基于深度学习方法的PM<sub>2.5</sub>精细化时空估算模型

摘要
图/表
参考文献
相关文章 (15)

全文: PDF (1214 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要提出一种基于深度学习方法的地面PM_2.5浓度时空估算模型（PM_2.5-DNN），该模型基于葵花-8卫星反演的AOD数据，结合PM_2.5监测站和气象站点观测数据对北京市地面PM_2.5浓度进行了逐时的高精度模拟，同时将PM_2.5-DNN模型的模拟性能与当前的主流方法进行了对比研究.结果表明，使用PM_2.5-DNN模型估算的北京地区1km分辨率每小时地面PM_2.5浓度与地表监测站观测数据对比的一致性较好，模型估算精度可达到R²=0.88，性能优于当前的主流方法.本文所提出的方法适用于区域尺度PM_2.5浓度时空分布细粒度建模与估算，采用端到端的训练方式构建模型，为精细的PM_2.5浓度估算提供了一个简便而有效的方法模型.

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	耿冰
	孙义博
	曾巧林
	商豪律
	刘霄宇
	单菁菁

关键词 ：大气细颗粒物浓度估算, 深度学习, 卫星遥感, 光学气溶胶厚度

Abstract：The concentration distribution of fine particulate matter (PM_2.5) on the surface of the atmosphere has a strong temporal and spatial heterogeneity. Due to the limited spatial coverage of traditional PM_2.5 monitoring sites, it is difficult to reflect the complexity of PM_2.5 concentration in time and space. This paper proposed a temporal and spatial prediction model of ground PM_2.5 concentration based on deep learning methods(PM_2.5-DNN). Based on the AOD data from Kuihua-8satellite and the observation data from PM_2.5 monitoring and meteorological station, hourly high-precision simulations of the surface PM_2.5 concentration in Beijing had been carried out. The results show that the 1km resolution hourly ground PM_2.5 concentration in Beijing area estimated by the PM_2.5-DNN model had good consistency with the observation data from the surface monitoring station. The model estimation accuracy could reach R²=0.88, which was better than the performance of current mainstream method. The method proposed in this paper was suitable for fine-grained modelling and estimation of the temporal and spatial distribution of PM_2.5 concentration at a regional scale. The end-to-end training method is used to construct the model, which provides a simple and effective method model for fine PM_2.5 concentration estimation.

Key words： PM_2.5 concentration estimation deep learning satellite remote sensing aerosol optical depth

收稿日期: 2020-12-28

PACS:

X513

基金资助:中国博士后科学基金资助项目（2018M631684）；国家自然科学基金资助项目（41590855）

通讯作者: 单菁菁,研究员,shanjj2008@126.com E-mail: shanjj2008@126.com

作者简介: 耿冰(1987-),女,吉林长春人,博士,主要从事国土规划和信息技术研究.发表论文10余篇.

引用本文:

耿冰, 孙义博, 曾巧林, 商豪律, 刘霄宇, 单菁菁. 基于深度学习方法的PM_2.5精细化时空估算模型[J]. 中国环境科学, 2021, 41(8): 3502-3510. GENG Bing, SUN Yi-bo, ZENG Qiao-lin, SHANG Hao-lv, LIU Xiao-yu, SHAN Jing-jing. Refined spatiotemporal estimation model of PM_2.5 based on deep learning method. CHINA ENVIRONMENTAL SCIENCECE, 2021, 41(8): 3502-3510.

链接本文:

http://www.zghjkx.com.cn/CN/ 或 http://www.zghjkx.com.cn/CN/Y2021/V41/I8/3502

[1]	何映月,张松林.2000~2016年中国PM2.5时空格局[J]. 中国环境科学, 2020,40(8):3284-3293. He Y Y, Zhang S L. Spatiotemporal pattern of PM2.5 from 2000 to 2016 in China[J]. China Environmental Science, 2020,40(8):3284-3293.
[2]	李名升,任晓霞,于洋,等.中国大陆城市PM2.5污染时空分布规律[J]. 中国环境科学, 2016,36(3):641-650. Li M S, Ren X X, Yu Y, et al. Spatiotemporal pattern of ground-level fine particulate matter (PM2.5) pollution in mainland China[J]. China Environmental Science, 2016,36(3):641-650.
[3]	Zhan Y, Luo Y, Deng X, et al. Spatiotemporal prediction of continuous daily PM2.5 concentrations across China using a spatially explicit machine learning algorithm[J]. Atmospheric environment, 2017,155:129-139.
[4]	郑国威,王腾军,杨友森,等.基于遗传小波神经网络的PM2.5浓度预测模型[J]. 测绘与空间地理信息, 2018,41(9):248-250. Zheng G W, Wang T J, Yang Y S, et al. PM2.5 concentration prediction model based on genetic wavelet neural network[J]. Geomatics & Spatial Information Technology, 2018,41(9):248-250.
[5]	梁泽,王玥瑶,岳远紊,等.耦合遗传算法与RBF神经网络的PM2.5浓度预测模型[J]. 中国环境科学, 2020,40(2):523-529. Liang Z, Wang Y Y, Yue Y W, et al. A coupling model of genetic algorithm and RBF neural network for the prediction of PM2.5 concentration[J]. China Environmental Science, 2020,40(2):523-529.
[6]	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.
[7]	李锋,朱彬,安俊岭,等.2013年12月初长江三角洲及周边地区重霾污染的数值模拟[J]. 中国环境科学, 2015,35(7):1965-1974. Li F, Zhu B, An J L, et al. Modeling study of a severe haze episode occurred over the Yangtze River Delta and its surrounding regions during early December, 2013[J]. China Environmental Science, 2015, 35(7):1965-1974.
[8]	李燚航,翟卫欣,颜寒祺,等.基于U-net神经网络模型的PM2.5逐小时浓度值预测模型[J]. 北京大学学报(自然科学版), 2020,56(5):796-804. Li Y H, Zhai W X, Yan H Q, et al. Prediction of PM2.5 hour concentration based on U-net neural network[J]. Acta Scientiarum Naturalium Universitatis Pekinensis, 2020,56(5):796-804.
[9]	付宏臣,孙艳玲,王斌,等.基于AOD数据和GWR模型估算京津冀地区PM2.5浓度[J]. 中国环境科学, 2019,39(11):4530-4537. Fu H C, Sun Y L, Wang B, et al. Estimation of PM2.5 concentration in Beijing-Tianjin-Hebei region based on AOD data and GWR model[J]. China Environmental Science, 2019,39(11):4530-4537.
[10]	Wu J, Yao F, Li W, et al. VIIRS-based remote sensing estimation of ground-level PM2.5 concentrations in Beijing-Tianjin-Hebei:A spatiotemporal statistical model[J]. Remote Sensing of Environment, 2016,184:316-328.
[11]	周亮,周成虎,杨帆,等.2000~2011年中国PM2.5时空演化特征及驱动因素解析[J]. 地理学报, 2017,72(11):2079-2092. Zhou L, Zhou C H, Yang F, et al. Spatio-temporal evolution and the influencing factors of PM2.5 in China between 2000 and 2011[J]. Acta Geographica Sinica, 2017,72(11):2079-2092.
[12]	陶金花,张美根,陈良富,等.一种基于卫星遥感AOT估算近地面颗粒物的方法[J]. 中国科学:地球科学, 2013,43(1):143-154. Tao J H, Zhang M G, Chen L F, et al. Method to estimate concentration of surface-level particulate matter from satellite-based aerosol optical thickness[J]. Science China:Earth Sciences, 2013,43(1):143-154.
[13]	李成才,毛节泰,刘启汉,等.MODIS卫星遥感气溶胶产品在北京市大气污染研究中的应用[J]. 中国科学:D辑, 2005,35(A01):177-186. Li C C, Mao J T, Liu Q H, et al. Application of MODIS remote sensing aerosol products in Beijing air pollution research[J]. Science in China Ser. D Earth Sciences, 2005,35(A01):177-186.
[14]	吴健生,王茜.基于AOD数据反演地面PM2.5浓度研究进展[J]. 环境科学与技术, 2017,40(8):68-76. Wu J S, Wang X. Research progress of retrieval ground-level PM2.5 concentration based on AOD data[J]. Environmental Science & Technology, 2017,40(8):68-76.
[15]	Xie Y, Wang Y, Zhang K, et al. Daily estimation of ground-level PM2.5 concentrations over Beijing using 3km resolution MODIS AOD[J]. Environmental Science & Technology, 2015,49(20):12280-12288.
[16]	Paciorek C J, Liu Y, Moreno-Macias H, et al. Spatiotemporal associations between GOES aerosol optical depth retrievals and ground-level PM2.5[J]. Environmental Science & Technology, 2008, 42(15):5800-5806.
[17]	Yang F, Wang Y, Tao J, et al. Preliminary investigation of a new AHI aerosol optical depth (AOD) retrieval algorithm and evaluation with multiple source AOD measurements in China[J]. Remote Sensing, 2018,10(5):748.
[18]	王琴,张大伟,刘保献,等.基于PMF模型的北京市PM2.5来源的时空分布特征[J]. 中国环境科学, 2015,35(10):2917-2924. Wang Q, Zhang D W, Liu BX, et al. Spatial and temporal variations of ambient PM2.5 source contributions using positive matrix factorization[J]. China Environmental Science, 2015,35(10):2917-2924.
[19]	周丽,徐祥德,丁国安,等.北京地区气溶胶PM2.5粒子浓度的相关因子及其估算模型[J]. 气象学报, 2003,61(6):761-768. Zhou L, Xu D, Ding G A, et al. The correlation factors and pollution forecast model for PM2.5 concentration in Beijing area[J]. Acta Meteorologica Sinica, 2003,61(6):761-768.
[20]	贺祥,林振山,刘会玉,等.基于灰色关联模型对江苏省PM2.5浓度影响因素的分析[J]. 地理学报, 2016,71(7):1119-1129. He X, Lin S, Liu H Y, et al. Analysis of the driving factors of PM2.5 in Jiangsu province based on grey correlation model[J]. Acta Geographica Sinica, 2016,71(7):1119-1129.
[21]	王敏,邹滨,郭宇,等.基于BP人工神经网络的城市PM2.5浓度空间预测[J]. 环境污染与防治, 2013,35(9):63-66. Wang M, Zou B, Guo Y, BP artificial neural network-based analysis of spatial variability of urban PM2.5 concentration[J]. Environment Pollution & Control, 2013,35(9):63-66.
[22]	宋国君,国潇丹,杨啸,等.沈阳市PM2.5浓度ARIMA-SVM组合预测研究[J]. 中国环境科学, 2018,38(11):4031-4039. Song G J, Guo X D, Yang X, et al. ARIMA-SVM combination prediction of PM2.5 concentration in Shenyang[J]. China Environmental Science, 2018,38(11):4031-4039.
[23]	Chu D A, Kaufman Y, 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(D21).
[24]	Gupta P, Christopher S A. Particulate matter air quality assessment using integrated surface, satellite, and meteorological products:2. A neural network approach[J]. Journal of Geophysical Research:Atmospheres, 2009,114(D20).
[25]	孙成,王卫,刘方田,等.基于线性混合效应模型的河北省PM2.5浓度时空变化模型研究[J]. 环境科学研究, 2019,32(9):1500-1509. Sun C, Wang W, Liu F T, et al. Spatial-temporal simulation of PM2.5 concentration in Hebei Province based on linear mixed effects model[J]. Research of Environmental Sciences, 2019,32(9):1500-1509.
[26]	Ma Z, Liu Y, Zhao Q, et al. Satellite-derived high resolution PM2.5 concentrations in Yangtze River Delta Region of China using improved linear mixed effects model[J]. Atmospheric Environment, 2016,133:156-164.
[27]	Kloog I, Koutrakis P, Coull B A, et al. Assessing temporally and spatially resolved PM2.5 exposures for epidemiological studies using satellite aerosol optical depth measurements[J]. Atmospheric Environment, 2011,45(35):6267-6275.
[28]	李娟,尉鹏,戴学之,等.基于机器学习方法的西安市数值模拟优化研究[J]. 环境科学研究, 2021,34(4):872-881. Li J, Wei P, Dai X Z, et al. Optimization of numerical simulation in Xi'an based on machine learning methods[J]. Research of Environmental Sciences, 2021,34(4):872-881.
[29]	Liu B-C, Binaykia A, Chang P-C, et al. Urban air quality forecasting based on multi-dimensional collaborative Support Vector Regression (SVR):A case study of Beijing-Tianjin-Shijiazhuang[J]. PloS one, 2017,12(7):1-17.
[30]	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.
[31]	Yu W, Liu Y, Ma Z, et al. Improving satellite-based PM2.5 estimates in China using Gaussian processes modeling in a Bayesian hierarchical setting[J]. Scientific reports, 2017,7(1):1-9.
[32]	Li T, Shen H, Zeng C, et al. Point-surface fusion of station measurements and satellite observations for mapping PM2.5 distribution in China:Methods and assessment[J]. Atmospheric Environment, 2017,152:477-489.
[33]	Hu X, Waller L A, Al-Hamdan M Z, et al. Estimating ground-level PM2.5 concentrations in the southeastern US using geographically weighted regression[J]. Environmental research, 2013,121:1-10.
[34]	Bengio Y. Learning deep architectures for AI[M]. Now Publishers Inc, 2009.
[35]	郑毅,朱成璋.基于深度信念网络的PM2.5预测[J]. 山东大学学报(工学版), 2014,44(6):19-25. Zheng Y, Zhu C Z. A prediction method of atmospheric PM2.5 based on DBNs[J]. Journal of Shandong University (Engineering Science), 2014,44(6):19-25.
[36]	马井会,曹钰,余钟奇,等.深度学习方法在上海市PM2.5浓度预报中的应用[J]. 中国环境科学, 2020,40(2):530-538. Ma J H, Cao Y, Yu Z Q, et al. The application of deep learning method in Shanghai PM2.5 prediction[J]. China Environmental Science, 2020,40(2):530-538.
[37]	Ian Goodfellow, Yoshua Bengio, Aaton Courville. Deep learning[M]. The MIT Press, 2016.