基于傅里叶频谱的臭氧干沉降多时间尺度模拟误差及来源诊断

Ozone at surface is a pollutant, harmful to both human health and ecosystem degradation. Due to a large number of error sources in complex models, current global or regional scale atmospheric chemistry models contain large uncertainties in applications. A comparison between observation and model output introduces an important way to evaluating model error and identifying its sources. Operational evaluation using statistical metrics neglects nonlinear interaction of error sources and the effect of random error, and are rooted in linear regression analysis and the assumption of normally distributed residuals, which has been proven to be unreliable. The usage of probabilistic methodologies such as sensitivity analysis and uncertainty analysis in identifying uncertainty sources, requires large computational resources, and has difficulties in identifying error sources due to lack of observational constraints. Spectral analysis serves as a useful tool to decompose the modelled and observed time series from noise, to effectively distinguish between multi-timescale patterns of error components (e.g., bias, variance and covariance), thus allowing for a clearer diagnosis of the processes that caused the error at multiple timescales. In this study, we established a methodology for error diagnostics based on spectrum, through comparison between measurement and model results from the regional scale chemistry transport modeling systems, apportioned error components on a range of timescales in , thereby identify the timescale at which it is most relevant and, when possible, to infer which process/es could have generated it. Information about the nature of error and the relevant process can significantly help modelers and developers to improve model performance.

O3作为一种重要大气污染物对人类健康和生态系统损害作用。由于复杂模式误差来源众多，现有的区域和全球大气数值模式对O3干沉降模拟的不确定性较高。模式模拟和外场观测结果的对比，是评估复杂模式模拟误差和识别其来源的主要途径。基于误差统计指标的模式评估忽略了误差来源的非线性交互作用和随机误差的影响，假定模拟与观测的线性响应和残差服从正态分布，其结果往往缺乏可靠性。敏感性分析和不确定性诊断在量化不确定来源方面所需的计算资源巨大，因无需观测资料约束往往无法提供误差来源信息。频谱分析，可以有效过滤模拟与观测时间序列中的随机噪声，便于识别多时间尺度误差和相应尺度下各大气过程模拟误差的关联。本项目拟建立一种基于频谱分析的误差诊断方法，量化区域模式O3干沉降模拟在多时间尺度上误差的分配，诊断关键时间尺度模拟误差，揭示造成误差的大气过程来源，为模拟精度提高和模式改进提供有效信息。

O3作为一种重要大气污染物对人类健康和生态系统损害作用。由于复杂模式误差来源众多 ，现有的区域和全球大气数值模式对O3干沉降模拟的不确定性较高。与误差统计指标、敏感性分析和不确定性分析等评估方法相比，频谱分析可以有效过滤模拟与观测时间序列中的随机噪声，靶向性指导多时间尺度误差来源诊断，实现相应尺度下关联大气过程模拟误差来源识别和量化，从而避免过多的计算资源浪费。本项目建立一种基于频谱分析的误差诊断方法，量化区域模式O3干沉降模拟在多时间尺度上误差的分配，诊断关键时间尺度模拟误差，揭示造成误差的大气过程来源。项目针对提高区域大气数值模型模拟精度的热点问题，研究多尺度关键大气过程对我国臭氧污染的影响，获得了较为普适性的科学结论和定量表达。

内容获取失败，请点击重试