大气污染物浓度时间序列中污染源及气象因素贡献的解析方法研究

Emission changes in sources/ precursors due to PM and O3 pollution control strategy are difficult to directly quantize. The benefits of pollution control are usually gauged by comparative analysis of the ambient concentrations before and after the implement of pollution control. However, the changes in ambient concentrations can hardly objectively reflect overall the emissions-related effectiveness of air pollution control regulations in the presence of meteorological variability. Time series of air pollutant concentrations include two signals: changes in source emissions and meteorological conditions. Because changes in ambient concentrations due to meteorological fluctuations are always greater than those due to changes in emissions, the effect of emissions is masked. Therefore, in such circumstances that emission changes are hard to be measured, it is very important to block the effect of meteorological factors and extract the emissions-relate signals in the time series of PM and O3 concentrations...In this project, filter analysis, neural network, semiparametric regression, and wavelet reconstruction are proposed to be applied to analyze the characteristics of time-frequency in time series of pollutant concentrations and key meteorological variables and to built the contribution apportionment model, which can separate the meteorological signal and emission-related signal mixing in the pollutant concentrations data and examine the effects of two signals respectively. The model can apportion the contributions to changes in pollutant concentrations between meteorological effects and emissions-related changes due to pollution control or increase of energy consumption, etc. Detecting the underlying emissions-related trends of pollutant concentrations will help to objectively evaluate the effectiveness of pollution control policy and make better air quality management decisions for the future.

对PM10或臭氧等污染物实施控制后所致的污染源（或前体物）排放的变化难以直接量化，一般通过对比控制措施实施前后污染物环境浓度的差异来反映污染控制的环境效果。但在气象因素的影响下，污染物环境浓度的变化一般很难全面客观反映污染控制措施是否有效。因此，在污染源排放变化难以定量的情况下，如何从大气污染物环境浓度变化的信息中屏蔽气象因素的影响，提取污染源排放变化的信息，具有重要研究意义。.本项目拟采用滤波分析、神经网络、半参数回归及小波重构等方法，研究大气污染物浓度及关键气象因子时间序列的频谱特征，建立解析模型，将混杂在污染物环境浓度时间序列中的气象信息与污染源信息分离，分别提取并评估人为因素（污染控制或能源消费增长等所致的污染源排放改变）与气象因素对环境空气质量变化（改善或恶化）所起的作用和贡献。研究成果将有助于更客观地认识污染控制的环境效果，为后续的环境管理决策提供科学依据。

污染控制的环境效果往往通过对比控制措施实施前后污染物环境浓度的差异来反映。但在气象因素的影响下，污染物环境浓度的变化一般很难全面客观反映污染控制措施是否有效。气象条件变化所致的大气污染物浓度变化的幅度往往大于由污染源排放变化引起的浓度变化的幅度，污染源排放变化的信息往往被“淹没”。因此，如何从大气污染物环境浓度变化的信息中屏蔽气象因素的影响，提取污染源排放的贡献，具有重要研究意义。本项目利用滤波分析、神经网络、小波重构等方法，研究大气污染物浓度及关键气象因子时间序列的频谱特征，建立解析模型，将混杂在污染物环境浓度时间序列中的气象信息与污染源信息分离，分别提取并定量评估污染源控制及气象因素对环境空气质量变化所起的作用和贡献。本研究建立的技术方法在天津市的应用研究表明，2010年及2013年后，气象条件总体上不利于SO2扩散，可使SO2及PM10日均浓度长期分量最大分别增加5μg/m3、18μg/m3左右。气象因素对NO2浓度的影响也有类似的规律。项目的研究成果可以为更客观地分析环境空气质量变化的原因、更有效地评估污染控制的环境绩效提供技术方法。也可为大气环境数值模拟中气象基准条件的确定提供参考。