首页 | 本学科首页   官方微博 | 高级检索  
     检索      

郑州市PM2.5浓度时空分布特征及预测模型研究
引用本文:陈强,梅琨,朱慧敏,蔡贤雷,张明华.郑州市PM2.5浓度时空分布特征及预测模型研究[J].中国环境监测,2015,31(3):105-112.
作者姓名:陈强  梅琨  朱慧敏  蔡贤雷  张明华
作者单位:温州医科大学水环境应用技术研究所, 浙江 温州 325035,温州医科大学水环境应用技术研究所, 浙江 温州 325035,温州医科大学水环境应用技术研究所, 浙江 温州 325035,温州医科大学水环境应用技术研究所, 浙江 温州 325035,温州医科大学水环境应用技术研究所, 浙江 温州 325035;加州大学戴维斯分校陆地、大气与水资源系, 美国戴维斯 CA 95616
摘    要:利用统计学原理和GIS技术,对郑州市2013年8月17—12月31日期间PM2.5浓度时空分布特征进行分析,同时结合气象资料与前一日污染数据,建立人工神经网络反向传播算法模型(BP-ANN)和多元线性回归模型用于该市细颗粒物污染的短期预测。结果表明,郑州市PM2.5浓度日变化呈单峰模式,随逆温现象的发生和交通的密集于上午11:00达到峰值,午后逐步下降。在工作日、周末与国庆节的对比中,国庆节期间颗粒物污染浓度高出平日32.8%,表明人为活动的加剧影响PM2.5的排放;周末与工作日期间无显著差异。在空间分布上,金水区、管城回族区污染最为严重,工业燃煤、地铁施工等源排放是造成污染的主要原因;位于远郊的岗里水库,受秸秆焚烧和市区污染输送等影响,PM2.5浓度亦维持较高水平。最后,研究将所构建的BP-ANN预测模型和多元线性回归模型对比,结果发现两模型在建模阶段预测值与真实值的拟合一致性指标分别为0.944、0.918,均方根误差分别为59.788、70.611;验证阶段拟合一致性指标分别为0.854、0.794,平均绝对误差分别为25.298、32.775,表明BP-ANN模型在预测郑州市PM2.5污染过程中更具优势。

关 键 词:PM2.5  预测模拟  BP-ANN模型  多元线性回归模型  GIS  郑州地区
收稿时间:2014/5/23 0:00:00
修稿时间:2014/10/15 0:00:00

Study on Spatiotemporal Variability of PM2.5 Concentrations and Prediction Model over Zhengzhou City
CHEN Qiang,MEI Kun,ZHU Hui-min,CAI Xian-lei and ZHANG Ming-hua.Study on Spatiotemporal Variability of PM2.5 Concentrations and Prediction Model over Zhengzhou City[J].Environmental Monitoring in China,2015,31(3):105-112.
Authors:CHEN Qiang  MEI Kun  ZHU Hui-min  CAI Xian-lei and ZHANG Ming-hua
Institution:Institute of Wenzhou Applied Technology in Environmental Research, Wenzhou Medical University, Wenzhou 325035, China,Institute of Wenzhou Applied Technology in Environmental Research, Wenzhou Medical University, Wenzhou 325035, China,Institute of Wenzhou Applied Technology in Environmental Research, Wenzhou Medical University, Wenzhou 325035, China,Institute of Wenzhou Applied Technology in Environmental Research, Wenzhou Medical University, Wenzhou 325035, China and Institute of Wenzhou Applied Technology in Environmental Research, Wenzhou Medical University, Wenzhou 325035, China;Dept. of Land, Air and Water Resources, College of Agricultural and Environmental Sciences, University of California Davis, CA 95616, USA
Abstract:This pape r is to identify the spatial and temporal patterns of PM2.5 concentrations and then attempt to model its distributions using Zhengzhou as an example. The spatial and temporal distributions of PM2.5 concentrations were analyzed using statistics theory and Geographic Information System (GIS) technology, from 17 August 2013 to 31 December 2013. Significant diurnal variations of PM2.5 concentrations were observed and showed a unimodal pattern with one marked peak at 11:00, due to the temperature inversion and dense traffic, and a declined trend in the afternoon. On National Day, particulate matter concentrations were found 32.8% higher than usual, suggesting the influence of intensification of anthropogenic activities on PM2.5 emissions; there was no significant differenc e between weekends and weekdays. The spatial distribution of PM2.5 concentration presented a severe pollution level both in Guancheng Hui District and Jinshui District, mainly due to the pollution source-emission such as fire coal and subway construction; High concentration of PM2.5 was also observed in Gangli Reservoir station which located in suburban area, due to external sources transport and the effect of straw burning. Finally, a back-propagation artificial neural network model (BP-ANN) and a multiple linear regression model were established and compared combined with meteorological data for a short-term estimation of fine particle pollution. The index of agreement of the two models' predicted and observed value in the modeling phase were 0.944, 0.918, root mean square error were 59.788, 70.611, respectively; in the validation phase the index of agreement were 0.854, 0.794, mean absolute error were 25.298, 32.775, respectively, which indicating that BP-ANN model has more advantages in predicting the process of PM2.5 pollution in Zhengzhou City.
Keywords:PM2  5  prediction model  BP-ANN model  multiple linear regression model  GIS  Zhengzhou
本文献已被 CNKI 等数据库收录!
点击此处可从《中国环境监测》浏览原始摘要信息
点击此处可从《中国环境监测》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号