Importance of secondary decomposition in the accurate prediction of daily-scale ozone pollution by machine learning

Xinyue Du; Zibing Yuan; Daojian Huang; Wei Ma; Jun Yang; Jianbin Mo

doi:10.1016/j.scitotenv.2023.166963

Importance of secondary decomposition in the accurate prediction of daily-scale ozone pollution by machine learning

Sci Total Environ. 2023 Dec 15:904:166963. doi: 10.1016/j.scitotenv.2023.166963. Epub 2023 Sep 9.

Authors

Xinyue Du¹, Zibing Yuan², Daojian Huang³, Wei Ma¹, Jun Yang¹, Jianbin Mo¹

Affiliations

¹ School of Environment and Energy, South China University of Technology, Guangzhou 510006, China.
² School of Environment and Energy, South China University of Technology, Guangzhou 510006, China. Electronic address: zibing@scut.edu.cn.
³ South China Institute of Environmental Sciences, Ministry of Ecology and Environment of China, Guangzhou 510655, China. Electronic address: huangdaojian@scies.org.

PMID: 37696411
DOI: 10.1016/j.scitotenv.2023.166963

Abstract

Machine learning (ML) models have been proven as a reliable tool in predicting ambient pollution concentrations at various places in the world. However, their performance in predicting the maximum daily 8-h averaged ozone (MDA8 O₃), the metric often used for O₃ pollution assessment and management, is relatively poorer. This is largely resulted from more irregular data fluctuations of the MDA8 O₃ levels governed collectively by the synoptic condition, local photochemistry, and long-range transport. In order to improve the prediction accuracy of MDA8 O₃, this study developed a secondary decomposition ML model framework which coupled the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) as the primary decomposition, the variational mode decomposition (VMD) as secondary decomposition, and the gate recurrent unit (GRU) ML model. By applying this secondary decomposition model framework on MDA8 O₃ prediction for the first time, we showed that the prediction accuracy of MDA8 O₃ is largely improved from R² of 0.46 and RMSE of 30.4 μg/m³ for GRU without decomposition to R² of 0.91 and RMSE of 12.6 μg/m³ over the Pearl River Delta of China. We also found that the prediction accuracy rate of O₃ pollution non-attainments, an essential indicator for initiating contingency O₃ pollution control, improved greatly from 14.9 % for GRU without decomposition to 72.5 %. The performance of O₃ pollution non-attainment prediction is relatively higher in southwestern PRD, which is mainly due to greater number and severity of O₃ non-attainments in southwestern cities located downwind of the emission hotspot area at central PRD. This study underscored the importance of secondary decomposition in accurately predicting daily-scale O₃ concentration and non-attainments over the PRD, which can be extended to other photochemically active region worldwide to improve their O₃ prediction accuracy and assist in O₃ contingency control.

Keywords: MDA8; Machine learning; Ozone; Pearl River Delta; Secondary decomposition.