Optimization of WRF-Chem model results by XGBoost algorithm
LI Jiang-tao1, AN Xing-qin1, LI Qing-yong2, YU Hao-min2, WANG Wei3, ZHOU Xin-yuan2, WANG Chao1, CUI Meng4
1. Institute of Atmospheric Composition and Environmental Meteorology, Chinese Academy of Meteorological Sciences, Beijing 100081, China; 2. School of Computer and Information Technology, Beijing Jiaotong University, Beijing 100044, China; 3. China National Environmental Monitoring Centre, Beijing 100012, China; 4. 78127 Unit of the Chinese People's Liberation Army, Chengdu 610000, China
Abstract:The artificial intelligence algorithm XGBoost was used to build the statistical prediction algorithm model, combined with the atmospheric chemical model WRF-Chem, the simulation results of air pollutants in Beijing and the site monitoring data. The PM2.5 and O3 were optimized and simulated, and their characteristic contribution factors were analyzed. The results showed that the statistical forecast model of XGBoost can optimize the atmospheric pollutant concentration simulated by the atmospheric chemical model and reduce the simulation error. Moreover, the simulation concentration optimization presented the optimization characteristics of urban> suburbs> outer suburbs in Beijing area site, and the algorithm model optimized the O3 concentration to a higher degree, and the correlation coefficient was increased by 128% after optimization. In addition, the contribution analysis of feature elements showed that CO was an important feature variable which affects the optimization of O3 and the feature contribution scores of urban and suburban areas were as high as 1000 or more. Q2 (2m specific humidity near the ground) was an important meteorological characteristic variable that influenced the optimization of PM2.5 and the characteristic contribution scores of urban and suburban areas were 950 and 824, respectively.
李江涛, 安兴琴, 李清勇, 余浩敏, 汪巍, 周心源, 王超, 崔萌. 基于XGBoost算法的WRF-Chem模式优化模拟[J]. 中国环境科学, 2021, 41(12): 5457-5466.
LI Jiang-tao, AN Xing-qin, LI Qing-yong, YU Hao-min, WANG Wei, ZHOU Xin-yuan, WANG Chao, CUI Meng. Optimization of WRF-Chem model results by XGBoost algorithm. CHINA ENVIRONMENTAL SCIENCECE, 2021, 41(12): 5457-5466.
贺克斌,张强,同丹,等.中国中长期空气质量改善路径及健康效益[R]. 北京:能源基金会, 2020. He K B, Zhang Q, Tong D, et al. China's medium and long-term air quality improvement path and health benefits[R]. Beijing: Energy Foundation, 2020.
[2]
薛涛,刘俊,张强,等.2013~2017年中国PM2.5污染的快速改善及其健康效益[J]. 中国科学:地球科学, 2020,50(4):441-452. Xue T, Liu J, Zhang Q, et al. Rapid improvement of PM2.5 pollution and associated health benefits in China during 2013~2017[J]. Science China Earth Sciences, 2020,50(4):441-452.
[3]
Sun Z B, An X Q, Tao Y, et al. Assessment of population exposure to PM10 for respiratory disease in Lanzhou (China) and its health-related economic costs based on GIS[J]. Bmc Public Health, 2013,13:891- 900.
[4]
Han L, Sun Z B, Gong T Y, et al. Assessment of the short-term mortality effect of the national action plan on air pollution in Beijing, China[J]. Environmental Research Letters, 2020,15:3-13.
[5]
Vautard R, Builtjes P, Thunis P, et al. Evaluation and intercomparison of ozone and PM10simulations by several chemistry transport models over four European cities within the CityDelta project[J]. Atmospheric Environment, 2007,41:173-188.
[6]
Russell A, Dennis R. NARSTO critical review of photochemical models and modeling[J]. Atmospheric Environment, 2000,34:2283- 2324.
[7]
Peter L. Weather prediction by numerical process[M]. Cambridge: Cambridge University Press, 2007.
[8]
Stern R, Builtjes P, Schaap M, et al. A model inter-comparison study focussing on episodes with elevated PM10 concentrations[J]. Atmospheric Environment, 2008,42:4567-4588.
[9]
Zhao Y, Wang S X, Duan L, et al. Primary air pollutant emissions of coal-fired power plants in China: Current status and future prediction[J]. Atmospheric Environment, 2008,42:8442-8452.
[10]
Fu X, Wang S X, Zhao B, et al. Emission inventory of primary pollutants and chemical speciation in 2010 for the Yangtze River Delta region, China[J]. Atmospheric Environment, 2013,70:39-50.
[11]
Gu J J, Yang B, Brauer M, et al. Enhancing the evaluation and interpretability of data-driven air quality models[J]. Atmospheric Environment, 2021,246:118125-118134.
[12]
Jiang T T, Chen B, Nie Z, et al. Estimation of hourly full-coverage PM2.5 concentrations at 1-km resolution in China using a two-stage random forest model[J]. Atmospheric Research, 2021,248:105146- 105158.
[13]
Zhao C, Liu Z R, Wang Q, et al. High-resolution daily AOD estimated to full coverage using the random forest model approach in the Beijing-Tianjin-Hebei region[J]. Atmospheric Environment, 2019, 203:70-78.
[14]
康俊锋,黄烈星,张春艳,等.多机器学习模型下逐小时PM2.5预测及对比分析[J]. 中国环境科学, 2020,40(5):1895-1905. Kang J F, Huang L X, Zhang C Y, et al. Hourly PM2.5 prediction and its comparative analysis under multi-machine learning model[J]. China Environmental Science, 2020,40(5):1895-1905.
[15]
康俊锋,谭建林,方雷,等.XGBoost-LSTM变权组合模型支持下短期PM2.5浓度预测——以上海为例[J]. 中国环境科学, 2021,41(9): 4016-4025. Kang J F, Tan J L, Fang L, et al. Short-term PM2.5 concentration prediction based on XGBoost and LSTM variable weight combination model: a case study of Shanghai[J]. China Environmental Science, 2021,41(9):4016-4025.
[16]
王勇.基于多源数据和XGBoost算法的上海市能见度预测模型研究[D]. 上海:华东师范大学, 2019. Wang Y. Research on Shanghai visibility prediction model based on multi-source data and XGBoost algorithm[D]. Shanghai: East China Normal University, 2019.
[17]
侯俊雄,李琦,朱亚杰,等.融机器学习与WRF大气模式的PM2.5预报方法[J]. 测绘科学, 2018,43:114-120,141. Hou J X, Li Q, Zhu Y J, et al. PM2.5 forecasting method based on machine learning and WRF hybrid model[J]. Science of Surveying and Mapping, 2018,43:114-120,141.
[18]
Chen T, Guestrin C. XGBoost: A scalable tree boosting system[C]. Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, 2016:785-794.
[19]
叶倩怡,饶泓,姬名书.基于Xgboost的商业销售预测[J]. 南昌大学学报(理科版), 2017,41(3):275-281. Ye Q Y, Rao H, Ji M S. Sales prediction of stores based on Xgboost algorithm[J]. Journal of Nanchang University (Natural Science), 2017,41(3):275-281.
[20]
张钰,陈珺,王晓峰,等.Xgboost在滚动轴承故障诊断中的应用[J]. 噪声与振动控制, 2017,37(4):166-170. Zhang Y, Chen J, Wang X F, et al. Application of Xgboost to fault diagnosis of rolling bearings[J]. Noise and Vibration contral, 2017, 37(4):166-170.
[21]
赵俊日,肖昕,吴涛,等.空气质量数值预报优化方法研究[J]. 中国环境科学2018,38(6):2047-2054. Zhao J R, Xiao X, Wu T, et al. A revised approach to air quality forecast based on Models-3/CMAQ[J]. China Environmental Science, 2018,38(6):2047-2054.