- 1Department of Environmental Science, Kangwon National University, Chuncheon, Republic of Korea
- 2Department of Integrated Particulate Matter Management, Kangwon National University, Chuncheon, Republic of Korea
PM2.5 component concentrations are critical indicators for identifying emission sources and formation pathways in the atmosphere. In Korea, the Ministry of Environment operates regional Air Quality Research Center to measure PM2.5 components every hour. However, relying solely on measurements has temporal and spatial constraints. The WRF-CMAQ model enables us to predict PM2.5 component concentrations continuously at high spatiotemporal resolutions. Therefore, this study aims to improve the predictive performance of the WRF-CMAQ model for PM2.5 components using machine learning. The WRF-CMAQ simulation results, including PM2.5 components, meteorology, geography, and emissions, were used as input variables in the machine learning model. Observational data of PM2.5 components from Air Quality Research Centers were used as target variables to build the machine learning model. The study period was from January 1 to March 31, 2022, and the study area included 10 regions where the Air Quality Research Center operates. The machine learning model showed a correlation coefficient above 0.83 which is quite reasonable to use for PM2.5 component prediction. We analyzed cases where PM2.5 episodes occurred nationwide. The original CMAQ model results mainly showed high PM2.5 concentrations in some regions. In contrast, the machine learning-corrected CMAQ model results captured nationwide high PM2.5 levels well. The results can be useful for providing information on PM2.5 characteristics in other regions where the Air Quality Research Centers do not exist.
Thank you for National Institute of Environmental Research, Republic of Korea for providing the measurement data from the Air Quality Research Center. This work was supported by a grant from the National Institute of Environment Research (NIER), funded by the Ministry of Environment (MOE) of the Republic of Korea(NIER-2021-03-03-007). This work was and supported by the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT) (RS-2023-00219830).
How to cite: Kim, J., Choi, M., Jeon, Y., and Kwak, K.-H.: Improving PM2.5 component prediction in the WRF-CMAQ model using Machine Learning, 12th International Conference on Urban Climate, Rotterdam, The Netherlands, 7–11 Jul 2025, ICUC12-667, https://doi.org/10.5194/icuc12-667, 2025.
Supporters & sponsors
