Detecting and Evaluating Dust‐Events in North China With Ground Air Quality Data

Abstract

Abstract We propose a dust‐event detection and tracking procedure based on air quality data from the ground monitoring network by detecting temporal and spatial change‐points in PM10 concentration. It supplements the existing remote sensing based approach with high temporal resolution and better weather adaptivity. Applications of the procedure on the labeled data showed its having high discriminating power for dust events, pollution events, and clean periods. Our study finds changing correlation patterns between PM10 and other air pollutants at the start of the dust events, which are utilized to enhance the discriminating power of the dust‐event detection procedure. The detection and tracking procedure allows the construction of transport networks of the dust‐events as well as the identification of the source regions and the transportation pattern, and assess the intensity and severity of the dust‐events in North China. Our analysis find the dust‐events contributed to 23.3%–34.6% for PM10 and 18.2–33.2% for PM2.5 in the source regions and 2.0%–7.3% and 0.8%–4.0%, respectively, in the down‐stream provinces in the spring season from 2015 to 2020.

Keywords

• air quality monitoring network data
• dust event detection and tracking
• statistical machine learning
• transportation network