:

• Meteorological normalization revealed a decreasing trend of -3.3% year⁻¹ (-1.9 μg m⁻³ year⁻¹) in PM2.5.
• Direct PM2.5 observation showed a steeper decreasing trend of -4.4% year⁻¹.
• PM2.5 was found to be primarily associated with anthropogenic emissions, with seasonal variations in chemical processing.

Conclusions:

• The study highlights the importance of meteorological normalization for accurate air pollution trend analysis.
• Achieving the WHO guideline value of 5 μg m⁻³ for PM2.5 in Xi'an may take approximately 25 years at the observed normalized rate.
• Reducing anthropogenic secondary aerosol precursors (e.g., NOx, VOCs) is crucial for more effective particulate pollution reduction.