时间尺度重构EEMD-GRNN改进模型预测PM2.5的研究

doi:10.12082/dqxxkx.2019.180598

摘要/Abstract

摘要：

合理构建PM_2.5浓度预测模型是科学、准确地预测PM_2.5浓度变化的关键。传统PM_2.5预测EEMD-GRNN模型具有较好的预测精度,但是存在过于关注研究数据本身而忽略其物理意义的不足。本研究基于南京市2014-2017年PM_2.5浓度时间序列数据,分析PM_2.5浓度多尺度变化特征及其对气象因子和大气污染因子的尺度响应,基于时间尺度重构进行EEMD-GRNN模型的改进与实证研究。南京市样本数据PM_2.5浓度变化表现为明显的天际尺度和月际尺度,从重构尺度（天际、月际）构建GRNN模型更具有现实意义;同时,PM_2.5对PM₁₀、NO₂、O₃、RH、MinT等因子存在多尺度响应效应,以其作为GRNN模型中的输入变量更具有时间序列上的解释意义。改进后的EEMD-GRNN模型具有更高的PM_2.5浓度预测精度,MAE、MAPE、RMSE和R²分别为6.17、18.41%、8.32和0.95,而传统EEMD-GRNN模型的模型有效性检验结果分别为8.37、27.56%、11.56、0.91。对于高浓度天（PM_2.5浓度大于100 μg/m³）的预测,改进模型更是全面优于传统EEMD-GRNN模型,MAPE为12.02%,相较于传统模型提高了9.03%。

关键词: 南京市, PM_2.5, 本征模函数, 时间尺度重构, 多尺度响应, 集合经验模态分解, 广义回归神经网络

Abstract:

Reasonable construction of PM_2.5 concentration prediction models is the key to scientifically and accurately predict PM_2.5 concentration dynamics. The traditional EEMD-GRNN model predicts PM_2.5 concentration with good prediction accuracy, but focuses more on research data and less on its physical meaning. Based on the time series data of PM_2.5 concentration in Nanjing duing 2014-2017, this study analyzed the multi-scale variations of PM_2.5 concentration and its scale response to meteorological and atmospheric pollution factors. Additionally, the EEMD-GRNN model was reconstructed and validated based on time scale reconstruction. Results show that: ① based on scale reconstruction, the improved EEMD-GRNN model was scientific in predicting the PM_2.5 concentration as it considered the variation of PM_2.5 concentration and its multi-scale responses to atmospheric pollutants and meteorological factors. The PM_2.5 concentration of Nanjing’s sample data had obvious inter-daily and inter-monthly variations. Thus, it is more practical to construct GRNN modeling at the reconstruction scales (i.e., daily and monthly). Meanwhile, PM_2.5 was sensitive to factors such as PM₁₀, NO₂, O₃, RH, and MinT. Thus, choosing these factors as input variables in the GRNN model can be more explanatory for time-series. ② the improved EEMD-GRNN model was more accurate in predicting PM_2.5 concentration. The MAE, MAPE, RMSE, and R² of the improved models are 6.17, 18.41%, 8.32, and 0.95, respectively, which are superior to the validity test results of the traditional EEMD-GRNN model. Furthermore, for the prediction of high-concentration days (i.e., PM_2.5 concentration greater than 100 μg/m³), the improved model was more comprehensive than the traditional EEMD-GRNN model. With the new EEMD-GRNN model, the MAPE is 12.02%, 9.03% higher than the traditional model. Our findings indicate that the improved EEMD-GRNN based on scale reconstruction is an efficient method to scientifically and accurately predict PM_2.5 concentration, especially in days with high PM_2.5concentration.

Key words: Nanjing, PM_2.5, intrinsic mode function(IMF), time scale reconstruction, multi-scale response, ensemble empirical mode decomposition, generalized regression neural network

符海月, 张祎婷. 时间尺度重构EEMD-GRNN改进模型预测PM_2.5的研究[J]. 地球信息科学学报, 2019, 21(7): 1132-1142.

Haiyue FU, Yiting ZHANG. Improving the EEMD-GRNN Model for PM_2.5 Prediction based on Time Scale Reconstruction[J]. Journal of Geo-information Science, 2019, 21(7): 1132-1142.

图/表 13

图1

图2

表1

图3

表2

图4

图5

表3

图6

图7

表4

图8

表5

参考文献 41

[1]	任阵海,万本太,苏福庆,等.当前我国大气环境质量的几个特征[J].环境科学研究,2004,17(1):1-6.
	[ Ren Z H, Wan B T, Su F Q, et al.Several characteristics of atmospheric environment quality in China at present[J]. Research of Environment Sciences, 2004,17(1):1-6. ]
[2]	Chan C K, Yao X H.Air pollution in mega cities in China[J]. Atmospheric Environment, 2008,42(1):1-42. doi: 10.1016/j.atmosenv.2007.09.003
[3]	Bandowe B A M, Meusel H, Huang R, et al. PM2.5-bound oxygenated PAHs, nitro-PAHs and parent-PAHs from the atmosphere of a Chinese megacity: Seasonal variation, sources and cancer risk assessment[J]. Science of The Total Environment, 2014,474(3):77-87.
[4]	Tao J, Zhang L, Ho K, et al.Impact of PM2.5 chemical compositions on aerosol light scattering in Guangzhou — the largest megacity in South China[J]. Atmospheric Research, 2014,136(1):48-58.
[5]	Hu X, Waller L A, Lyapustin A, et al.Estimating ground-level PM2.5 concentrations in the Southeastern United States using MAIAC AOD retrievals and a two-stage model[J]. Remote Sensing of Environment, 2014,140(1):220-232. doi: 10.1016/j.rse.2013.08.032
[6]	Xin J Y, Zhang Q, Wang L L, et al.The empirical relationship between the PM2.5 concentration and aerosol optical depth over the background of North China from 2009 to 2011[J]. Atmospheric Research, 2014,138(3):179-188. doi: 10.1016/j.atmosres.2013.11.001
[7]	Wu S W, Deng F R, Hao Y, et al.Fine particulate matter, temperature, and lung function in healthy adults: Findings from the HVNR study[J]. Chemosphere, 2014,108:168-174. doi: 10.1016/j.chemosphere.2014.01.032
[8]	韩素芹,张裕芬,李英华,等.天津市春季气溶胶消光特征和辐射效应的数值模拟[J].中国环境科学,2011,31(1):8-12.
	[ Han S Q, Zhang Y F, Li Y H, et al.Simulation of extinction and radiant effect of aerosol in spring of Tianjin City[J]. China Environment Science, 2011,31(1):8-12. ]
[9]	Tie X X, Wu D, Brasseur G.Lung cancer mortality and exposure to atmospheric aerosol particles in Guangzhou[J], China Atmospheric Environment, 2009,43(14):2375-2377. doi: 10.1016/j.atmosenv.2009.01.036
[10]	Jung J, Lee H, Kim Y J, et al.Aerosol chemistry and the effect of aerosol water content on visibility impairment and radiative forcing in Guangzhou during the 2006 Pearl River Delta campaign[J]. Journal of Environmental Management, 2009,90(11):3231-3244.
[11]	Díaz-Robles L A, Ortega J C, Fu J S, et al. A hybrid ARIMA and artificial neural networks model to forecast particulate matter in urban areas: The case of Temuco, Chile[J]. Atmospheric Environment, 2008,42(35):8331-8340. doi: 10.1016/j.atmosenv.2008.07.020
[12]	Jian L, Zhao Y, Zhu Y P, et al.An application of ARIMA model to predict submicron particle concentrations from meteorological factors at a busy roadside in Hangzhou, China[J]. Science of The Total Environment, 2012,426(2):336-345. doi: 10.1016/j.scitotenv.2012.03.025
[13]	Vlachogianni A, Kassomenos P, Karppinen A, et al.Evaluation of a multiple regression model for the forecasting of the concentrations of NOx and PM10 in Athens and Helsinki[J]. Science of The Total Environment, 2011,409(8):1559-1571. doi: 10.1016/j.scitotenv.2010.12.040
[14]	Cobourn W G.An enhanced PM2.5 air quality forecast model based on nonlinear regression and back-trajectory concentrations[J]. Atmospheric Environment, 2010,44(25):3015-3023. doi: 10.1016/j.atmosenv.2010.05.009
[15]	Baker K R, Foley K M.A nonlinear regression model estimating single source concentrations of primary and secondarily formed PM2.5[J]. Atmospheric Environment, 2011,45(22):3758-3767. doi: 10.1016/j.atmosenv.2011.03.074
[16]	Corani G.Air quality prediction in Milan: Feed-forward neural networks, pruned neural networks and lazy learning[J]. Ecological Modelling, 2005,185(2):513-529. doi: 10.1016/j.ecolmodel.2005.01.008
[17]	Voukantsis D, Karatzas K, Kukkonen J, et al.Intercomparison of air quality data using principal component analysis, and forecasting of PM10 and PM2.5 concentrations using artificial neural networks, in Thessaloniki and Helsinki[J]. Science of The Total Environment, 2011,409(7):1266-1276. doi: 10.1016/j.scitotenv.2010.12.039
[18]	Pérez P, Trier A, Reyes J.Prediction of PM2.5 concentrations several hours in advance using neural networks in Santiago, Chile[J]. Atmospheric Environment, 2000,34(8):1189-1196. doi: 10.1016/S1352-2310(99)00316-7
[19]	Sousa S, Martins F, Alvimferraz M, et al.Multiple linear regression and artificial neural networks based on principal components to predict ozone concentrations[J]. Environmental Modelling & Software, 2007,22(1):97-103.
[20]	Fernando H J S, Mammarella M C, Grandoni G, et al. Forecasting PM10 in metropolitan areas: Efficacy of neural networks[J]. Environmental Pollution, 2012,163(4):62-67. doi: 10.1016/j.envpol.2011.12.018
[21]	Antanasijević D Z, Pocajt V V, Povrenović D S, et al.PM10 emission forecasting using artificial neural networks and genetic algorithm input variable optimization[J]. Science of The Total Environment, 2013,443(3):511-519. doi: 10.1016/j.scitotenv.2012.10.110
[22]	Perez P.Combined model for PM10 forecasting in a large city[J]. Atmospheric Environment, 2012,60(60):271-276. doi: 10.1016/j.atmosenv.2012.06.024
[23]	Zhou Q P, Jiang H Y, Wang J Z, et al.A hybrid model for PM2.5 forecasting based on ensemble empirical mode decomposition and a general regression neural network[J]. Science of the Total Environment, 2014,496(2):264-274. doi: 10.1016/j.scitotenv.2014.07.051
[24]	Ausati S, Amanollahi J.Assessing the accuracy of ANFIS, EEMD-GRNN, PCR, and MLR models in predicting PM2.5[J]. Atmospheric Environment, 2016,142:465-474. doi: 10.1016/j.atmosenv.2016.08.007
[25]	赵娜,焦毅蒙.基于TRMM降水数据的空间降尺度模拟[J].地球信息科学学报,2018,20(10):1388-1395.
	[ Zhao N, Jiao Y M.Spatial downscaling simulation of TRMM satellite precipitation data[J]. Jouural of Geo-information Science, 2018,20(10):1388-1395. ]
[26]	徐凯健,田庆久,杨闫君,等.遥感土地覆被分类的空间尺度响应研究[J].地球信息科学学报,2018,20(2):246-253.
	[ Xu K J, Tian Q J, Yang Y J, et al.Response of spatial scale for land cover classification of remote sensing[J]. Jouural of Geo-information Science, 2018,20(2):246-253. ]
[27]	Huang N E, Shen Z, Long S R, et al.The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Proceedings of the Royal Society of London[J]. Series A: Mathematical, Physical and Engineering Sciences. 1998,454(1971):903-995. doi: 10.1098/rspa.1998.0193
[28]	Wu Z H, Huang N E.Ensemble empirical mode decomposition:A noise-assisted data analysis method[J]. Advances in Adaptive Data Analysis, 2009,1(1):1-41. doi: 10.1142/S1793536909000047
[29]	Wu Z H, Huang N E.A study of the characteristics of white noise using the empirical mode decomposition method[J]. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 2004,460(2046):1597-1611. doi: 10.1098/rspa.2003.1221
[30]	Wu Z H, Huang N E.Statistical significance test of intrinsic mode functions[M]. Singapore: World Scientific, 2005.
[31]	Specht D F.A general regression neural network[J]. IEEE Transactions on Neural Networks, 1991,2(6):568-576. doi: 10.1109/72.97934
[32]	王占山,李云婷,陈添,等. 2013年北京市PM2.5的时空分布[J].地理学报,2015,70(1):110-120.
	[ Wang Z S, Li Y T, Chen T, et al.Spatial-temporal characteristics of PM2.5 in Beijing in 2013[J]. Acta Geographica Sinica, 2015,70(1):110-120. ]
[33]	王振波,方创琳,许光,等.2014年中国城市PM2.5浓度的时空变化规律[J].地理学报,2015,70(11):1720-1734.
	[ Wang Z B, Fang C L, Xv G, et al.Spatial-temporal characteristics of PM2.5 in China in 2014[J]. Acta Geographica Sinica, 2015,70(11):1720-1734. ]
[34]	宓科娜,庄汝龙,梁龙武,等.长三角PM2.5时空格局演变与特征——基于2013-2016年实时监测数据[J].地理研究,2018,37(8):1641-1654.
	[ Mi K N, Zhuang R L, Liang L W, et al.Spatio-temporal evolution and characteristics of PM2.5 in the Yangtze River Delta based on real-time monitoring data during 2013-2016[J]. Geographical Research, 2018,37(8):1641-1654. ]
[35]	陈辉,厉青,李营,等.京津冀及周边地区PM2.5时空变化特征遥感监测分析[J].环境科学,2019,40(1):33-43.
	[ Chen H, Li Q, Li Y, et al.Monitoring and analysis of the spatio-temporal change characteristics of PM2.5 concentration over Beijing-Tianjin-Hebei and its surrounding regions based on remote sensing[J]. Environmental Science, 2019,40(1):33-43. ]
[36]	Kang H Q, Zhu B, Su J F, et al.Analysis of a long-lasting haze episode in Nanjing, China[J]. Atmospheric Research, 2013,120:78-87.
[37]	Zu Y, Huang L, Hu J, et al.Investigation of relationships between meteorological conditions and high PM10, pollution in a megacity in the western Yangtze River Delta, China[J]. Air Quality Atmosphere & Health, 2017,10(6):713-724.
[38]	Zu Y, Huang L, Hu J, et al.Investigation of relationships between meteorological conditions and high PM10, pollution in a megacity in the western Yangtze River Delta, China[J]. Air Quality Atmosphere & Health, 2017,10(6):713-724.
[39]	Hu J L, Wang Y G, Ying Q, et al.Spatial and temporal variability of PM2.5 and PM10 over the North China Plain and the Yangtze River Delta, China[J]. Atmospheric Environment, 2014,95(1):598-609. doi: 10.1016/j.atmosenv.2014.07.019
[40]	李宗恺,潘云仙,孙澜桥.空气污染气象学原理及应用[M].北京:气象出版社,1985.
	[ Li Z K, Pan Y X, Sun L Q.Principles and applications of air pollution meteorology[M]. Beijing: China Meteorological Press, 1985. ]
[41]	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. ]

	单位	均值	标准差
PM_2.5	μg/m³	54.75	36.88
PM₁₀	μg/m³	96.36	55.21
SO₂	μg/m³	19.08	10.02
CO	mg/m³	0.99	0.34
NO₂	μg/m³	48.07	18.72
O₃	μg/m³	103.55	51.48
日平均风速（WS）	m/s	2.68	1.79
日平均大气压（AP）	kPa	100.60	0.88
每日地表空气相对湿度（RH）	%	78.40	15.25
每日最高温度（MaxT）	℃	20.86	8.98
每日最低温度（MinT）	℃	13.41	8.89
每日总降水量（PR）	mm	4.10	11.29

IMF分量	IMF1	IMF2	IMF3	IMF4	IMF5	IMF6	IMF7	IMF8	IMF9	RES
周期/d	3	7	14	27	58	172	417	974	974
显著性检验/%	>95	>95	>95	>95	>95	>95	>95	>95	<50
贡献率/%	21.46	15.06	7.85	3.94	4.75	10.57	10.18	2.58	0.01	23.61

周期/d	PM_2.5	大气污染因子					气象因子
周期/d	PM_2.5	PM₁₀	SO₂	CO	NO₂	O₃	WS	AP	RH	MaxT	MinT	PR
天际尺度	3	3*	3*	3*	3*	3*	3*	4	3*	3*	3*	3*
	7	7*	7*	7*	7*	7*	7*	7*	7*	7*	7*	7*
	14	13	13	14*	14*	14*	13	15	15	14*	13	12
	27	27*	27*	29	26	28	26	29	32**	30	28	24
月际尺度	58	56	60	54	62	58*	54	52	60	55	56	47
	172	133	122	162	162	139	112	417	162	325	417	97
	417	417*	417*	325	417*	417*	195	584	417*	487	974	172
	974	974*	731	584	974*	974*	417	974*	974*	1461	1461	417
	2922	1461	1461	1461	1461	1461	1461	2922*	1461	1461	2922*	974

	PM_2.5与PM₁₀	PM_2.5与O₃	PM_2.5与MinT	PM_2.5与NO₂	PM_2.5与RH
原始数据	0.917**	-0.065*	-0.343**	0.618**	-0.254**
天际尺度	0.892**	0.271**	0.236**	0.529**	-0.039
月季尺度	0.939**	-0.431**	-0.628**	0.715**	-0.560**