受全球变暖影响，极端性天气事件和气候灾害频发，给自然生态和人类社会的可持续发展带来了严峻挑战。人类活动排放大量CO2是引起全球变暖的主要原因。卫星观测发现大气CO2浓度分布存在非均匀分布，在时间上有年际和季节性的差异，在空间上也存在着显著差异。部分空间上差异与大气环流遥相关有关。以往关于遥相关研究多集中于经验正交函数分解（EOF）和统计相关分析，前者只发现了一些最强的信号，而对这些信号的物理解释困难，后者研究限定在有限数量的CO2监测点尺度或者区域尺度。由于气候系统多个遥相关之间的相互依赖性，地区遥相关的分析不能表示全球尺度的遥相关，全球大气CO2浓度变化整体上的遥相关仍未得到清楚的认识。卫星观测发现大气CO2浓度存在非均匀分布，而国际上关于大气CO2浓度和地表温度之间关系的研究都是基于全球CO2浓度均匀分布这一假设。近年来，复杂网络理论在分析气候系统中复杂的非线性作用上取得了丰硕的成果，在描述地球系统内部相互关系作用方面具有优越性，这为大气CO2浓度遥相关路径分析及CO2浓度非均匀分布与地表温度之间关系的研究提供了新的思路。 本文围绕CO2浓度非均匀分布及其与地表温度关系这一主题，基于复杂网络方法，以CO2浓度网络和地表温度网络为核心，在CO2浓度非均匀分布特征、CO2浓度遥相关路径、地表温度遥相关路径以及CO2浓度和地表温度遥相关路径等方面开展了相关研究。本论文的研究工作对正确理解CO2的气候效应以及积极应对气候变暖具有科学的指导意义，对降低灾害风险以及未来国际气候变化协议的实施也有着重要的现实意义。本文主要研究工作如下： 第一，从时间和空间维度上开展了全球遥感卫星CO2浓度非均匀分布特征研究，并进一步探究柱总层与对流层中层大气CO2浓度的空间差异。 基于GOSAT和AIRS卫星探测的大气CO2浓度数据，分别分析了大气CO2浓度的时间和空间分布变化特征以及这两种卫星反演的大气CO2浓度的差异。结果表明，全球总柱层CO2浓度的空间分布受到人类活动和自然过程CO2排放的强烈影响，而对流层中层CO2的空间特征既与地表来源有关，也与大规模环流系统有关。二者（GOSAT–AIRS）空间差异处于2–10 ppm之间，差异较大的区域分布在北极地区，尤其是格陵兰岛。季节上看，春季GOSAT CO2浓度高值区集中东亚、南亚、北美洲东北部以及太平洋地区，AIRS CO2浓度高值区集中于北半球中高纬度地区；夏季GOSAT CO2浓度明显下降，高值区主要集中于赤道非洲和南美洲中部地区，AIRS CO2浓度高值区集中于北半球中高纬度地区；秋季GOSAT CO2浓度高值区分布特征和夏季类似，AIRS CO2浓度高值区集中于南非南部、澳大利亚东南部延至南太平洋中部；冬季GOSAT CO2浓度高值区分布在东亚、北太平洋、北美洲东部、西欧、西亚和赤道非洲区域，AIRS CO2高值区分布在北半球的中纬度地区。从时间序列分析发现，GOSAT CO2月平均浓度始终低于AIRS CO2浓度，且在春季最高，夏季最低。 第二，构建大气CO2浓度网络，分析大气CO2浓度遥相关路径。 基于复杂网络方法，分析不同区域CO2浓度之间存在的相关关系及其时滞，通过打散重排的显著性测试，分析CO2浓度中显著的连接。应用节点加权入度和加权出度等网络结构指标，衡量CO2浓度节点间相互影响与传播中的关键区域。计算网络中长程距离边的作用方向和权重，并将这些路径与一些已知的气候现象相关联。研究发现，节点的度、加权度和连接长度概率密度函数的分布符合对数正态分布。节点加权度高的节点集中在45°–60° N纬度带和50° S纬度带上，整体空间分布与瞬态热通量的分布模式相似。四条典型遥相关路径与不同环流作用有明确对应：（1）东亚到东北太平洋和北美西部的传输路径与亚热带西风急流有关；（2）北美向东延伸至北大西洋西部和地中海的传播遵循北大西洋风暴的路径；（3）俄罗斯西部传向中国南部的路径与北大西洋涛动（NAO）密切关联；（4）南半球的连接反映了大气Rossby波的影响。此外，本文还构建了夏季和冬季CO2浓度气候网络，分析遥相关的季节性特征。结果表明，连接权值峰值距离、时滞、相位速度、空间分布以及季节性均和大气Rossby波的性质一致。遥相关路径的结果显示北半球路径从北太平洋中部，向东扩展到北美和北大西洋，然后继续进入西北欧洲、地中海地区和东亚；南半球路径遵照大气Rossby波。这些传输路径反映了中纬度风（西风）、跨欧亚波列、NAO、东亚西风急流（ESWJS）、太平洋-北美模式（PNA）以及东大西洋/西俄罗斯（EA/WR）遥相关。 第三，构建全球地表温度网络，分析地表温度的遥相关路径，揭示大气CO2浓度与地表温度联系的物理机制。 分析不同区域地表温度之间存在的相关关系及其时滞，构建全球地表温度网络。应用节点加权入度和加权出度等网络结构指标，衡量地表温度节点间相互影响的关键区域。计算网络中长程距离边的作用方向和权重，并将这些路径与一些已知的气候现象相关联。研究表明，网络中地表温度相关作用强的节点分布在东亚及西延的北太平洋、美国东海岸东侧大西洋地区以及南半球的亚热带地区。进一步遥相关路径的分析发现，北半球的路径类似于CO2浓度网络中的传输路径，这为大气CO2浓度与地表温度的联系提供了有力的支持。此外，夏季和冬季地表温度网络的构建进一步揭示了地表温度与大气Rossby波的联系。实际上，连接权重峰值、节点加权度的空间分布、季节性特征都反应了Rossby波的作用。研究发现网络中遥相关路径在北半球从北太平洋中部向东扩展到北美和北大西洋，然后继续进入西北欧洲、地中海地区和东亚，在南半球的传输路径沿着Rossby波方向传播。其中，夏季北半球的遥相关路径与中纬度西风、跨欧亚波列、NAO和ESWJS有明显联系，冬季北半球则表现为EPA、EA/WR、NAO、ESWJS遥相关作用。值得注意的是，夏季和冬季地表温度的网络中连接权重峰值距离、传输方向、加权度的分布模式以及遥相关路径都与夏季和冬季CO2浓度网络相应的指标类似。因此，大气CO2浓度和地表温度之间存在明显的联系，并且这种相似模式是Rossby波的作用。 第四，构建大气CO2浓度和地表温度耦合网络，分析二者遥相关路径。 分析CO2浓度和地表温度节点之间相关性及其时滞，构建双层耦合气候网络模型。采用打散重排的方法对相关性进行显著性测试，分析显著的、可信的连接。应用节点加权入度和加权出度等网络结构指标，衡量CO2浓度节点和地表温度节点间相互影响的关键区域，确定关联影响最大区域及其遥相关路径，并用气候现象阐述原因。研究表明，连接权重在短距离值较大，说明CO2浓度与距离较近的地表温度联系密切。网络中连接权重峰值距离、时滞天数、速度、加权度分布特征都和大气Rossby波的性质一致，表明Rossby波是联系大气CO2浓度和地表温度的重要纽带。去除Rossby波的影响后，重新构建CO2浓度和地表温度的耦合网络。结果表明，CO2浓度和地表温度关键关联区域位于北美洲东部、北大西洋、东亚及北太平洋东部。遥相关路径的研究发现，北美洲北部的CO2浓度可以影响大西洋中部的地表温度，北冰洋的温度反过来影响美国东部的CO2浓度。这可能是北美洲北部的CO2浓度影响当地的温度，当地的温度通过西风带的作用影响北美洲东部的温度。此外，西伯利亚地区的CO2浓度传到中国东南部地表温度的路径和亚洲中部的地表温度向中国东部至太平洋地区的CO2浓度的路径与CO2浓度和地表温度网络中遥相关路径一致。这表明NAO在连接CO2浓度和温度中起着重要作用。

Nowadays, the frequent occurrence of the extreme weather and climate disasters is due to global warming, which exactly, leads to severe challenges to the sustainable development of natural ecology and human society. The main cause of global warming is the large amount of CO2 emissions from human activities. Satellite observations have found that atmospheric CO2 concentrations are not uniformly distributed, and there are interannual and seasonal variability in time and significant variability in spatial. The spatial variability is partly related to large-scale circulations. Previous studies on telecorrelation mainly focused on empirical orthogonal function decomposition (EOF) and statistical correlation analysis. The former could not capture the nonlinear dependence of climate system, while the latter is confined to limited number of CO2 monitoring station or regional scales. Overall, the telecorrelations of global atmospheris CO2 concentration is not clear. In addition, international studies on the relationship between atmospheric CO2 concentration and surface air temperature (SAT) are based on the global average CO2 concentration, without taking into account the non-uniform distribution of atmospheric CO2 concentration. In the recent decade, good results have been achieved in the analysis of complex nonlinear effects in the climate system and the advantage of describing the interaction relationship within the earth system, which provides a new idea for analyzing telecorrelations path of the atmospheric CO2 concentrations and their relationship with SAT. This study focuses on the non-uniform distribution of CO2 concentration and its relationship with SAT, aiming to analyse long-range connection between CO2 concentration and SAT. The thesis includes the features of non-uniform distribution of CO2 concentration, the long-distance connection of CO2 concentration, the long-distance connection of SAT, and the key areas of the CO2 concentration and SAT. The research not only has scientific guiding significance for correctly understanding the climate effect of CO2 and responsing to climate warming, but also has important practical significance for reducing disaster risk and implementing the future international climate change agreement. The main contents are as follows: Firstly, the characteristics of the non-uniform distribution of global satellite CO2 concentration as well as their spatial differences are studied from the perspective of temporal and spatial dimension. Based on the atmospheric CO2 concentration retrieved form GOSAT and AIRS satellites, the temporal and spatial distribution of atmospheric CO2 concentration and their spatial difference are analyzed.The results demonstrate that the distribution of GOSAT CO2 is strongly influenced by CO2 emissions from human activities and natural processes, while the characteristics of AIRS CO2 are concerned with both surface sources and large-scale circulation systems. Meanwhile, the spatial difference (GOSAT minus AIRS) ranges from 2 ppm to 10 ppm, and the greater discrepancy is concentrated at higher latitudes (above 60° N). More specifically, the largest difference occurs over the Arctic region, especially over the Greenland region. Seasonally, high values of GOSAT CO2 are concentrated in East Asia, South Asia, northeast parts of North America and Pacific region in spring, larger AIRS CO2 are concentrated in high latitudes of the Northern Hemisphere (NH). In summer, higher GOSAT CO2 concentrations occur in the Equatorial Africa and Equatorial South America, while high AIRS CO2 concentrations occur in high latitudes in NH. In autumn, the pattern of high GOSAT CO2 concentrations is similar to that in summer, and high AIRS CO2 concentrations are observed in southern South Africa, southeast Australia and neighbouring Pacifict. In winter, high GOSAT CO2 concentrations are distributed in East Asia and neighbouring north Pacific, Eastern North America, Western Europe and equatorial Africa, and high AIRS CO2 values are observed in the middle latitudes of the NH. It’s clear to see that GOSAT CO2 concentration is lower than AIRS CO2 concentration. Values in spring is highest while in summer is the lowest. Secondly, constructing a CO2 concentration network, identifying teleconnections paths of CO2 concentrations. Based on complex network theory, we analyzes the correlation and the time delay between different regions of the CO2 concentration, and study the significant CO2 concentration links by randomized significance test. Further, the indices of network structure, in-/out- weighted degrees, are applied to measure critical area calculation of interaction and communication between CO2 concentration nodes. The direction and weights of teleconnections paths are also calculated to associate with some already known climate phenomena. The results show that the distribution of degrees, weighted degrees, and edge lengths can be described by log-normal distribution. The map of the total weighted degrees exhibits high values in the latitude band 45° N–60° N and in the band centered near 50° S. The locations of the total weighted degrees hubs of the climate network are qualitatively similar to the transient heat flux. Careful analyses further reveal that long distance links spreading from west to east are the most dominant in the climate network. More specifically, we captured an eastward link from East Asia to the Northeastern Pacific Ocean and even the western North America region, following the subtropical westerly jet. And the long-term link from North America extending eastward to the western North Atlantic and Mediterranean belongs to southern part of the storm track in the North Atlantic. The connection between the western Russia and southern China generated by North Atlantic Oscillation (NAO) from the North Atlantic to East Asia is also apparent. Besides, links along the pathway of atmospheric Rossby waves perform evidently in the Southern Hemisphere (SH). By further constructing the climate network of CO2 concentration in summer and winter, it is found that the large links weight distribution, time delay, propagation direction, velocity, nodes distribution characteristics and seasonality are all consistent with the properties of atmospheric Rossby waves. The dominant long range links propagate from the central North Pacific, eastward to North America and the North Atlantic, and then to Northwest Europe, the Mediterranean, and even East Asia. The long distance transmission paths in SH are also associated with the Rossby wave. These propagation paths are associated with the mid-latitude wind (westerly), Eurasian wave train, NAO and East Asian Westerly Jet (ESWJS), Pacific-North American (PNA) and East Atlantic/West Russian (EA/WR). Thirdly, constructing SAT network, identifying the teleconnections paths, revealling the possible physical mechanism between atmospheric CO2 concentration and SAT. By analyzing the correlation and the time delay between different regions of the SAT, the SAT network is built. The indices of network structure, in-/out- weighted degree, is applied to measure critical area calculation of interaction and communication between SAT nodes. It is shown that the connectivity pattern shows a dense stripe of links in the Northern Hemisphere (NH) over East and West Asia, Northeast Pacific Ocean, Eastern parts of U.S. and Northwest Atlantic as well as in the extra tropics of the SH. Further analysis of the long-range connections present that the paths in NH are similar to that of the CO2 concentration network, which provide strong supports for understanding the relationship between atmospheric CO2 concentration and SAT. In addition, the establishment of the SAT in summer and winter periods further reveal the connection with an atmospheric Rossby waves. It apparent that the links weight, spatial distribution of weighting degrees nodes and seasonal characteristics yield a clear association with the pattern of atmospheric Rossby wave. Moreover, the long-distance links in the NH showed an eastward propagation from the central north Pacific, extending to North America and North Atlantic, and continuing to Northwest Europe, Mediterranean region, and even to East Asia. The paths in SH follow the pattern of the Rossby wave. It is worth noting that the characteristics of SAT network in summer and winter months are similar to that of summer and winter CO2 concentration network. It can be found in the distance of link weight peaks, direction, the centers of weight degrees nodes, as well as the long distance links. Therefore, it is reasonable to believe that there is a significant relationship between atmospheric CO2 concentration and SAT, and the similar patterns can be attributed to the influence of Rossby wave. Fourthly, building the coupling network of atmospheric CO2 concentration and SAT, identifying the key relation region. By analyzing correlation and the time delay between of CO2 concentration nodes and SAT nodes, we build a coupled climate network. We test the significance correlation with the method of shuffling. The results show that the correlation between CO2 concentration and SAT is relatively larger in a short distance, indicating that CO2 concentration mainly affected nearby location of SAT. The probability density function of degrees, weighted degrees and edge lengths of nodes follow the power-law distribution. The expected direction of energy flow, time delay days, group velocity, and weighted degrees distribution are consistent with the properties of Rossby waves. These factors confirm that the connection between atmospheric CO2 concentration and SAT and the reseaon is the influence of Rossby wave. Next, the coupling network is re-established after the effects of atmospheric Rossby waves removed. We identify the critical areas in the CO2 and SAT impacts and propagations by utlizing in-/out- weighted degrees and the important nodes of CO2 transmission to SAT occurs in East Asia, Mediterranean Sea, North Atlantic and East North America. The directions and weights of long-range paths are also calculated. More specifically, CO2 concentrations in Northern North America can affect the SAT in mid-Atlantic, and SAT in Arctic Ocean can affect CO2 in the Eastern U.S. It can be discerned that CO2 influence the SAT by affecting the local SAT, and then have a role in remote SAT through atmospheric circulation, which means CO2 and SAT can be affected each other by atmospheric large-scale circulation. Note that the routes from Siberia CO2 expand to SAT in southeast China, and the path form SAT in central Asia extend to East China CO2 as well as nerghboring Pacific Ocean are consistent with the long-range links in the CO2 concentration network and SAT network analyzed above. These factors reveal that NAO plays an importance role in linking CO2 and SAT.