Estimation of Near-surface Ozone Concentration in the Beijing-Tianjin-Hebei Region Based on XGBoost-LME Model

High spatiotemporal resolution data on near-surface ozone concentration distribution is of great significance for monitoring and controlling atmospheric ozone pollution and improving the living environment. Using TROPOMI-L3 NO2，HCHO products，and ERA5-land high-resolution data as estimation variables，an XGBoost-LME model was constructed to estimate the near-surface ozone concentration in the Beijing-Tianjin-Hebei Region. The results showed that：① Through correlation analysis，surface 2 m temperature（T2M），2 m dewpoint temperature（D2M），surface solar radiation downwards（SSRD），tropospheric formaldehyde（HCHO），and tropospheric nitrogen dioxide（NO2）were important factors affecting the near-surface ozone concentration in the Beijing-Tianjin-Hebei Region. Among them，T2M，SSRD，and D2M had strong correlations，with correlation coefficients of 0.82，0.75，and 0.71，respectively. ② Compared with that of other models，the XGBoost-LME model had the best performance in terms of various indicators. The ten-fold cross-validation evaluation indicators R2，MAE，and RMSE were 0.951，9.27 μg·m−3，and 13.49 μg·m−3，respectively. At the same time，the model performed well at different time scales. ③ In terms of time，there was a significant seasonal difference in near-surface ozone concentration in the Beijing-Tianjin-Hebei Region in 2019，with the concentration changing in the order of summer > spring > autumn > winter. The monthly average ozone concentration in the region showed an inverted“V”trend，with a slight increase in September. The highest value occurred in July，whereas the lowest value occurred in December. In terms of spatial distribution，the near-surface ozone concentrations in the Beijing-Tianjin-Hebei Region during the months of February and March were generally at the same levels. In January，November，and December，there was a relatively insignificant trend of higher concentrations in the north and lower concentrations in the south. For the remaining months，the spatial distribution of near-surface ozone concentrations in this area predominantly exhibited a pattern of higher concentrations in the south and lower concentrations in the north. High-value areas were predominantly found in the plain regions of the southern part with lower altitudes，dense population，and higher industrial emissions；low-value areas，on the other hand，were primarily located in mountainous areas of the northern part with higher altitudes，sparse population，higher vegetation coverage，and lower industrial emissions. © 2024 Science Press. All rights reserved.