• 关键词：大气污染物  小波分析  时间序列  多分辨率

• 基金项目：

• 徐鸣
• 1. 四川大学环境科学与工程系, 成都 610065; 2. 乌鲁木齐环境监测中心站, 乌鲁木齐 830000

• 王斌
• 四川大学环境科学与工程系, 成都 610065

• 吕爱华
• 乌鲁木齐环境监测中心站, 乌鲁木齐 830000

• 赵柳生
• 乌鲁木齐环境监测中心站, 乌鲁木齐 830000

• 梅自良
• 西华大学能源与环境学院, 成都 610039

• 摘要：利用Morlet小波函数进行小波变换,利用Daubechies小波函数进行分解及滤波,以乌鲁木齐市的大气污染物时间序列为例,分析了当地单个大气监测站点NO₂的多分辨率的演变特性,并对NO₂的时间序列进行了滤波消噪.研究结果显示,NO₂年际变化和季节性变化的时间尺度在尺度空间中分布不均匀,具有较明显的局部化特征;除了季节性变化的时间尺度和变化趋势之外,还存在大气污染物中、长期变化趋势.基于滤波消噪后的时间序列建立的预测模型可用于有效地提高大气污染物的预测预报精度.

• Abstract：As an example of an air pollutant time series, the evolution characteristics of local NO₂ at a single air monitoring site in Urumchi were analyzed based on a wavelet transform with a Morlet wavelet function at multi-timescale resolution. Atime series of NO₂ data was filtered and denoised by decomposition and filtering with the Daubechies wavelet function. The results indicated that the variance distributions on multi-timescales for inter-years or inter-seasons were asymmetric, and possessed distinct local characteristics. Besides seasonal time scales or trends, the data also showed some medium-term and long-term variation trends for air pollutant concentration. Anew forecasting model was set up based on the filtered and denoised time series, which could effectively improve the accuracy of prediction for air pollutants.

• Key words：air pollutant  wavelet transform  time series  multi-timescales resolution

• 
  摘要点击次数： 2485 全文下载次数： 2959