基于可解释机器学习的交通CO2和O3的驱动机制

张煌帆; 谢郑一; 温晓娟; 王硕; 翁大维; 徐艺诺; 胡喜生; 张兰怡

PDF(1669 KB)

中国环境科学 ›› 2025, Vol. 45 ›› Issue (11) : 6058-6067.

大气污染与控制

基于可解释机器学习的交通CO₂和O₃的驱动机制

张煌帆, 谢郑一, 温晓娟, 王硕, 翁大维, 徐艺诺, 胡喜生, 张兰怡

作者信息 +

Driving mechanisms of traffic-related CO₂ and O₃ based on interpretable machine learning

ZHANG Huang-fan, XIE Zheng-yi, WEN Xiao-juan, WANG Shuo, WENG Da-wei, XU Yi-nuo, HU Xi-sheng, ZHANG Lan-yi

Author information +

文章历史 +

摘要

以华东地区为研究区域,围绕交通产生的CO₂以及其对O₃的贡献(下文简称“O₃”)构建15个驱动因素指标体系.首先,采用Lasso回归模型筛选关键变量;其次,引入RF(随机森林)与XGBoost(极端梯度提升)模型开展特征重要性分析,并结合SHAP方法对各驱动因素的贡献度进行可视化解释,构建了系统的驱动因素识别与排序分析框架.研究表明,华东地区交通CO₂排放总量持续增长,至2022年,上海、江苏与山东三省合计占比达62.18%; O₃浓度高值区域主要集中于东北部.其中安徽省O₃浓度增幅高达92.25%.在模型性能方面,RF模型在变量解释力和稳定性上优于XGBoost.其中,其CO₂排放与O₃浓度的R²均大于0.8,MSE均为0.001,MAE均小于0.1.SHAP分析结果显示,交通CO₂排放的主要驱动因素包括人口密度、货运周转量、第三产业产值和城镇化水平;而O₃浓度则受人口数量、交通运输业产值、载货汽车数量和城市道路长度等因素影响.

Abstract

This study focuses on East China, establishing a 15-indicator system to investigate the driving mechanisms of traffic-related CO₂ emissions and their contributions to ozone (O₃) concentrations. First, the Lasso regression model is employed to screen key variables. Then, feature importance is assessed using Random Forest (RF) and Extreme Gradient Boosting (XGBoost) models, with the SHAP method used to visualize the marginal contributions of each factor. A systematic framework for identifying and ranking driving factors is thus constructed. Results indicate that traffic-related CO₂ emissions in East China have shown a continuous upward trend, with Shanghai, Jiangsu, and Shandong together accounting for 62.18% of total emissions by 2022. High O₃ concentration zones are primarily located in the northeastern subregion, with Anhui Province experiencing a remarkable increase of 92.25%. In terms of model performance, the RF model outperforms XGBoost in both explanatory power and stability, achieving R² values above 0.8 for both CO₂ and O₃, with MSE values of 0.001 and MAE values below 0.1. SHAP analysis reveals that the main drivers of CO₂ emissions include population density, freight turnover, tertiary industry output, and urbanization level; while O₃ concentrations are primarily influenced by total population, transport sector output, the number of freight vehicles, and urban road length.

导出引用

张煌帆, 谢郑一, 温晓娟, 王硕, 翁大维, 徐艺诺, 胡喜生, 张兰怡. 基于可解释机器学习的交通CO₂和O₃的驱动机制[J]. 中国环境科学. 2025, 45(11): 6058-6067

ZHANG Huang-fan, XIE Zheng-yi, WEN Xiao-juan, WANG Shuo, WENG Da-wei, XU Yi-nuo, HU Xi-sheng, ZHANG Lan-yi. Driving mechanisms of traffic-related CO₂ and O₃ based on interpretable machine learning[J]. China Environmental Science. 2025, 45(11): 6058-6067

中图分类号： X734 U491.9

参考文献

[1] He H, Lu D, Zhao H, et al. Characterizing CO₂ and NO_x emission of vehicles crossing toll stations in highway [J]. Transportation Research Part D: Transport and Environment, 2024,126:104024.
[2] Zhang S, Xiong Y, Liang X, et al. Spatial and cross-sectoral transfer of air pollutant emissions from the fleet electrification in China by 2030 [J]. Environmental Science & Technology, 2023,57(50):21249- 21259.
[3] Weng D, Zhang H, Wen X, et al. Co-benefits of carbon and pollutant emission reduction in urban transport: Sustainable pathways and economic efficiency [J]. Urban Climate, 2025,60:102348.
[4] 张兰怡,徐艺诺,王硕,等.碳达峰背景下沿海城市群道路交通节能减排潜力 [J]. 交通运输系统工程与信息, 2024,24(5):45-55. Zhang L, Xu Y, Wang S, et al. Energy saving and emission reduction potential of road traffic in coastal urban agglomerations under background of carbon peak [J]. Journal of Transportation Systems Engineering and Information Technology, 2024,24(5):45-55.
[5] Huang Y, Wang Y, Peng J, et al. Can China achieve its 2030 and 2060 CO₂ commitments? Scenario analysis based on the integration of LEAP model with LMDI decomposition [J]. Science of the Total Environment, 2023,888:164151.
[6] Shangguan Z. Investigating the spatiotemporal dynamics and interplay mechanisms between population urbanization and carbon dioxide emissions in China [J]. Journal of Geochemical Exploration, 2024,266: 107571-107571.
[7] 王硕,徐艺诺,翁大维,等.中国道路交通碳排放驱动因素及碳达峰 [J]. 生态学报, 2025,45(3):1315-1327. Wang S, Xu Y, Weng D, et al. Driving factors and peak analysis of road traffic carbon emissions in China [J]. Acta Ecologica Sinica, 2025,45(3):1315-1327.
[8] 翁大维,谢郑一,温晓娟,等.福建省城市群道路交通碳排放时空异质性 [J]. 环境科学, 2025,46(8):4826-4838. Weng D, Xie Z, Wen X, et al. Spatial and temporal heterogeneity of road transportation carbon emissions in Urban Agglomerations of Fujian Province [J]. Environmental Science, 2025,46(8):4826-4838.
[9] 张兰怡,翁大维,王硕,等.赣闽粤地区交通碳排放时空演化及驱动力 [J]. 环境科学, 2025,46(5):2886-2896. Zhang L, Weng D, Wang S, et al. Spatio-temporal evolution and driving forces of transportation carbon emissions in Jiangxi, Fujian, and Guangdong regions [J]. Environmental Science, 2025,46(5): 2886-2896.
[10] 张兰怡,徐艺诺,翁大维,等.经济高质量发展背景下中国省域物流业碳排放时空分异 [J]. 环境科学, 2024,45(9):5086-5096. Zhang L, Xu Y, Weng D, et al. Spatiotemporal differentiation of carbon emissions from logistics industry at provincial scale in China under the background of high-quality economic development [J]. Environmental Science, 2024,45(9):5086-5096.
[11] Zhang L, Weng D, Xu Y, et al. Spatio-temporal evolution characteristics of carbon emissions from road transportation in the mainland of China from 2006 to 2021 [J]. Science of the Total Environment, 2024,917:170430.
[12] 戈秋虞,张兰怡,邱荣祖,等.基于GDIM法的福建省交通碳排放驱动因素分析 [J]. 福建农林大学学报(自然科学版), 2022,51(1):131- 136. Ge Q, Zhang L, Qiu R, et al. Analysis on driving factors of transportation carbon emissions in Fujian Province based on GDIM [J]. Journal of Fujian Agriculture and Forestry University (Natural Science Edition), 2022,51(1):131-136.
[13] He Z, Liu P, Zhao X, et al. Responses of surface O₃ and PM_2.5 trends to changes of anthropogenic emissions in summer over Beijing during 2014~2019: A study based on multiple linear regression and WRF- Chem [J]. Science of the Total Environment, 2022,807(P2):150792- 150792.
[14] Shi Z, Huang L, Li J, et al. Sensitivity analysis of the surface ozone and fine particulate matter to meteorological parameters in China [J]. Atmospheric Chemistry and Physics, 2020,20(21):13455-13466.
[15] 陈坚,陈嘉果,陈琦,等.山地城市道路交通碳排放影响因素非关系模型 [J]. 交通运输系统工程与信息, 2024,24(5):237-245. Chen J, Chen J, Chen Q, et al. Nonlinear relationship model of factors influencing carbon emissions in mountainous [J]. Journal of Transportation Systems Engineering and Information, 2024,24(5): 237-245.
[16] 陈镇,刘润,罗征,等.基于机器学习的珠三角秋季臭氧浓度预测 [J]. 环境科学, 2024,45(1):1-7. Chen Z, Liu R, Luo Z, et al. Prediction of autumn ozone concentration in the Pearl River Delta based on machine learning [J]. Environmental Science, 2024,45(1):1-7.
[17] Vigneshkumar B, Vinothkumar B, Jia C. Importance of ozone precursors information in modelling urban surface ozone variability using machine learning algorithm [J]. Scientific Reports, 2022,12(1): 5646-5646.
[18] Ainsworth A, Yendrek R, Sitch S, et al. The effects of tropospheric Ozone on net primary productivity and implications for climate change [J]. Annual Review of Plant Biology, 2012,63(1):637-661.
[19] Lombardozzi D, Levis S, Bonan G, et al. The influence of chronic Ozone exposure on global carbon and water cycles [J]. Journal of Climate, 2014,28(1):292-305.
[20] 曾晓莹,邱荣祖,林丹婷,等.中国交通碳排放及影响因素时空异质性 [J]. 中国环境科学, 2020,40(10):4304-4313. Zeng X, Qiu R, Lin D, et al. Spatio-temporal heterogeneity of transportation carbon emissions and its influencing factors in China [J]. China Environmental Science, 2020,40(10):4304-4313.
[21] 易睿,王亚林,张殷俊,等.长江三角洲地区城市臭氧污染特征与影响因素分析 [J]. 环境科学学报, 2015,35(8):2370-2377. Yi R, Wang Y, Zhang Y, et al. Pollution characteristics and influence factors of Ozone in Yangtze River Delta [J]. Journal of Environmental Sciences, 2015,35(8):2370-2377.
[22] 陈玉红.中国臭氧污染时空分布特征及影响因素研究 [D]. 大连:大连理工大学, 2021. Chen Y. Research on the spatiotemporal distribution characteristics and influencing factors of ozone pollution in China [D]. Dalian: Dalian University of Technology, 2021.
[23] Wu R, Xie S. Spatial distribution of ozone formation in China derived from emissions of speciated volatile organic compounds [J]. Environmental Science & Technology, 2017,51(5):2574-2583.
[24] Ding S, He J, Liu D. Investigating the biophysical and socioeconomic determinants of China tropospheric O₃ pollution based on a multilevel analysis approach [J]. Environmental Geochemistry and Health, 2021, 43(8):2835-2849.
[25] 杨寒雨,赵晓永,王磊.数据归一化方法综述 [J]. 计算机工程与应用, 2023,59(3):13-22. Yang H, Zhao X, Wang L, et al. Review of data normalization methods [J]. Computer Engineering and Applications, 2023,59(3): 13-22.
[26] Shan Y, Guan D, Zheng H, et al. China CO₂ emission accounts 1997~ 2015 [J]. Scientific Data 5, 2018:17021.
[27] Liu M, Wen Y, Wu X, et al. Province-level decarbonization potentials for China’s road transportation sector [J]. Environmental Science & Technology, 2024,58(41):18213-18221.
[28] Zhang L, Weng D, Wen X, et al. Unveiling intra-provincial cities carbon gap: Exploring transportation emissions in ecological Vanguard [J]. Transportation Research Part D: Transport and Environment, 2025,144,104799.
[29] 杨晓彤,康平,王安怡,等.基于随机森林模型的四川盆地臭氧污染预测 [J]. 环境科学, 2024,45(5):2507-2515. Yang X, Kang, P, Wang A, et al. Prediction of Ozone pollution in Sichuan Basin based on random forest model [J]. Environmental Science, 2024,45(5):2507-2515.
[30] 卢鋆镆,曾穗平,曾坚,等.基于随机森林的高分辨率PM_2.5浓度时空变化模拟—以中原城市群核心区为例 [J]. 中国环境科学, 2023, 43(7):3299-3311. Lu Y, Zeng H, Zeng J, et al. High resolution simulation of temporal and spatial variation of PM_2.5concentration based on random forest — A case study of central plains urban agglomeration core area [J]. China Environmental Science, 2023,43(7):3299-3311.
[31] 赵楠,卢毅敏.基于XGBoost算法的近地面臭氧浓度遥感估算 [J]. Journal of Environmental Sciences, 2022,42(5):95-108. Zhao N, Lu Y. Remote-sensing estimation of near-surface Ozone concentration based on XGBoost [J]. 环境科学学报, 2022,42(5): 95-108.
[32] 陈启香,吕斌,李显林.站域建成环境对出租车-地铁组合出行的非线性影响 [J]. 交通运输工程学报, 2024,24(5):285-300. Chen Q, Lu Q, Li X. Nonlinear effect of station-area built environment on taxi-metro combined travel [J]. Journal of Traffic and Transportation Engineering, 2024,24(5):285-300.
[33] 阎莹,王玉莹,周旋,等.基于混合模型的自动驾驶车辆事故严重程度影响因素分析 [J]. 交通运输工程学报, 2025,25(1):184-196. Yan Y, Wang Y, Zhou X, et al. Analysis of autonomous vehicle accident severity factors based on hybrid model [J]. Journal of Traffic and Transportation Engineering, 2025,25(1):184-196.
[34] 宋永朝,舒秦,金程容,等.中国区域交通碳排放预测与碳达峰路径规划 [J]. 环境科学, 2025,46(4):1995-2008. Song Y, Shu Q, Jin C, et al. Regional transport carbon emission forecasting and peak carbon pathway planning in China [J]. Environmental Science, 2025,46(4):1995-2008.
[35] Silva R, Adelman Z, Fry M, et al. The impact of individual anthropogenic emissions sectors on the global burden of human mortality due to ambient air pollution [J]. Environmental health perspectives, 2016,124(11):1776-1784.

基金

福建省自然科学基金项目(2023J01475);福建省社科规划项目(FJ2022B065);福建农林大学“杰出青年科研人才”计划项目(社会科学类)(xjq2020S4);福建农林大学科技创新专项(KFb22101XA);国家级大学生创新创业训练计划项目(202410389030,202410389034)