Investigating the nonlinear and non-stationary relationship between PM2.5 and air pollutants by wavelet signal analysis in central Taiwan

• Published in: Environmental geochemistry and health 2023-07, Vol.45 (7), p.5195-5211
• Main Authors: Feng, Shan-Non, Chen, Yi-Ho, Weng, Tzu-Han, Lin, Yuan-Chien
• Format: Article
• Language: English
• Subjects: air; Air pollution; Analysis; Carbon monoxide; climate; Correlation; Cross correlation; Data analysis; Data processing; Earth and Environmental Science; Environment; Environmental Chemistry; Environmental factors; Environmental Health; Environmental indicators; Geochemistry; human health; Industrial areas; Lag time; landscapes; Nitrogen compounds; Nitrogen oxides; Original Paper; Ozone; particle size; Particulate matter; Photochemicals; Pollutants; Pollution sources; Prediction models; Public Health; Signal analysis; Soil Science & Conservation; Spatiotemporal data; Sulfur; Sulfur dioxide; Sulphur; Sulphur dioxide; Taiwan; Terrestrial Pollution; wavelet; Wavelet analysis
• ISSN: 0269-4042; 1573-2983
• DOI: 10.1007/s10653-023-01560-5

In recent years, PM 2.5 has become a critical factor as an environmental indicator, causing severe air pollution that has negatively impacted nature and human health. This study used hourly data gathered in central Taiwan from 2015 to 2019 and applied spatiotemporal data analysis and wavelet analysis methods to investigate the cross-correlation between PM 2.5 and other air pollutants. Furthermore, it explored the correlation differences between adjacent stations after excluding major environmental factors such as climate and terrain. Wavelet coherence shows that PM 2.5 and air pollutants mostly have a significant correlation at the half-day and one-day frequencies, while the differences between PM 2.5 and PM 10 are only particle size; hence, not only is the correlation the most consistent among all air pollutants but also the lag time is the most negligible. Carbon monoxide (CO) is the primary source pollutant of PM 2.5 as it is also significantly correlated with PM 2.5 at most timescales. Sulfur dioxide (SO 2 ) and nitrogen oxide (NO x ) are related to the generation of secondary aerosols, which are important components of PM 2.5 ; therefore, the consistency of significant correlations improves as the timescale increases and the lag time becomes amplified. The pollution source mechanism of ozone (O 3 ) and PM 2.5 is not identical, so the correlation is lower than for other air pollutants; the lag time is also obviously influenced by the season changes that have significant fluctuations. At stations near the ocean such as Xianxi station and Shulu station, PM 2.5 and PM 10 have a higher correlation in the 24-h frequency, while the SO 2 and PM 2.5 at Sanyi station and Fengyuan station, which are close to industrial areas, have significant correlations in the 24-h frequency. This study hopes to help better understand the impact mechanisms behind different pollutants, and thus construct a better reference for establishing a complete air pollution prediction model in the future.