气溶胶散射吸湿增长因子的BP神经网络模型

佟景哲; 倪长健; 米家媛; 蒋梦姣; 张莹; 石荞语

doi:10.13205/j.hjgc.202307018

气溶胶散射吸湿增长因子的BP神经网络模型

doi: 10.13205/j.hjgc.202307018

佟景哲^1,2,
倪长健^1,2,
米家媛^1,2,
蒋梦姣^1,2,
张莹^1,2,
石荞语^3, ,

1. 成都信息工程大学大气科学学院, 成都 610225;
2. 成都平原城市气象与环境四川省野外科学观测研究站, 成都 610225;
3. 中国气象局成都高原气象研究所, 成都 610072

基金项目:

四川省科技厅应用基础研发项目（2021YJ0314）；国家重点研发计划项目（2018YFC0214004，2018YFC1506006）

详细信息

作者简介:
佟景哲(1998-),男,硕士,主要研究方向为大气物理学与大气环境。tongjz1998@163.com

通讯作者:
石荞语(1994-),男,硕士,研究实习员,主要研究方向为大气物理学与大气环境。1803426355@qq.com

计量
- 文章访问数: 64
- HTML全文浏览量: 7
- PDF下载量: 2
- 被引次数: 0
出版历程
- 收稿日期: 2022-10-18

A BP NEURAL NETWORK MODEL OF AEROSOL SCATTERING HYGROSCOPIC GROWTH FACTOR

1. College of Atmospheric Science, Chengdu University of Information Technology, Chengdu 610225, China;
2. Chengdu Plain Urban Meteorology and Environment Observation and Research Station of Sichuan Province, Chengdu 610225, China;
3. Institute of Plateau Meteorology, China Meteorology of Administration, Chengdu 610072, China

摘要

摘要: 气溶胶组分、结构以及形态的复杂性对高湿条件下气溶胶散射吸湿增长因子统计模型的适用性提出了挑战。基于成都市2017年10—12月浊度计和黑碳仪的逐时观测资料，结合同时次的环境气象监测数据，利用"光学综合法"计算气溶胶散射吸湿增长因子。以相对湿度（RH）、C_BC、C_BC/C_{PM_2.5}、C_PM₁/C_{PM_2.5}以及C_{PM_2.5}/C_PM₁₀作为输入因子[C_BC、C_PM₁、C_{PM_2.5}、C_PM₁₀分别为黑碳（BC）、PM₁、PM_2.5、PM₁₀的质量浓度]，构建了气溶胶散射吸湿增长因子的BP神经网络模型。结果表明：气溶胶散射吸湿增长因子BP神经网络模型、多变量GAM模型、双变量模型以及单变量二次多项式模型的判决系数（R²）分别为0.870、0.792、0.744和0.650，其中高湿条件（RH>85%）下模拟值的判决系数（R²）分别为0.749、0.685、0.638和0.538。多模型对比表明，气溶胶散射吸湿增长因子BP神经网络模型的模拟效果最优，显著降低了统计模型在高湿条件（RH>85%）下的模拟误差。
- 气溶胶 /
- 散射吸湿增长因子 /
- 人工神经网络 /
- BP模型 /
- 成都
Abstract: The complexity of aerosol components, structure and morphology challenges the applicability of statistical models for aerosol scattering hygroscopic growth factor under high humidity conditions. Based on the hourly observational data of nephelometer and aethalometer, as well as the simultaneous monitored data of environmental meteorology from October to December, 2017 in Chengdu, the aerosol scattering hygroscopic growth factor was calculated by optical synthesis method. The relative humidity (RH), C_BC, C_BC/C_{PM_2.5}, C_PM₁/C_{PM_2.5} and C_{PM_2.5}/C_PM₁₀ were used as input factors (C_BC, C_PM₁, C_{PM_2.5} and C_PM₁₀ representing mass concentrations of BC, PM₁, PM_2.5 and PM₁₀ respectively), and a BP neural network model of aerosol scattering hygroscopic growth factor was proposed. The comparison results of multiple models showed that:the corresponding determination coefficient (R²) of the univariate quadratic polynomial model, bivariate model, multivariate GAM model and BP neural network model for the aerosol hygroscopic growth factor were 0.650, 0.744, 0.792 and 0.870 respectively, and the corresponding determination coefficient (R²) for the simulated values under high humidity conditions of RH>85% were 0.538, 0.638, 0.685 and 0.749, respectively. The BP neural network model of aerosol scattering hygroscopic growth factor achieved the best fitting effect and the simulation error of aerosol scattering hygroscopic growth factor under high humidity condition (RH>85%) was significantly reduced.
- aerosol /
- scattering hygroscopic growth factor /
- artificial neural network /
- BP model /
- Chengdu

HTML全文

参考文献(40)

[1]	孙俊英,张璐,沈小静, 等.大气气溶胶散射吸湿增长特性研究进展[J].气象学报,2016,74(5):672-682.
[2]	ZHANG Q N, ZHAO L J, CHEN S H, et al.Hygroscopic property of inorganic salts in atmospheric aerosols measured with physisorption analyzer[J].Atmospheric Environment, 2020, 247:118171.
[3]	HE Q S, ZHOU G Q, GENG F H, et al.Spatial distribution of aerosol hygroscopicity and its effect on PM_2.5 retrieval in East China[J].Atmospheric Research, 2016, 170:161-167.
[4]	LIU X G, ZHANG Y H.Research on the measurement of aerosol hygroscopic growth factor in Guangzhou city[J].China Environmental Science, 2009, 29(11):1128-1133.
[5]	WU Y F, WANG X J, YAN P, et al.Investigation of hygroscopic growth effect on aerosol scattering coefficient at a rural site in the southern North China Plain[J].Science of the Total Environment, 2017, 599/600:76-84.
[6]	TANG I N.Chemical and size effects of hygroscopic aerosolson light scattering coefficients[J].Journal of Geophysical Research, 1996, 101(D4):19245-19250.
[7]	TANG I N.Thermodynamic and optical properties of mixed-salt aerosols of atmospheric importance[J].Journal of Geophysical Research, 1997, 102:1883-1893.
[8]	谭天怡, 郭松, 吴志军, 等.老化过程对大气黑碳颗粒物性质及其气候效应的影响[J].科学通报, 2020, 65(36):4235-4250.
[9]	KHALIZOV A F, XUE H X, WANG L, et al.Enhanced light absorption and scattering by carbon soot aerosol internally mixed with sulfuric acid[J].Journal of Physical Chemistry A, 2009, 113(6):1066-1074.
[10]	曾晨.黑碳混合-吸湿对其光学和辐射特性的影响[D].南京:南京信息工程大学,2019,18-23.
[11]	MAGI B I, HOBBS P V.Effects of humidity on aerosols in Southern Africa during the biomass burning season[J].Journal of Geophysical Research Atmospheres, 2003, 108(D13):8495.
[12]	SONG C H, PARK M E, LEE K H, et al.An investigation into seasonal and regional aerosol characteristics in East Asia using model-predicted and remotely-sensed aerosol properties[J].Atmospheric Chemistry and Physics, 2008, 8(22):6627-6654.
[13]	CHEN J C, ZHAO C S, MA N, et al.Aerosol hygroscopicity parameter derived from the light scattering enhancement factor measurements in the North China Plain[J].Atmospheric Chemistry and Physics, 2014, 14(15):8105-8118.
[14]	尹单丹, 倪长健, 邓也, 等.成都秋冬季气溶胶散射吸湿增长模型研究[J].生态与农村环境学报, 2020, 36(4):542-548.
[15]	张城语,倪长健,佟景哲,等.成都地区气溶胶散射吸湿增长因子双变量模型[J].中国环境科学,2021,41(12):5467-5475.
[16]	佟景哲,倪长健,杜云松, 等.成都气溶胶散射吸湿增长因子多变量影响分析[J].生态与农村环境学报,2022,38(5):636-644.
[17]	莫祖斯,卜令兵,王勤,等.基于GRNN神经网络模型结合气溶胶消光系数和气象要素评估颗粒物质量浓度[J].中国激光,2022,49(17):140-150.
[18]	孙宝磊,孙暠,张朝能,等.基于BP神经网络的大气污染物浓度预测[J].环境科学学报,2017,37(5):1864-1871.
[19]	郭庆春,何振芳,寇立群,等.BP神经网络在北京市API预报中的应用[J].环境工程,2011,29(4):106-108.
[20]	KOSCHMIEDER H.Therie der horizontalen sichtweite[J].Beitragezur Physik der freien Atmosphare, 1924:33-53.
[21]	孙景群.能见度与相对湿度的关系[J].气象学报,1985(2):230-234.
[22]	杨寅山, 倪长健, 邓也, 等.成都市冬季大气消光系数及其组成的特征研究[J].环境科学学报, 2019, 39(5):1425-1432.
[23]	伯广宇, 刘东, 吴德成, 等.双波长激光雷达探测典型雾霾气溶胶的光学和吸湿性质[J].中国激光, 2014, 41(1):207-212.
[24]	BODHAINE B A.Aerosol absorption measurements at Barrow, Mauna Loa and the south pole[J].Journal of Geophysical Research, 1995, 100(D5):8967-8975.
[25]	李梅芳, 叶芝祥.基于太阳光度计的成都双流地区夏季气溶胶光学特性研究[J].成都信息工程学院学报, 2014, 29(2):213-216.
[26]	吴兑,毛节泰,邓雪娇,等.珠江三角洲黑碳气溶胶及其辐射特性的观测研究[J].中国科学(D辑:地球科学),2009,39(11):1542-1553.
[27]	BERGSTROM R W, RUSSELL P B, HIGNETT P.Wavelength dependence of the absorption of black carbon particles:predictions and results from the TARFOX experiment and implications for the aerosol single scattering albedo[J].Journal of the Atmospheric Sciences, 2002,59(3):567-577.
[28]	YAN P, TANG J, HUANG J, et al.The measurement of aerosol optical properties at a rural site in Northern China[J].Atmospheric Chemistry and Physics, 2008,8(8):2229-2242.
[29]	颜鹏,刘桂清,周秀骥,等.上甸子秋冬季雾霾期间气溶胶光学特性[J].应用气象学报,2010,21(3):257-265.
[30]	PENNDORF R.Tables of the refractive index for standard air and the rayleigh scattering coefficient for the spectral region between 0.2 and 20.0μ and their application to atmospheric optics[J].Journal of the Optical Society of America, 1957,47(2):176-182.
[31]	HODKINSON R J.Calculations of color and visibility in urban atmospheres polluted by gaseous NO₂[J].International Journal of Air Pollution, 1966, 10:137-144.
[32]	张雪慧,官莉,王振会,等.利用人工神经网络方法反演大气温度廓线[J].气象,2009,35(11):137-142.
[33]	申浩洋,韦安磊,王小文,等.BP人工神经网络在环境空气SO₂质量浓度预测中的应用[J].环境工程,2014,32(6):117-121.
[34]	OLDEN J D, JACKSON D A.Illuminating the "black box":a randomization approach for understanding variable contributions in artificial neural networks[J].Ecological Modelling, 2002, 154:135-150.
[35]	CAPPA C D, ZHANG X L, RUSSELL L M, et al.Light absorption by ambient black and brown carbon and its dependence on black carbon coating state for two California, USA, cities in winter and summer[J].Journal of Geophysical Research:Atmospheres, 2019, 124:1550-1557.
[36]	LIU S, AIKEN A C, GORKOWSKI K, et al.Enhanced light absorption by mixed source black and brown carbon particles in UK winter[J].Nature Communications, 2015, 6:8435.
[37]	姚婷婷, 黄晓锋, 何凌燕, 等.深圳市冬季大气消光性质与细粒子化学组成的高时间分辨率观测和统计关系研究[J].中国科学:化学, 2010, 40(8):1163-1171.
[38]	张智察,倪长健,刘新春,等.干气溶胶复折射率的参数化方案[J].中国环境科学,2021,41(2):580-587.
[39]	刘凡, 谭钦文, 江霞, 等.成都市冬季相对湿度对颗粒物浓度和大气能见度的影响[J].环境科学, 2018, 39(4):1466-1472.
[40]	常永波, 陈新军.基于GAM和BP模型的智利竹筴鱼资源量影响因子与模型比较[C]//"一带一路"战略与海洋科技创新——中国海洋学会2015年学术论文集,2015:290-296.