Machine learning-based quantification and separation of emissions and meteorological effects on PM2.5 in Greater Bangkok

Abstract

This study presents the first-ever application of machine learning (ML)-based meteorological normalization and Shapley additive explanations (SHAP) analysis to quantify, separate, and understand the effect of meteorology on PM_2.5 over Greater Bangkok (GBK). Six ML models namely random forest (RF), adaptive boosting (ADB), gradient boosting (GB), extreme gradient boosting (XGB), light gradient boosting machine (LGBM), and cat boosting (CB) were used with meteorological factors, fire activity, land use, and socio-economic data as predictor variables. The LGBM outperformed other models achieving ρ = 0.9 (0.95), MBE = 0 (- 0.01), MAE = 5.5 (3.3) μg m^-3, and RMSE = 8.7 (4.9) μg m^-3 for hourly (daily) PM_2.5 prediction. LGBM was used for spatiotemporal PM_2.5 estimation, and meteorological normalization was applied to calculate PM_{2.5_emis} (emission-related PM_2.5) and PM_{2.5_met} (meteorology-related PM_2.5). Diurnal variation reveals higher PM_2.5 levels in the morning (08-10 LT) due to increased traffic emissions and thermal inversion and a decrease in PM_2.5 as the day progresses due to decreased emission and inversion dissipation. Monthly variation suggests higher PM_2.5 in winter (December and January) due to emissions and stagnant meteorological conditions. Negative PM_{2.5_met} during November, March, and April values show meteorology improves air quality, while positive values from December to February indicate stagnant winter conditions worsen it. During winter, PM_{2.5_emis} and PM_2.5 showed an increasing trend in 15.6% and 67.8% of the area while decreasing trends fell from 23.2 to 1.9%. In summer, the percentage of areas with an increasing trend rose from 18.7 to 34.6%, and decreasing areas fell from 12.6 to 6.5%. Increase in PM_2.5 despite decreasing emission over a larger area, indicating limited effectiveness of mitigation measures. Winter exhibits greater PM_2.5 variability due to episodic increases from changing meteorological conditions. In Bangkok and nearby areas, higher variability is mainly driven by meteorology, with more consistent emissions in Bangkok compared to rural areas affected by agricultural burning. PM_2.5 and PM_{2.5_emis} showed stronger persistence in winter than in summer, with weaker effects in Bangkok. Hurst exponent averages were 0.75, 0.76, and 0.72 for PM_2.5 and 0.79, 0.8, and 0.73 for PM_{2.5_emis} in dry, winter, and summer seasons, respectively. SHAP analysis suggested relative humidity, planetary boundary layer height, v wind, temperature, u wind, global radiation, and aerosol optical depth as the key variables affecting PM_2.5 with mean absolute SHAP values of 5.29, 4.79, 4.29, 3.68, 2.37, 2.22, and 2.03, respectively. Based on these findings, some policy recommendations have been proposed.

Keywords: Explainable machine learning; Himawari-8; Hurst exponent; Meteorological normalization; PM2.5 mapping; SHAP.

Fig. 1

(a) Thailand and Greater Bangkok (GBK), (b) PM_2.5 monitoring stations (PCD in Black and BMA in Blue) and AERONET stations (Red) in GBK, (c) Monthly variation of PM_2.5 over averaged over PCD stations and BMA stations. In (c) the axis labels (N, D, J, F, M, …..S, O) denote the months of the year from November to October.

Fig. 2

Hourly spatial distribution of (a) PM_2.5, (b) PM_{2.5_met}.

Fig. 3

Monthly spatial distribution of (a) PM_2.5, (b) PM_{2.5_met}.

Fig. 4

(a) Trend and p-value in (a) PM_2.5, (b) PM_{2.5_emis}.

Fig. 5

Coefficient of variation (COV) of (a) PM_2.5 (b) PM_{2.5_emis}.

Fig. 6

Hurst exponent (HE) for (a) PM_2.5 (b) PM_{2.5_emis}.

Fig. 7

(a) Feature importance as measured by SHAP analysis (b) dependence plots of SHAP values for the key influencing variables.