基于BP人工神经网络法沈阳市PM2.5质量浓度集成预报试验

doi:10.3969/j.issn.1673-503X.2018.02.013

气象与环境学报 ›› 2018, Vol. 34 ›› Issue (2): 100-106.doi: 10.3969/j.issn.1673-503X.2018.02.013

基于BP人工神经网络法沈阳市PM_2.5质量浓度集成预报试验

李晓岚¹, 刘旸², 栾健³, 马雁军¹, 王扬锋¹, 张婉莹⁴

1. 中国气象局沈阳大气环境研究所, 辽宁沈阳 110166;
2. 辽宁省人工影响天气办公室, 辽宁沈阳 110166;
3. 中国气象局气象干部培训学院辽宁分院, 辽宁沈阳 110166;
4. 辽宁省气象服务中心, 辽宁沈阳 110166

收稿日期:2016-12-15 修回日期:2017-03-07 出版日期:2018-04-30 发布日期:2018-04-30
通讯作者: 马雁军,E-mail:mayanjun0917@163.com E-mail:mayanjun0917@163.com
作者简介:李晓岚,女,1986年生,副研究员,主要从事大气边界层与大气湍流、大气环境研究,E-mail:leexl.ouc@163.com。
基金资助:
国家重点研发计划课题（2016YFC0203304）、辽宁省气象局科学技术研究项目（博士科研专项）（D201603）、国家科技支撑计划课题（2014BAC16B04）、公益性行业（气象）科研专项经费（GYHY201406031）和中央级公益性科研院所基本科研业务费专项（2016SYIAEZD3）共同资助。

Integration forecast experimentation for PM_2.5 mass concentration in Shenyang based on BP artificial neural network

LI Xiao-lan¹, LIU Yang², LUAN Jian³, MA Yan-jun¹, WANG Yang-feng¹, ZHANG Wan-ying⁴

1. Institute of Atmospheric Environment, CMA, Shenyang 110166, China;
2. Liaoning Weather Modification Office, Shenyang 110166, China;
3. Liaoning Branch of China Meteorological Administration Training Centre, Shenyang 110166, China;
4. Liaoning Meteorological Service Center, Shenyang 110166, China

Received:2016-12-15 Revised:2017-03-07 Online:2018-04-30 Published:2018-04-30

摘要/Abstract

摘要： 基于CUACE（CMA Unified Atmospheric Chemistry Environment）和CMAQ（Community Multiscale Air Quality）空气质量模式预报产品，应用BP（Back-Propagation）人工神经网络法建立沈阳市不同地点小风和高湿条件下PM_2.5浓度集成预报模型，并对预报结果进行检验。结果表明：与单一空气质量模式相比，集成模型预报的PM_2.5浓度更接近实测值，预报的PM_2.5浓度的平均偏差和归一化均方误差均明显减小，预报的PM_2.5浓度的模拟值在观测值两倍范围内的百分比（FAC2）明显提高。集成模型能较好地预报PM_2.5浓度高值的变化，且显著提高了沈阳市外围城区PM_2.5浓度的预报水平。集成预报模型可以实现CUACE和CMAQ两种空气质量模式产品的最优综合，对空气质量的实时预报具有一定的参考价值。

关键词: PM_2.5质量浓度, 集成预报, CUACE, CMAQ, BP神经网络

Abstract: Based on the forecasting products of CUACE (CMA Unified Atmospheric Chemistry Environment) and CMAQ (Community Multiscale Air Quality) models,integration forecast models for PM_2.5 at different positions in Shenyang under conditions of small wind speed and high relative humidity were developed and validated using an artificial neural network method with back-propagation (BP) algorithm.The results indicate that PM_2.5 concentrations predicted by integration models are much closer to their observational values than those predicted by CUACE and CMAQ.The values of mean deviation and NMSE (Normalized Mean Square Error) of modelling results decrease significantly,and the values of FAC2 increase obviously.The PM_2.5 forecast from integration models can better reflect the variation of high PM_2.5 concentrations,and its development at surrounding urban areas of Shenyang is significant.The integration models based on BP artificial neural network are a kind of effective method for PM_2.5 forecast,which can provide a reference to the real-time operational forecast of air quality.

Key words: PM_2.5 mass concentration, Integration forecast, CMA Unified Atmospheric Chemistry Environment (CUACE) model, Community Multiscale Air Quality (CMAQ) model, Back-propagation neural network

中图分类号:

P457.7

李晓岚, 刘旸, 栾健, 马雁军, 王扬锋, 张婉莹. 基于BP人工神经网络法沈阳市PM_2.5质量浓度集成预报试验[J]. 气象与环境学报, 2018, 34(2): 100-106.

LI Xiao-lan, LIU Yang, LUAN Jian, MA Yan-jun, WANG Yang-feng, ZHANG Wan-ying. Integration forecast experimentation for PM_2.5 mass concentration in Shenyang based on BP artificial neural network[J]. Journal of Meteorology and Environment, 2018, 34(2): 100-106.

0
/ / 推荐

导出引用管理器 EndNote|Reference Manager|ProCite|BibTeX|RefWorks

链接本文: http://www.jme1984.net.cn/CN/10.3969/j.issn.1673-503X.2018.02.013

http://www.jme1984.net.cn/CN/Y2018/V34/I2/100

参考文献

[1] 马雁军,刘宁微,王扬锋,等.辽宁中部城市群大气环境研究进展[J].气象科技进展,2012,2(2):19-24.
[2] Cheng Y,He K B,Du Z Y,et al.Humidity plays an im-portant role in the PM_2.5 pollution in Beijing[J].Envi-ronmental Pollution,2015,197(2):68-75.
[3] Kim Y J,Kim K W,Kim S D,et al.Fine particulate matter characteristics and its impact on visibility impair-ment at two urban sites in Korea:Seoul and Incheon[J].Atmospheric Environment,2006,40(S2):593-605.
[4] Zhao H J,Che H Z,Zhang X Y,et al.Characteristics of visibility and particulate matter (PM) in an urban area of Northeast China[J].Atmospheric Pollution Re-search,2013,4(4):427-434.
[5] Limin S H,Takahashi H,Usup A D,et al.Impacts of haze in 2002on social activity and human health in Palangka Raya[J].Tropics,2007,16(3):275-282.
[6] 杜钧.集合预报的现状和前景[J].应用气象学报,2002,13(1):16-28.
[7] 黄思,唐晓,徐文帅,等.利用多模式集合和多元线性回归改进北京PM₁₀预报[J].环境科学学报,2015,35(1):56-64.
[8] 刘永和,严中伟,冯锦明,等.基于TIGGE资料的沂沭河流域6小时降水集合预报能力分析[J].大气科学,2013,37(3):539-551.
[9] 马学款,普布次仁,唐叔乙,等.人工神经网络在西藏中短期温度预报中的应用[J].高原气象,2007,26(3):491-495.
[10] 李倩,胡邦辉,王学忠,等.基于BP人工神经网络的区域温度多模式集成预报试验[J].干旱气象,2011,29(2):231-235,250.
[11] 胡江林,张礼平,宇如聪.神经网络模型预报湖北汛期降水量的应用研究[J].气象学报,2001,59(6):776-783.
[12] 农吉夫,金龙.基于MATLAB的主成分RBF神经网络降水预报模型[J].热带气象学报,2008,24(6):713-717.
[13] 刘旸,官莉.人工神经网络法反演晴空大气湿度廓线的研究[J].气象,2011,37(3):318-324.
[14] Sohn S H,Oh S C,Jo B W,et al.Prediction of ozone formation based on neural network[J].Journal of Envi-ronmental Engineering,2000,126(8):688-696.
[15] Mok K M,Tam S C.Short-term prediction of SO₂ concentration in Macau with artificial neural networks[J].Energy and Buildings,1998,28(3):279-286.
[16] 马雁军,杨洪斌,张云海.BP神经网络法在大气污染预报中的应用研究[J].气象,2003,29(7):49-52.
[17] 白鹤鸣,沈润平,师华定,等.基于BP神经网络的空气污染指数预测模型研究[J].环境科学与技术,2013,36(3):186-189.
[18] Hecht-Nielsen R.Theory of the backpropagation neural network[C]//Proceedings of International Joint Confer-ence on Neural Networks.Washington,DC,USA:IEEE,1989:593-605.
[19] Hanna S R.Uncertainties in air quality model predic-tions[J].Boundary-Layer Meteorology,1993,62(1/4):3-20.
[20] 李晓岚,马雁军,王扬锋,等.基于CUACE系统沈阳地区春季空气质量预报的校验及修正[J].气象与环境学报,2016,32(6):10-18.
[21] 马雁军,王江山,王扬锋,等.辽宁中部城市群可吸入颗粒物PM₁₀和PM_2.5的污染特征研究[J].气象与环境学报,2008,24(5):11-15.
[22] 洪也,马雁军,韩文霞,等.沈阳市冬季大气颗粒物元素浓度及富集因子的粒径分布[J].环境科学学报,2011,31(11):2336-2346.
[23] Zhao H J,Che H Z,Ma Y J,et al.Temporal variability of the visibility,particulate matter mass concentration and aerosol optical properties over an urban site in Northeast China[J].Atmospheric Research,2015,166(13):204-212.
[24] Hu X M,Zhang Y,Jacobson M Z,et al.Coupling and e-valuating gas/particle mass transfer treatments for aero-sol simulation and forecast[J].Journal of Geophysical Research:Atmospheres,2008,113(D11):D11208.

基于BP人工神经网络法沈阳市PM_2.5质量浓度集成预报试验

Integration forecast experimentation for PM_2.5 mass concentration in Shenyang based on BP artificial neural network

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 9

Metrics

本文评价

推荐阅读 0

[1]	王新伟, 张宇星, 魏芳芳. 山岳型风景区最高最低气温预报方法研究[J]. 气象与环境学报, 2019, 35(2): 92-96.
[2]	刘博;党冰;张楠;王式功;尹岭;张晓云; 黎檀实;卢震华. 多种气象统计模型对比研究——以气象敏感性疾病脑卒中预报为例[J]. 气象与环境学报, 2018, 34(4): 126-133.
[3]	王焕毅, 谭政华, 杨萌, 张翘, 蒋林杉. 三种数值模式气温预报产品的检验及误差订正方法研究[J]. 气象与环境学报, 2018, 34(1): 22-29.
[4]	李红斌, 傅瑜, 张靖萱, 赵繁盛, 盖晓波, 何阳. 大连市冬季大气污染数值模拟及其对人工增雨(雪)作业的指示作用[J]. 气象与环境学报, 2017, 33(5): 10-16.
[5]	杨关盈, 邓学良, 吴必文, 霍彦峰, 翟菁, 于彩霞, 田磊. 基于CUACE模式的合肥地区空气质量预报效果检验[J]. 气象与环境学报, 2017, 33(1): 51-57.
[6]	李晓岚, 马雁军, 王扬锋, 赵胡笳, 刘洪利. 基于CUACE系统沈阳地区春季空气质量预报的校验及修正[J]. 气象与环境学报, 2016, 32(6): 10-18.
[7]	张敏,边海,韩素芹,张华,崔振雷,朱彬. INTEX-B NO₂排放源在中国地区的应用与检验[J]. 气象与环境学报, 2012, 28(1): 65-70.
[8]	杨永生,何平. 投影寻踪回归与BP神经网络方法在前汛期降水预测中的比较研究[J]. 气象与环境学报, 2008, 24(1): 14-17.
[9]	陈静,郭卫红,王素丽. 石家庄市空气污染预警天气背景分析及模型的建立[J]. 气象与环境学报, 2007, 23(6): 62-67.

基于BP人工神经网络法沈阳市PM2.5质量浓度集成预报试验

Integration forecast experimentation for PM2.5 mass concentration in Shenyang based on BP artificial neural network

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 9

Metrics

本文评价

推荐阅读 0

基于BP人工神经网络法沈阳市PM_2.5质量浓度集成预报试验

Integration forecast experimentation for PM_2.5 mass concentration in Shenyang based on BP artificial neural network