基于深度学习的中国连续空间覆盖PM2.5浓度预报

doi:10.11947/j.AGCS.2022.20200385

摘要/Abstract

摘要： 实现全国连续空间覆盖未来小时尺度的PM_2.5浓度实时、高精度预报是一个难题。本文建立基于深度学习的多层长短期记忆迭代模型和改进的空间反向传播神经网络S-BPNN模型来实现全国小时尺度PM_2.5浓度的空间预报。首先，研究基于空间相关性将全国1286个空气质量监测站点在空间上进行自适应分区，并对各个分区分别构建多层LSTM迭代预报模型实现未来24 h各个监测站点的PM_2.5浓度的实时预报。其次，应用改进的S-BPNN空间化模型实现未来24 h全国连续空间覆盖的PM_2.5浓度精细化制图。然后，利用2016—2019年中国PM_2.5监测站的历史数据进行训练和验证，结果显示预报模型和空间化模型的相关系数R²分别为0.88和0.87，表明模型都能实现较高的精度。最后，基于提出的预报模型和空间化模型，辅助从监测站实时获取的大气污染数据和气象数据，通过搭建的大气污染物浓度预报智能化在线信息原型系统可实时发布预报结果并可进行空间化展示。研究实现了全国连续空间覆盖的PM_2.5浓度高时空精度的实时预测，以支持大气污染联防联控和公众环境空间质量信息服务。

关键词: 大气污染, 时空预报, PM_2.5, 深度学习

Abstract: To achieve real-time and high-precision spatiotemporal forecast of PM_2.5 concentration in China with spatial coverage is still a difficult problem. In this paper, two models based on deep learning were established to realize the spatiotemporal forecast in the hourly scale of PM_2.5 concentration in China:a multi-layer long and short-term memory iterative model and an improved spatial back-propagation neural network (S-BPNN) model. First of all, we based on spatial correlation to divide 1286 air quality monitoring stations across the country in space adaptively and built a multi-layer LSTM iterative model for each region to achieve the PM_2.5 concentration forecast of each monitoring site in the next 24 hours. Secondly, the improved S-BPNN spatialization model was applied to realize the refined mapping of PM_2.5 concentration in a large continuous spatial coverage across the country in the next 24 hours. We integrated historical data of PM_2.5 monitoring stations in China from 2016 to 2019 for training and verification. The results show that the correlation coefficients R² of the forecast model and the spatialized model are respectively 0.88 and 0.89, indicating that the two models could achieve high accuracy. Finally, based on the proposed two models and related data crawled from monitoring stations in real-time, the intelligent online information prototype system for air pollutant concentration forecast can be built to release the forecast results of stations in real-time and display spatial results. The study has realized the real-time prediction of PM_2.5 concentration with high spatio-temporal accuracy across the country and has strongly supported joint prevention and control of air pollution and public environmental spatial quality information services.

Key words: air pollution, spatio-temporal forecast, PM_2.5, deep learning

中图分类号:

P208

毛文婧, 王卫林, 焦利民, 刘安宝. 基于深度学习的中国连续空间覆盖PM_2.5浓度预报[J]. 测绘学报, 2022, 51(3): 361-372.

MAO Wenjing, WANG Weilin, JIAO Limin, LIU Anbao. Continuous spatial coverage PM_2.5 concentration forecast in China based on deep learning[J]. Acta Geodaetica et Cartographica Sinica, 2022, 51(3): 361-372.

参考文献

[1] 李伟, 郑新奇. 结合VIIRS和监测数据插值的北京雾霾监测方法[J]. 测绘学报, 2015, 44(S1):123-128. DOI:10.11947/j.AGCS.2015.F022. LI Wei, ZHENG Xinqi. A haze monitoring method combined VIIRS images with real-time observation data interpolation in Beijing[J]. Acta Geodaetica et Cartographica Sinica, 2015, 44(S1):123-128. DOI:10.11947/j.AGCS.2015.F022.
[2] LI Can, HSU N C, TSAY S C. A study on the potential applications of satellite data in air quality monitoring and forecasting[J]. Atmospheric Environment, 2011, 45(22):3663-3675.
[3] 沈焕锋, 李同文. 大气PM_2.5遥感制图研究进展[J]. 测绘学报, 2019, 48(12):1624-1635. DOI:10.11947/j.AGCS.2019.20190456. SHEN Huanfeng, LI Tongwen. Progress of remote sensing mapping of atmospheric PM_2.5[J]. Acta Geodaetica et Cartographica Sinica, 2019, 48(12):1624-1635. DOI:10.11947/j.AGCS.2019.20190456.
[4] KAMPA M, CASTANAS E. Human health effects of air pollution[J]. Environmental Pollution, 2008, 151(2):362-367.
[5] KURT A, OKTAY A B. Forecasting air pollutant indicator levels with geographic models 3 days in advance using neural networks[J]. Expert Systems with Applications, 2010, 37(12):7986-7992.
[6] DENG Xuejiao, TIE Xuexi, WU Dui, et al. Long-term trend of visibility and its characterizations in the Pearl river delta (PRD) region, China[J]. Atmospheric Environment, 2008, 42(7):1424-1435.
[7] KIM Y, FU J S, MILLER T L. Improving ozone modeling in complex terrain at a fine grid resolution:partⅠ:examination of analysis nudging and all PBL schemes associated with LSMs in meteorological model[J]. Atmospheric Environment, 2010, 44(4):523-532.
[8] GENG Guannan, ZHANG Qiang, MARTIN R V, et al. Estimating long-term PM_2.5 concentrations in China using satellite-based aerosol optical depth and a chemical transport model[J]. Remote Sensing of Environment, 2015, 166:262-270.
[9] PAN Lin, SUN Baosheng, WANG Wei. City air quality forecasting and impact factors analysis based on grey model[J]. Procedia Engineering, 2011(12):74-79.
[10] ENGEL-COX J, OANH N T K, VAN DONKELAAR A, et al. Toward the next generation of air quality monitoring:particulate matter[J]. Atmospheric Environment, 2013, 80:584-590.
[11] ZHU Suling, LIAN Xiuyuan, WEI Lin, et al. PM_2.5 forecasting using SVR with PSOGSA algorithm based on CEEMD, GRNN and GCA considering meteorological factors[J]. Atmospheric Environment, 2018, 183:20-32.
[12] STADLOBER E, HÖRMANN S, PFEILER B. Quality and performance of a PM₁₀ daily forecasting model[J]. Atmospheric Environment, 2008, 42(6):1098-1109.
[13] 付倩娆. 基于多元线性回归的雾霾预测方法研究[J]. 计算机科学, 2016, 43(S1):526-528. FU Qianrao. Research on haze prediction based on multivariate linear regression[J]. Computer Science, 2016, 43(S1):526-528.
[14] HU Xuefei, BELLE J H, MENG Xia, et al. Estimating PM_2.5 concentrations in the conterminous united states using the random forest approach[J]. Environmental Science & Technology, 2017, 51(12):6936-6944.
[15] PEREZ P, REYES J. An integrated neural network model for PM10 forecasting[J]. Atmospheric Environment, 2006, 40(16):2845-2851.
[16] FENG Yu, ZHANG Wenfang, SUN Dezhi, et al. Ozone concentration forecast method based on genetic algorithm optimized back propagation neural networks and support vector machine data classification[J]. Atmospheric Environment, 2011, 45(11):1979-1985.
[17] MAO Wenjing, WANG Weilin, JIAO Limin, et al. Modeling air quality prediction using a deep learning approach:method optimization and evaluation[J]. Sustainable Cities and Society, 2021, 65(2):102-118.
[18] KOLEHMAINEN M, MARTIKAINEN H, RUUSKANEN J. Neural networks and periodic components used in air quality forecasting[J]. Atmospheric Environment, 2001, 35(5):815-825.
[19] VOUKANTSIS D, KARATZAS K, KUKKONEN J, et al. Intercomparison of air quality data using principal component analysis, and forecasting of PM₁₀ and PM_2.5 concentrations using artificial neural networks, in Thessaloniki and Helsinki[J]. Science of the Total Environment, 2011, 409(7):1266-1276.
[20] WANG Xinpeng, SUN Wenbin. Meteorological parameters and gaseous pollutant concentrations as predictors of daily continuous PM_2.5 concentrations using deep neural network in Beijing-Tianjin-Hebei, China[J]. Atmospheric Environment, 2019, 211:128-137.
[21] ZHAN Yu, LUO Yuzhou, DENG Xunfei, et al. Spatiotemporal prediction of continuous daily PM_2.5 concentra-tions across China using a spatially explicit machine learning algorithm[J]. Atmospheric Environment, 2017, 155:129-139.
[22] WANG Weilin, ZHAO Suli, JIAO Limin, et al. Estimation of PM_2.5 concentrations in China using a spatial back propagation neural network[J]. Scientific Reports, 2019, 9(1):1378-1387.
[23] MA Jun, DING Yuexiong, CHENG J C P, et al. A Lag-FLSTM deep learning network based on Bayesian optimization for multi-sequential-variant PM_2.5 prediction[J]. Sustainable Cities and Society, 2020, 60(9):1022-1037.
[24] XU Fulong, LI Zishen, ZHANG Kefei, et al. An investigation of optimal machine learning methods for the prediction of ROTI[J]. Journal of Geodesy and Geoinformation Science, 2020, 3(2):1-15. DOI:10.11947/j.JGGS.2020.0201.
[25] HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8):1735-1780.
[26] LI Xiang, PENG Ling, YAO Xiaojing, et al. Long short-term memory neural network for air pollutant concentration predictions:method development and evaluation[J]. Environmental Pollution, 2017, 231:997-1004.
[27] ZHAO Jiachen, DENG Fang, CAI Yeyun, et al. Long short-term memory-Fully connected (LSTM-FC) neural network for PM_2.5 concentration prediction[J]. Chemosphere, 2019, 220:486-492.
[28] PAK U, KIM C, RYU U, et al. A hybrid model based on convolutional neural networks and long short-term memory for ozone concentration prediction[J]. Air Quality, Atmosphere & Health, 2018, 11(8):883-895.
[29] WEN Congcong, LIU Shufu, YAO Xiaojing, et al. A novel spatiotemporal convolutional long short-term neural network for air pollution prediction[J]. Science of the Total Environment, 2019, 654:1091-1099.
[30] BAI Yun, LI Yong, WANG Xiaoxue, et al. Air pollutants concentrations forecasting using back propagation neural network based on wavelet decomposition with meteorological conditions[J]. Atmospheric Pollution Research, 2016, 7(3):557-566.
[31] ZHANG Boen, JIAO Limin, XU Gang, et al. Influences of wind and precipitation on different-sized particulate matter concentrations (PM_2.5, PM₁₀, PM_2.5-10)[J]. Meteorology and Atmospheric Physics, 2018, 130(3):383-392.
[32] RESHEF D N, RESHEF Y A, FINUCANE H K, et al. Detecting novel associations in large data sets[J]. Science, 2011, 334(6062):1518-1524.
[33] LI Tongwen, SHEN Huanfeng, ZENG Chao, et al. Point-surface fusion of station measurements and satellite observations for mapping PM_2.5 distribution in China:methods and assessment[J]. Atmospheric Environment, 2017, 152:477-489.
[34] LI Tiantian, GUO Yuming, LIU Yang, et al. Estimating mortality burden attributable to short-term PM_2.5 exposure:a national observational study in China[J]. Environment International, 2019, 125:245-251.