Abstract：In this study, two newly developed long short-term memory (LSTM) model and Light Gradient Boosting Machine (LGBM) algorithms were introduced for application in extended-range forecasting of PM2.5 in Shanghai by incorporating three members of the sub-seasonal-to-seasonal prediction project (S2S) forecasting, six prediction models were obtained. Therefore, based on six models forecast results, an accurate ~40d PM2.5 prediction fusion model over Shanghai was developed by LSTM algorithm, providing new insights for air pollution extended-range forecasting. The evaluation results indicated the fusion model exhibited not only much better accuracy but also captured the pollution process more closely compared to than any of the six single model. The correlation coefficients for the fusion model forecasts on lead times of 11~40 days ranged from 0.47 to 0.76, 23.5%~31.1% higher than other six single model. The root mean square errors (RMSE) for the fusion model forecasts on lead times of 11~40 days ranged from 19 to 25.1µg/m3, 19%~19.3% lower than other six single model. The fusion model could not only better predict the overall trend of PM2.5 concentration, but also the occurrence time of peak and valley, and its Heidke Skill Scores (HSS) was between 0.18 and 0.5, showing a good prediction skill. The fusion model could predict pollution episodes at a lead of 11~40 days, and the overall prediction accuracy was 75.5% at a lead of 11~40 days. For three pollution episodes lasting for 3 days or more, the prediction accuracy reached 100% at a lead of 11~40 days .The prediction efficiency of the fusion model was up to 40days, nearly four times that of the current pollution numerical prediction model (generally 96~240h), and the calculation speed was fast, which could save a lot of computing resources and time costs.
马井会, 瞿元昊, 余钟奇, 许建明. 上海市PM2.5浓度延伸期预测模型的构建及评估[J]. 中国环境科学, 2023, 43(7): 3290-3298.
MA Jing-hui, QU Yuan-hao, YU Zhong-qi, XU Jian-ming. Establishment and evaluation of prediction model for PM2.5 concentration extension period in Shanghai. CHINA ENVIRONMENTAL SCIENCECE, 2023, 43(7): 3290-3298.
王 静,赵雪艳,殷宝辉,等.城市间PM2.5源的碳成分谱比较 [J]. 中国环境科学, 2022,42(12):5523-5532. Wang J, Zhao X Y, Yin B H, et al. Comparison of carbon composition spectra of PM2.5 sources between cities [J]. Chinese Environmental Science, 2022,42(12):5523-5532.
[2]
杜亚彬,赵红梅,张学磊,等.基于WRF-CMAQ评估秸秆禁烧对区域空气质量的影响 [J]. 中国环境科学, 2022,42(12):5578-5588. Du Y B, Zhao H M, Zhang X L, et al. Assessing the impact of straw burning ban on regional air quality based on WRF-CMAQ [J] China Environmental Science, 2022,42(12):5578-5588.
[3]
樊文雁,蔡子颖,姚 青,等.区域输送对天津臭氧污染的影响 [J]. 中国环境科学, 2022,42(11):4991-4999. Fan W Y, Cai Z Y, Yao Q, et al. The impact of regional transportation on ozone pollution in Tianjin [J] China Environmental Science, 2022, 42(11):4991-4999.
[4]
吴 进,李 琛,王志立,等.两类污染型下中低空风切变对PM2.5浓度影响 [J]. 中国环境科学, 2022,42(11):5016-5022. Wu J, Li C, Wang Z L, et al. The impact of middle and low level wind shear on PM2.5 concentration under two types of pollution [J]. China Environmental Sciences, 2022,42(11):5016-5022.
[5]
刘旭艳,韩秀珍,梁林林,等.2015~2021年景洪市PM2.5的传输路径及潜在源区 [J]. 中国环境科学, 2022,42(11):5023-5035. Liu X Y, Han X Z, Liang L L, et al. The transmission path and potential source area of PM2.5 in Jinghong City from 2015 to 2021 [J]. China Environmental Science, 2022,42(11):5023-5035.
[6]
Baklanov A, Mestayer P G, Clappier A, et al. Towards improving the simulation of meteorological fields in urban areas through updated/advanced surface fluxes description [J]. Atmospheric Chemistry and Physics, 2008,8(3):523-543.
[7]
Kim Y, Fu J S, Miller T L. Improving ozone modeling in complex terrain at a fine grid resolution: part I-examination of analysis nudging and all PBL schemes associated with LSMs in meteorological model [J]. Atmospheric Environment, 2010,44(4):523-532.
[8]
Woody M C, Wong H W, West J J, et al. Multiscale predictions of aviation-attributable PM2.5 for U.S. airports modeled using CMAQ with plume-in-grid and an aircraft-specific 1-D emission model [J]. Atmospheric Environment, 2016,147:384-394.
[9]
Carlo P D, Pitari G, Mancini E, et al. Evolution of surface ozone in central Italy based on observations and statistical model [J]. Journal of Geophysical Research, 2007,112:D10316.
[10]
Castellano M, Franco A, Cartelle D, et al. Identification of NOx and ozone episodes and estimation of ozone by statistical analysis [J]. Water Air Soil Pollut, 2009,198:95-110.
[11]
Gennaro G, Trizioa L, Gilioa A D, et al. Neural network model for the prediction of PM10 daily concentrations in two sites in the Western Mediterranean [J]. Science of the Total Environment, 2013,463-464:875-883.
[12]
Donnelly A, Misstear B, Broderick B. Real time air quality forecasting using integrated parametric and non-parametric regression techniques [J]. Atmospheric Environment, 2015,103:53-65.
[13]
Vitart F, Coauthors. The subseasonal to seasonal (S2S) prediction project database [J]. Bulletin of the American Meteorological Society, 2017,98:163−173.
[14]
Seng D, Zhang Q, Zhang X, et al. Spatiotemporal prediction of air quality based on LSTM neural network [J]. Alexandria Engineering Journal, 2021,60(2):2021−2032.
[15]
Qin D, Yu J, Zou G, et al. A novel combined prediction scheme based on CNN and LSTM for urban PM2.5 concentration [J]. IEEE Access, 2019,7:20050-20059.
[16]
Qi Y, Li Q, Karimian H, et al. A hybrid model for spatiotemporal forecasting of PM2.5 based on graph convolutional neural network and long short-term memory [J]. Science of the Total Environment, 2019, 664:1-10.
[17]
Zhou Y, Chang F J, Chang L C, et al. Explore a deep learning multi-output neural network for regional multi-step-ahead air quality forecasts [J]. Journal of Cleaner Production. 2019,209:134-145.
[18]
Wen C, Liu S, Yao X, 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.
[19]
Wang J, Song G. A deep spatial-temporal ensemble model for air quality prediction [J]. Neurocomputing, 2018,314:198-206.
[20]
Qu Y H, Ma J H, Yu Z Q. Extended-Range Forecasting of PM2.5 Based on the S2S: A Case Study in Shanghai, China [J]. Frontiers in Environmental Science, 2022,10(5):1-14.
[21]
Fan Z, Zhan Q, Yang C, et al. Estimating PM2.5 concentrations using spatially local xgboost based on full-covered SARA AOD at the urban scale. Remote Sensing, 2020,12(20):3368.
[22]
Kianian B, Liu Y, Chang H H. Imputing satellite-derived aerosol optical depth using a multi-resolution spatial model and random forest for PM2.5 prediction [J]. Remote Sensing, 2021,13(1):126.
[23]
Ma J H, Zhang R H, Xu J M, et al. MERRA-2PM2.5 mass concentration reconstruction in China mainland based on LightGBM machine learning [J]. Science of the Total Environment, 2022,827:154363.
[24]
Yu Z Q, Ma J H, Qu Y H, et al. PM2.5 extended-range forecast based on MJO and S2S using LightGBM [J]. Science of the Total Environment 2023,880:163358.
[25]
Ke G, Meng Q, Finley T, et al. LightGBM: A highly ef-fificient gradient boosting decision tree Proceedings of the 31st Annual Conference on Neural Information Processing Systems, Long Beach, CA, USA, 4-9December 2017:3146-3154.
[26]
Zhong J, Zhang X, Gui K, et al. Robust prediction of hourly PM2.5 from meteorological data using LightGBM [J]. National science review, 2021,8(10):8-10.
[27]
Hochreiter S, Schmidhuber J. Long short-term memory [J]. Neural Computation, 1997,9(8):1735-1780.
[28]
Graves Alex. Long short-term memory supervised sequence labelling with recurrent neural networks [Z]. Springer Berlin Heidelberg, 2012: 1735−1780.
[29]
Heidke P. Calculation of the success and goodness of strong wind forecasts in the storm warning service [J]. Geografiska Annaler, 1926,8:301−349.
