Inversion of PM2.5 in the Beijing-Tianjin-Hebei region using a random forest model based on GF-5B satellite
HANG Rui-jie1,2, ZHOU Chun-yan2, CHEN Hui2, ZHOU Wei2, XIE Hui-zhen2, CHEN Rui-zhi1,2, WANG Zhong-ting2
1. Chinese Research Academy of Environmental Sciences, Beijing 100012, China; 2. Satellite Application Center for Ecology and Environment, Ministry of Ecology and Environment, Beijing 100094, China
摘要 This study utilizes aerosol optical depth (AOD) data derived from the domestic GF-5B DPC payload, ERA- 5meteorological data, NDVI data, DEM data, nighttime light data, and ground-based PM2.5 measurements. Based on a random forest model, an AOD-PM2.5 inversion model is developed to estimate the PM2.5 concentration in the Beijing-Tianjin-Hebei region for the year 2022. The results showed that the model's coefficient of determination (R2) for the entire year was 0.80, with a root mean square error (RMSE) of 11.43 μg/m3. For the seasonal models, the R2 values were 0.84, 0.71, 0.88, and 0.87 for spring, summer, autumn, and winter, respectively, with corresponding RMSE values of 9.50, 7.37, 9.71, and 11.32μg/m3. Both the annual and seasonal models demonstrated good simulation accuracy. The PM2.5 concentrations in the Beijing-Tianjin-Hebei region in 2022 exhibited significant seasonal variation, with the highest concentrations in winter and the lowest in summer. The estimated annual average PM2.5 concentration was 30.86μg/m3. During the summer and autumn, AOD was the primary factor influencing PM2.5 concentrations, while in spring and winter, the boundary layer height and relative humidity were the main influencing factors.
Abstract:This study utilizes aerosol optical depth (AOD) data derived from the domestic GF-5B DPC payload, ERA- 5meteorological data, NDVI data, DEM data, nighttime light data, and ground-based PM2.5 measurements. Based on a random forest model, an AOD-PM2.5 inversion model is developed to estimate the PM2.5 concentration in the Beijing-Tianjin-Hebei region for the year 2022. The results showed that the model's coefficient of determination (R2) for the entire year was 0.80, with a root mean square error (RMSE) of 11.43 μg/m3. For the seasonal models, the R2 values were 0.84, 0.71, 0.88, and 0.87 for spring, summer, autumn, and winter, respectively, with corresponding RMSE values of 9.50, 7.37, 9.71, and 11.32μg/m3. Both the annual and seasonal models demonstrated good simulation accuracy. The PM2.5 concentrations in the Beijing-Tianjin-Hebei region in 2022 exhibited significant seasonal variation, with the highest concentrations in winter and the lowest in summer. The estimated annual average PM2.5 concentration was 30.86μg/m3. During the summer and autumn, AOD was the primary factor influencing PM2.5 concentrations, while in spring and winter, the boundary layer height and relative humidity were the main influencing factors.
张瑞杰, 周春艳, 陈辉, 周伟, 谢慧真, 陈睿智, 王中挺. 基于GF-5B卫星的随机森林模型反演京津冀地区PM2.5[J]. 中国环境科学, 2024, 44(11): 5961-5970.
HANG Rui-jie, ZHOU Chun-yan, CHEN Hui, ZHOU Wei, XIE Hui-zhen, CHEN Rui-zhi, WANG Zhong-ting. Inversion of PM2.5 in the Beijing-Tianjin-Hebei region using a random forest model based on GF-5B satellite. CHINA ENVIRONMENTAL SCIENCECE, 2024, 44(11): 5961-5970.
[1] 郑 俊.基于MODIS反演的西安市气溶胶光学厚度与颗粒物浓度研究 [D]. 西安: 长安大学, 2019. Zheng J. Study on aerosol optical depth and particulate matter concentration in Xi'an based on MODIS retrieval [D]. Xi'an: Chang'an University, 2019. [2] Huang K, Zhuang G, Li J, et al. Mixing of Asian dust with pollution aerosol and the transformation of aerosol components during the dust storm over China in spring 2007 [Z]. Journal of Geophysical Research: Atmospheres, VOL.115,D00K13. [3] Zhang X, Wang J, Wang Y, et al. Changes in chemical components of aerosol particles in different haze regions in China from 2006 to 2013 and contribution of meteorological factors [J]. Atmos. Chem. Phys., 15,12935-12952. [4] Li L, Dubovik O, Derimian Y, et al. Retrieval of aerosol components directly from satellite and ground-based measurements [J]. Atmos. Chem. Phys., 19,13409-13443. [5] 刘泽阳.基于地基观测的气溶胶光学特性研究 [D]. 合肥:中国科学技术大学,2020. [6] Liu Z Y. Aerosol optical properties study based on ground observation [D]. Hefei: University of Science and Technology of China, 2020. [7] 王 桥.PM2.5卫星遥感技术及其应用 [M]. 北京:中国环境出版社, 2017:27-28. Wang Q. PM2.5 monitoring technology and application using satellite remote sensing [M]. Beijing: China Environmental Publishing House, 2017:27-28. [8] Geng G. Estimating long-term PM2.5 concentrations in China using satellite-based aerosol optical depth and a chemical transport model [J]. Remote Sensing of Environment, 2015,166:262-270. [9] Lee H J, Coull B A, Bell M L, et al. Use of satellite-based aerosol optical depth and spatial clustering to predict ambient PM2.5 concentrations [J]. Remote Sensing of Environment, 2012,118:8-15. [10] Chu D A, Kaufman Y J, Zibordi G, et al. Global monitoring of air pollution over land from the Earth Observing System-Terra Moderate Resolution Imaging Spectroradiometer (MODIS) [J]. Journal of Geophysical Research: Atmospheres, 2003,108. [11] Benas N, Beloconi A, Chrysoulakis N. Estimation of urban PM10concentration, based on MODIS and MERIS/AATSR synergistic observations [J]. Atmospheric Environment, 2013,79:448-454. [12] Strawa A W, Chatfield R B, Legg M, et al. Improving PM2.5 Retrievals in the San Joaquin Valley Using A-Train Multi-Satellite Observations [J]. Atmospheric Chemistry and Physics, 2011,11: 30563-30598. [13] Lee H J, Liu Y, Coull B A, et al. A novel calibration approach of MODIS AOD data to predict PM< sub>2.5</sub> concentrations [J]. Atmospheric Chemistry and Physics, 2011,11: 7991-8002. [14] Beckerman B S, Jerrett M, Serre M, et al. A hybrid approach to estimating national scale spatiotemporal variability of PM2.5 in the contiguous United States [J]. Environmental Science Technology, 2013,47:7233-7241. [15] Hu X, Waller L A, Al-hamdan M Z, et al. Estimating ground-level PM2.5 concentrations in the southeastern U.S. using geographically weighted regression [J]. Environmental Research, 2013,121:1-10. [16] HuX M, MaZ, Lin, et al. Impact of the Loess Plateau on the atmospheric boundary layer structure and air quality in the North China Plain: a case study [J]. Science of the Total Environment, 2014, 499:228-237. [17] Liu, Park J, Jacob D J, et al. Mapping annual mean ground-level PM2.5 concentrations using Multiangle Imaging Spectroradiometer aerosol optical depth over the contiguous United States [J]. Journal of Geophysical Research: Atmospheres, 2004,109: D22 206. [18] 吴 迪,高枞亭,李建平,等.东北地区PM2.5质量浓度遥感估算与时空分布特征 [J]. 地理科学, 2023,43(10):1869-1878. Wu D, Gao Z T, Li J P, et al. Remote sensing estimation and spatial-temporal distribution of PM2.5 concentration in Northeast China. Scientia Geographica Sinica, 2023,43(10):1869-1878. [19] Wei J, Huang W, Li Z Q, et al. Estimating 1-km-resolution PM<sub>2.5</sub> concentrations across China using the space-time random forest approach [J]. Remote Sensing of Environment, 2019; 231. [20] Xue W H, Wei J, Zhang J, et al. Inferring Near-Surface PM<sub>2.5</sub> Concentrations from the VIIRS Deep Blue Aerosol Product in China: A Spatiotemporally Weighted Random Forest Model [J]. Remote Sensing, 2021;13(3). [21] 李婷苑,谭浩波,王春林,等.卫星遥感AOD反演地面细颗粒物浓度方法与效果 [J]. 中国环境科学, 2020,40(1):13-23. Li T, Tan H, Wang C, et al. Research on daily PM2.5 simulation based on MODIS and multiple machine learning algorithms [J]. China Environmental Science, 2020,40(1):13-23. [22] 叶昊天,邹春辉,田宏伟.一种基于机器学习和葵花8号数据的PM2.5浓度监测方法 [J]. 气象与环境科学, 2023,46(3):89-97. Ye H, Zou C, Tian H. A Method for PM2.5 Concentration Monitoring Based on Machine Learning and Himawari-8Satellite Data [J]. Meteorological and Environmental Sciences, 2023,46(3):89-97. [23] 康新礼,张文豪,刘原萍,等.基于随机森林的京津冀地区PM2.5遥感反演及变化分析 [J]. 遥感技术与应用, 2022,37(2):424-435. Kang X L, Zhang W H, Liu Y P, et al. PM2.5 remote sensing retrieval and change analysis in Beijing-Tianjin-Hebei Region based on random forest model [J]. Remote Sensing Technology and Application, 2022,37(2):424-435. [24] Tian Z, Wei J, Li Z, et al. How Important Is Satellite-Retrieved Aerosol Optical Depth in Deriving Surface PM<sub>2.5</sub> Using Machine Learning? [J]. Remote Sensing, 2023;15(15). [25] Chen D, Gu X, Guo H, et al. Spatiotemporally continuous PM2.5 dataset in the Mekong River Basin from 2015 to 2022 using a stacking model [J]. Science of The Total Environment, Volume 914, 2024. [26] Gündoğdu S, Tuna T G, Li, Z. et al. Estimating daily PM2.5 concentrations using an extreme gradient boosting model based on VIIRS aerosol products over southeastern Europe [J]. Air Qual Atmos Health 2022,15, 2185-2198. [27] Zhang R, Zhou W, Chen H, et al. Aerosol Information Retrieval from GF-5B DPC Data over North China Using the Dark Dense Vegetation Algorithm [J]. Atmosphere 2023,14,241. [28] Sokolik I N, Toon O B, et al. Direct radiative forcing by anthropogenic airborne mineral aerosols [J]. Nature, 1996,381(6584):681-683. [29] Charlson R J, Schwartz S E, Hales J M, et al. Climate forcing by anthropogenic aerosols [J]. Science, 1992,255(5043):423-430. [30] 柴文雯,贾 夏,赵永华,等.黄土高原人类活动强度与植被覆盖时空关联性 [J]. 生态学报, 2024,44(15):6708-6721. Chai W, Jia X, Zhao Y, et al. Spatio-temporal correlation between human activity intensity and vegetation cover on the Loess Plateau[J]. Acta Ecologica Sinica, 2024,44(15):6708-6721. [31] Hu X, Belle J H, Meng X, et al. Estimating PM2.5 concentrations in the conterminous United States using the random forest approach [J]. Environmental Science & Technology, 2017,51(12):6936-6944. [32] 徐发昭,李 净,褚馨德,等.基于MODIS数据与多机器学习法的日PM2.5模拟研究 [J]. 中国环境科学, 2022,42(6):2523-2529. Xu F, Li J, Chu X, et al. Research on daily PM2.5 simulation based on MODIS and multiple machine learning algorithms [J]. China Environmental Science, 2022,42(6):2523-2529. [33] 陈 辉,厉 青,王中挺,等.MERSI和MODIS卫星监测京津冀及周边地区PM2.5浓度.遥感学报, 2018,22(5):822-832 Chen H, Li Q, Wang Z T, et al. Utilization of MERSI and MODIS data to monitor PM2.5 concentration in Beijing-Tianjin-Hebei and its surrounding areas. Journal of Remote Sensing, 2018,22(5):822-832. [34] AQG2021 全球空气质量准则 [S]. AQG2021 Global air quality guidelines [S]. [35] Guo B, Zhang D, Pei L, et al. Estimating PM2.5 concentrations via random forest method using satellite, auxiliary, and ground-level station dataset at multiple temporal scales across China in 2017 [J]. Science of The Total Environment, 2021,778:146288.