Temporal-Spatial Evolutions of Low-Frequency Climate Variability and Warming Trend

本项目的目标是：(1)从空间和时间上适应地、局地分离变化的全球变暖趋势和多年代际气候变率；（ii）揭示1850年至今观测到的一些低频气候模态，如太平洋多年代际变率、大西洋多年代际变率和北大西洋涛动的时间演变特征；（iii）将这些特征与气候模式模拟中对应的低频气候模态进行比较；(4)寻找低频气候模态演化的物理机制。将使用的主要工具是多维集成经验模态分解（MEEMD） （MEEMD），由PI在不到一年前开发。MEEMD是一种新的扩展稀疏时间序列分解方法，没有先验确定的基础，经验模式分解，用于时空多维数据分析。通过对历史气候场和耦合模式比对项目（CMIP3/CMIP5）模式的模拟，研究了一个半世纪以来气候的多年代际变率和趋势演变的时空特征，以及模式模拟中的对应特征。这将为进一步改进气候模式提供见解和线索。该研究具有以下更广泛的影响：(i)自适应和时空局部分析方法的改进可能在涉及数据分析的其他科学、工程或生物医学领域被证明是有用的；（ii）在数据分析理论和方法方面培训未来的科学家和技术专业人员；（iii）研究将有助于传播本项目开发的知识、技术和工具，通过在开放文献中发表，并使新的软件工具开放源代码，供更广泛的社区使用。

This project has the following goals: (i)to adaptively and locally separate the varying global warming trend and the multi-decadal climate variability, both spatially and temporally; (ii)to reveal the characteristics of temporal evolution from 1850 to present of some observed low-frequency climate modes, e.g., Pacific multidecadal variability, Atlantic multidecadal variability, and the North Atlantic Oscillation; (iii) to compare these characteristics to their counterparts of low-frequency climate modes in climate model simulations; and (iv)to search for physical mechanisms of the evolution of low frequency climate modes. The main tool that will be used is the Multi-dimensional Ensemble Empirical Mode Decomposition (MEEMD) (MEEMD), developed less than one year ago by the PI. The MEEMD is a novel extension of a sparse time series decomposition method without a priori determined bases, the Empirical Mode Decomposition, for spatial-temporal multi-dimensional data analysis. By applying the method to historical climate fields and Coupled Model Intercomparison Project (CMIP3/CMIP5) model simulations, the researcher will examine spatial-temporal characteristics of the evolutions of multidecadal variability and trend of the climate of the past one and a half centuries, as well as their counterparts in model simulations. This will provide insights and clues for further improvement of climate models.The research has the following broader impact: (i) improvement of the adaptive and spatial-temporally local analysis method may prove useful in other scientific, engineering, or biomedical field that involves data analysis; (ii) there is training of future scientists and technology professionals in both data analysis theory and methods; and (iii) the research will help to disseminate the knowledge, techniques and tools developed in this project through publishing in the open literature, and making new software tools open source for use by a wider community.