CMIP6与CMIP5全球气候模式对中国东北地区气温模拟性能的比较评估

本文基于CN05.1逐月气温观测资料和CMIP6计划中34个模式资料、CMIP5中39个模式资料,利用泰勒图、技能得分(S值)、综合评级指标(M_r),系统地评估了相比于CMIP5模式,CMIP6模式对1961—2005年中国东北地区(黑龙江省、吉林省、辽宁省)气温模拟能力.结果表明：1)相较于CMIP5模式,CMIP6中大部分模式能更好地模拟出区域平均气温多年变化、年平均气温气候态空间分布及年平均气温气候倾向率的空间分布的特征,但普遍存在低估的现象;2)经过优选后得到的CMIP5与CMIP6最优模式集合平均(MME5、MME6)对年平均气温的模拟优于大部分单个模式和所有模式的集合平均模拟结果. MME6比MME5能更好地模拟出年平均气温气候态及气温多年变化趋势的空间分布特征,但对区域平均气温多年变化的模拟能力要略低.总体来说,CMIP6模式相对于CMIP5有所进步,MME6对中国东北地区气温的时空变化特征具有一定的模拟能力.     

CMIP6气候模式对21世纪初全球增暖减缓现象模拟能力评估与归因分析

自工业革命以来,大气中温室气体浓度持续升高,但1998～2013年期间却出现了全球增暖减缓现象,挑战了现有的对全球温度变化机理的认识.第五次国际耦合模式比较计划(Coupled Model Inter-comparison Project Phase5, CMIP5)中众多当时最先进的气候模式都没能合理地模拟出该现象,引发了公众对气候模拟和预测的质疑.文章基于六种常用的全球表面温度观测资料集,评估了最近发布的28个CMIP6气候模式对21世纪初全球增暖减缓现象的模拟能力.结果显示,相对前一代的CMIP5模式,新一代的CMIP6模式尽管在年代尺度增暖速率模拟能力上有所提高,但大多数CMIP6模式依然未能重现出全球增暖减缓现象.在28个CMIP6模式中,只有4个成功模拟出了增暖减缓,其他模式普遍高估了1998～2013年期间的增暖速率,呈现出明显的增暖加速,而不是观测到的增暖减缓.进一步分析表明,气候模式能否成功模拟出增暖减缓现象的关键在于其能否准确区分并模拟出两种不同来源的温度变化信号,即人类排放温室气体导致的长期增暖信号和气候系统内部短期的自然变率信号,特别是年际、年代际和多年代际这三种尺度的自然变率. 4个复现出增暖减缓现象的模式都较为合理地模拟出了长期增暖趋势和三种关键尺度自然变率.上述评估结果为改进气候模式对短期气候变化的模拟和预测能力提供了一个新的视角.     

多区域模式集合的东亚陆地区域的平均和极端降水未来预估

利用东亚区域联合降尺度计划(CORDEX-EA) 15个区域模式的模拟结果,集合预估了高排放情景RCP8.5下东亚陆地区域平均和极端降水的未来时空变化,并量化未来预估的不确定性.结果 表明:区域模式基本上能够再现东亚及各个区域平均和极端降水的多年平均分布.未来多模式集合预估的平均和极端强降水在东亚各区域多表现为增加,连续无降水日数(CDD)表现为南增北减,且变幅多随时间增大.到21世纪末期,冬季和年平均降水的增幅大值都位于中国西部(WC),冬季降水的变化在WC、蒙古(MG)、中国东北(NE)和中国华北及西北地区东部(NC)的确定性都较高,年降水的变化仅在WC和MG确定性较高.夏季降水增幅大值位于朝鲜半岛和日本(KJ),且仅在这一区域确定性较高.最大5日降水量(Rx5day)和大雨日数(R20)以增加为主且变化的空间分布较为均匀,除去中国江南及华南(SC)和KJ的R20变化,其余区域两个变量的变化确定性都较高.CDD的增幅和减幅大值分别位于SC和MG,其变化在MG、NE和SC确定性较高.     

中国平均降水和极端降水对气候变暖的响应:CMIP5模式模拟评估和预估

基于24个CMIP5全球耦合模式模拟结果,分析了中国区域年平均降水和ETCCDI强降水量(R95p)、极端强降水量(R99p)对增暖的响应.定量分析结果显示,CMIP5集合模拟的当代中国区域平均降水对增温的响应较观测偏弱,而极端降水的响应则偏强.对各子区域气温与平均降水、极端降水的关系均有一定的模拟能力,并且极端降水的模拟好于平均降水.RCP4.5和RCP8.5情景下,随着气温的升高,中国区域平均降水和极端降水均呈现一致增加的趋势,中国区域平均气温每升高1℃,平均降水增加的百分率分别为3.5%和2.4%,R95p增加百分率为11.9%和11.0%,R99p更加敏感,分别增加21.6%和22.4%.就各分区来看,当代的区域性差异较大,未来则普遍增强,并且区域性差异减小,在持续增暖背景下,中国及各分区极端降水对增暖的响应比平均降水更强,并且越强的极端降水敏感性越大.未来北方地区平均降水对增暖的响应比南方地区的要大,青藏高原和西南地区的R95p和R99p增加最显著,表明未来这些区域发生暴雨和洪涝的风险将增大.     

CMIP5模式对东亚冬季风指数变化及其与冬季大气环流和气温关系的模拟评估

本文评估了44个CMIP5模式对东亚冬季风环流系统,特别是东亚冬季风指数及其对应的环流和气温特征的模拟能力.结果表明:CMIP5模式对地表气温和500 hPa位势高度场模拟效果最好,对200 hPa纬向风的模拟次之,而对海平面气压和850 hPa经向风的模拟相对较差.与单个模式相比,多模式集合(MME)的模拟能力要更优,其能够很好地再现西伯利亚高压、阿留申低压、东亚低层偏北风、中层东亚大槽、高层东亚西风急流以及地表气温的空间分布.不过,模拟的环流系统偏强,造成东亚地表气温总体偏低.对于东亚冬季风指数,分别选取基于300 hPa纬向风(I_(Jhun))、850 hPa风场(I_(Wang))、500 hPa位势高度(I_(Cui))、以及海平面气压(I_(Guo))定义的四个指数表征东亚冬季风强度.MME能很好地模拟I_(Cui)和I_(Wang)指数的长期变化,还能合理再现四个指数所指示的东亚冬季风环流和气温的变化特征:对应冬季风偏强年份,西伯利亚高压、阿留申低压、东亚沿岸低层北风、东亚大槽和高空西风急流加强,东亚大陆地表气温和极端低温降低,但变化的幅度比观测结果偏弱.     

SRES B2情景下中国区域最高、最低气温及日较差变化分布特征初步分析

本文利用Hadley气候预测与研究中心的区域气候模式系统PRECIS进行中国区域气候基准时段(1961～1990年)和SRES B2情景下2071～2100年(2080s)最高、最低气温及日较差变化响应的分析.气候基准时段的模拟结果与观测资料的对比分析表明：PRECIS具有对中国区域最高、最低气温及日较差的模拟能力,能够模拟出中国区域最高、最低气温及日较差的局地分布特征.对SRES B2情景下相对于气候基准时段的最高、最低气温及日较差变化响应分析表明：中国区域2080s时段年、冬季和夏季平均最高、最低气温变化均呈一致增加的趋势,北方地区增温幅度普遍大于南方地区.夏季东北地区极端高温事件发生的频率将会增加,而冬季华北地区极端冷害事件发生频率将会减少.未来中国区域年平均日较差将出现北方地区减小而南方地区增加的趋势.冬季长江中下游以南地区日较差呈增加趋势,而夏季华东地区、西北地区及内蒙古中部日较差将呈减小趋势,其中在青藏高原北部地区存在一个较强的低值中心.     

气候系统模式对北极涛动的模拟

本文系统分析了参与IPCC AR4 “20世纪气候模拟"（20C3M）的23个气候系统模式模拟的1950～1999年冬季（JFM）北极涛动（AO）.结果表明,在22个模式中,AO模态都表现为北半球中高纬大气年际变率的第一模态.不过,18个模式模拟的AO模态在北太平洋地区表现偏强,有两组模式结果均表明,提高模式的水平分辨率能够克服此偏差.所有模式模拟的AO指数均未出现与观测相当的增强趋势,只有两个模式（ECHO-G和UKMO-HadGEM1）模拟的AO指数与观测存在显著正相关.多数模式能够模拟出纬向风伴随AO位相变化在中高纬出现的偶极子异常特征,部分模式对平流层AO特征的模拟能力仍需要提高.对AO垂直结构模拟较好的模式,例如CCSM3、MRI CGCM2-32和UKMO-HadGEM1,能够较为合理地再现伴随AO指数增强出现的极区平流层变冷和中纬度对流层增暖现象.AO与欧亚大陆地表气温和降水的相关分布,在多数模式中有较好的体现,个别模式还对AO与东亚气候的相关关系具有一定的模拟能力.对AO各指标均有较好模拟能力的模式是UKMO-HadGEM1.本文结果为改进气候系统模式对北极涛动的模拟能力提供了依据,亦为学术界利用IPCC AR4的耦合模式数据、开展与AO相关的气候变率研究提供了基础参考.     

SRES情景下多模式集合对淮河流域未来气候变化的预估

采用偏差修正/空间降尺度方法处理后的IPCC AR4中8个全球海气耦合模式的集合平均结果,分析了SRESA2、A1B和B1情景下淮河流域未来30 a(2011 2040年)相对于现状(1961 1990年)地面温度和降水的可能变化.结果表明:(1)多模式集合能较好地反映流域现状年、季温度和降水的大尺度空间分布特征;对温度和降水的年内分配过程模拟较好,各月温度集合平均与观测值相差0.2℃左右(冬季各月除外),而降水集合平均与观测值相对误差在5%左右(9月除外).(2)不同情景下未来流域年、季温度一致增加,年温度增加幅度在0.85～1.12℃之间;冬、春季温度增加相对明显,而夏、秋季温度增加并不显著;年际和年代际温度增加趋势显著.(3)不同情景下未来流域年降水有增加趋势,增加幅度为0.13%～5.24%,增幅不明显;降水季节变化有增有减,季节、年际和年代际降水变化较为复杂,不同情景下降水空间变化差异显著.     

CMIP5多模式集合对东北三省未来气候变化的预估研究

应用CN05观测资料,以及参与国际耦合模式比较计划第5阶段(CMIP5)中的26个模式,评估了新一代全球气候模式对东北三省气候变化模拟能力并选出4个较优模式,发现经过筛选得出的较优模式集合平均模拟结果的可靠性得到进一步加强,尤其体现在对气温的模拟上.在此基础上着重分析了多模式集合在不同典型浓度路径(RCPs)下对未来气候变化特征的预估.结果表明:21世纪的未来阶段,东北三省将处于显著增温的状态,且RCP8.5情景下的增温速率(0.53℃/10a)明显高于RCP4.5情景下的速率(0.22℃/10a);空间上,北部地区将成为增温幅度最大、增温速率最高的区域.未来降水将会相对增加,但波动较大,21世纪末期RCP4.5和RCP8.5情景下的降水增加幅度分别为11.24%和15.95%;空间上,辽宁省西部地区将成为降水增加最为显著的区域.根据水分盈亏量,21世纪未来阶段,RCP4.5情景下的东北三省绝大多数地区未来将相对变湿,尤其到了中后期;RCP8.5情景下则是中西部地区将相对变干,其余地区则会相对变湿.     

东亚地区晴空湍流未来变化趋势预估：基于CORDEX-WRF模式降尺度

利用耦合模式比较计划(CMIP5)的全球环流模式CNRM-CM5、GFDL-ESM2M、EC-EARTH和MPI-ESM-LR分别驱动区域气候模式(RCM)WRF,在国际间区域气候降尺度试验(CORDEX)框架下对东亚地区当代和RCP 8.5情景下未来气候进行高分辨率模拟.通过计算21个晴空湍流指数研究了东亚地区2041—2060年三个高度层上(200 hPa、250 hPa和300 hPa)五种强度湍流(轻度、轻-中度、中度、中-重度和重度)的季节变化特征.首先检验RCM对1981—2000年气候的模拟能力,结果显示,模式能够较好再现东亚地区高空变量气候平均态空间分布特征,同时对晴空湍流指数分布特征亦具有较好模拟能力.晴空湍流未来变化趋势预估结果显示,东亚地区所有季节中均存在湍流强度越强、湍流频率增幅越大的特征.各强度湍流增幅冬季最大,其次为春季和秋季,夏季最小.从不同高度对比来看,200 hPa和250 hPa各强度湍流增幅普遍大于300 hPa,且在夏季和秋季增幅随高度降低而减小,春季和冬季则250 hPa上增幅最大,300 hPa上增幅最小.就中国不同子区域而言,纬度越高的区域...     

Analysis to significant climate change in aerosol influence domain of Beijing and its peripheral areas by EOF mode

Using the total ozone mapping spectrometer (TOMS) aerosol optical depth (AOD)data and the sunshine duration, fog days, Iow cloud cover (LCC), etc. meteorological data in 1979-2000 in North China, as well as empirical orthogonal function (EOF) mode statistical analyses method, the winter aerosol distributive character of Beijing and peripheral city agglomeration and its influence effect on regional climate are investigated in this paper, especially the relation between aerosol influence effect and distinct change regions of eigenvectors of EOF mode. It is found from analyzing the regional distribution of the long-term averaged winter TOMS AOD that there is a large-scale relatively stable high value zone of aerosol concentration in the valley of the Beijing and peripheral U-shape megarelief. A high correlation area of AOD between Beijing and its southern peripheral exists in winter, and in this significant region of aerosol interaction, there is "in-phase" character of the interannual variations of winter AOD, fog days, and LCCs. It indicates that the variations of aerosol in Beijing and its peripheral areas have impacts on interannual changes of fog days and LCCs in this area. The EOF analyses of the meteorological data further reveal the climate change regions and long-term trends of winter sunshine duration-reducing, and LCC- and fog days-increasing in North China. The areas of significant changes of the first EOF eigenvectors (FEE) of sunshine duration, fog days, LCCs almost superpose on corresponding marked regions of interdecadal differences between the 1990s and 1980s, and all accord with the S-N zonal high value pattern and high correlation region of winter AOD in Beijing and its peripheral areas. Interannual variations of their associated time coefficients (ATC) are in phase with that of regional mean AOD, and both of them have a secular rising trend. Results by EOF mode analyses depict the regional climatic change principal character of winter sunshine duration-reducing, and LCC- and fog days-increasing in peripheral areas to the south of Beijing, and reveal the regional influence effect of aerosol, i.e. the high value zone of long-term averaged winter AOD, significant change regions of FEE of sunshine duration, fog days, and LCC all lie in peripheral city agglomeration to the south of Beijing. These distributive features above suggest that there exists a regional strengthening trend of aerosol climatic effect within influence domain in peripheral city agglomeration to the south of Beijing.     

1951～2002年中国东、西部地区地面气温变化对比

利用1951～2002年全国733个测站经过非均一性检验的月平均气温资料,在剔除50万以上人口大城市测站后,分析了52年来中国东、西部及青藏高原地区的气温变化趋势的一致性和差异性,并讨论了其可能原因.结果表明我国东、西部地区年、季平均气温变化有着较好的一致性;近52年来,我国东、西部和青藏高原地区年平均气温均呈升温趋势,年平均气温的增温速率东部为026℃/10a,西部018℃/10a,东部比西部高008℃/10a;季平均气温东部地区冬、春季的增暖趋势大于西部和青藏高原,而其夏、秋季的增暖趋势小于西部和青藏高原.我国东、西部地区年、季平均气温变化关系密切,说明其主要是受全球气候变化的影响而变化,但东部年平均气温的增暖总趋势大于西部,又说明地域差异在气温变化中也有重要作用.     

MM5V3模式对IPCCA1B情景下中国地区极端事件的模拟和预估

利用MM5V3区域气候模式单向嵌套ECHAM5全球环流模式,对中国地区1978-2000年及IPCC A1B情景下2038-2070年气候分别进行了水平分辨率为50 km的模拟试验.文章首先检验了模式模拟的当代极端气候结果,在此基础上对6个极端温度指数和6个极端降水指数的未来变化进行了预估.检验结果表明:MM5V3模式对中国地区当代日最高、最低温度及强降水(大雨和暴雨)日数的空间分布和概率特征均具有一定的模拟能力,但模拟的日最高温度在大部分地区偏低,日最低温度在南方地区偏低、西北地区偏高.概率统计结果显示日最高温度向低值频段偏移,日最低温度在0℃的峰值附近明显偏高.模式对大雨和暴雨年平均日数的模拟在东部地区偏多,概率统计结果则为一致偏大.未来中国地区极端气候预估结果表明:极端高温、极端低温和相对高温在全国范围内都将升高,且线性趋势均为上升;霜日日数则为减少,并具有下降趋势;暖日日数和相对低温在青藏高原和新疆部分地区有所减少、其它地区均为增加,且线性趋势暖日日数为上升,相对低温不明显.极端降水指数的变化具有区域特征,其中单日最大降水、连续五日最大降水、最长无雨期、强降水日数、简单降水强度和极端降水总量均在江淮、华南及西南地区有所增多,而在东北及内蒙古地区有所减少,未来中国南方地区降水的极端化趋势将更加显著.极端降水指数的线性趋势除最长无雨期外其它均为上升.     

Singular value decomposition analysis of spatial relationships between monthly weather and air pollution index in China

Spatial patterns are important features for understanding regional air quality variability. Statistical analysis tools, such as empirical orthogonal function (EOF), have been extensively used to identify and classify spatial patterns. These tools, however, do not directly reveal the related weather conditions. This study used singular value decomposition (SVD) to identify spatial air pollution index (API) patterns related to meteorological conditions in China, one of world’s regions facing catastrophic air pollution. The monthly API and four meteorological variables (precipitation, surface air temperature, humidity, and wind speed) during 2001–2012 in 42 cities in China were used. The two leading SVD spatial patterns display the API anomalies with the same sign across China and opposite signs between northern and southern China, respectively. The meteorological variables have different relationships with these patterns. For the first pattern, wind speed is the most important. The key regions, where the correlations between the API field and the wind speed’s SVD time series are significant at the 99% confidence level, are found nationwide. Precipitation and air temperature are also important in the southern and northern portions of eastern China, respectively. For the second pattern, the key regions occur mainly in northern China for temperature and humidity and southern China for wind speed. Air humidity has the largest contribution to this pattern. The weather-API relationships characterized by these spatial patterns are useful for selecting factors for statistical air quality prediction models and determining the geographic regions with high prediction skills.     

Evaluating spatio-temporal representations in daily rainfall sequences from three stochastic multi-site weather generation approaches

Many hydrological and agricultural studies require simulations of weather variables reflecting observed spatial and temporal dependence at multiple point locations. This paper assesses three multi-site daily rainfall generators for their ability to model different spatio-temporal rainfall attributes over the study area. The approaches considered consist of a multi-site modified Markov model (MMM), a reordering method for reconstructing space–time variability, and a nonparametric k-nearest neighbour (KNN) model. Our results indicate that all the approaches reproduce adequately the observed spatio-temporal pattern of the multi-site daily rainfall. However, different techniques used to signify longer time scale observed temporal and spatial dependences in the simulated sequences, reproduce these characteristics with varying successes. While each approach comes with its own advantages and disadvantages, the MMM has an overall advantage in offering a mechanism for modelling varying orders of serial dependence at each point location, while still maintaining the observed spatial dependence with sufficient accuracy. The reordering method is simple and intuitive and produces good results. However, it is primarily driven by the reshuffling of the simulated values across realisations and therefore may not be suited in applications where data length is limited or in situations where the simulation process is governed by exogenous conditioning variables. For example, in downscaling studies where KNN and MMM can be used with confidence.     

Spatial and Temporal Variations in Indian Summer Monsoon Rainfall and Temperature: An Analysis Based on RegCM3 Simulations

Regional climate models are important tools to examine the spatial and temporal characteristics of rainfall and temperature at high resolutions. Such information has potential applications in sectors like agriculture and health. In this study, the Regional Climate Model Version 3 (RegCM3) has been integrated in the ensemble mode at 55 km resolution over India for the summer monsoon season during the years 1982–2009. Emphasis has been given on the validation of the model simulation at the regional level. In Central India, both rainfall and temperature show the best correlations with respective observed values. The model gives rise to large wet biases over Northwest and Peninsular India. RegCM3 slightly underestimates the summer monsoon precipitation over the Central and Northeast India. Nevertheless, over these regions, RegCM3 simulated rainfall is closer to the observations when compared to the other regions where rainfall is overestimated. The position of the monsoon trough simulated by the model lies to the north of its original observed position. This is similar to the usual monsoon break conditions leading to less rainfall over Central India. RegCM3 simulated surface maximum temperature shows a large negative bias over the country while the surface minimum temperature is close to the observation. Nevertheless, there is a strong correlation between the all India weighted average surface temperature simulated by RegCM3 and IMD observed values. While examining the extreme weather conditions in Central India, it is found that RegCM3 simulated frequencies of occurrence of very wet days, extremely wet days, warm days and warm nights more often as compared to those in IMD observed values. However, these are systematic biases. The model biases in the frequencies of distribution of rainfall extremes explain the wet and dry biases in different regions in the country. Overall, the inter-annual characteristics of both the rainfall and temperature extremes simulated by RegCM3 in Central India are well in phase with those found in the observed data.     

Relationship and its instability of ENSO — Chinese variations in droughts and wet spells

Monthly data of Self-Calibrated Palmer Drought Severity Index (PDSI) from 1951 to 2000 are calculated using historical precipitation and temperature data for Chinese 160 stations. Temporal and spatial patterns of the first empirical orthogonal function (EOF) of the PDSI reveals a fairly linear trend resulting from trends in precipitation and surface temperature, which is similar to the linear PDSI trend during 1951–2000 calculated using all monthly data. The EOF analysis also reveals that the leading mode correlates significantly with ENSO events in time and space. The ENSO EOF shows that during the typical warm phase of ENSO, surface conditions are drier in most regions of China, especially North China, but wetter than normal in the southern regions of Changjiang River, and Northwest China. During the typical cold phase of ENSO, these anomalies reverse sign. From 1951 to 2000, there are large multi-year to decadal variations in droughts and wet spells over China, which are all closely related to strong El Niño events. In other words, when one strong El Niño event happens, there is a possible big variability in droughts and wet spells over China on the multi-year or decadal scale. Studies also suggest that during the last 2–3 decades climate changes over China, especially North China’s drying and northwest China’s wetting, are closely related to the shift in ENSO towards warmer events and global warming since the late 1970s. The instability of the relationship is also studied. It is revealed that there is a good correlation between ENSO and Chinese variations in droughts and wet spells in the 3–8-year band, but the correlation between ENSO and Chinese variations in droughts and wet spells is instable. Studies suggest that there are decadal changes in the correlation: the wavelet coherency between ENSO and Chinese variations in droughts and wet spells is high during 1951–1962 and 1976–1991, but low during 1963–1975 and 1992–2000.     

Mixed layer depth (MLD) variability in the southern Bay of Biscay. Deepening of winter MLDs concurrent with generalized upper water warming trends?

Mixed layer depth (MLD) variability from seasonal to decadal time scales in the Bay of Biscay is studied in this work. A hydrographic time series running since 1991 in the study area, a climatology of the upper layer vertical structure based on the topology of this temperature profile time series and a one-dimensional water column model have been used for this purpose. The prevailing factors driving MLD variability have been determined with detail, and agreement with observations is achieved. Tests carried out to investigate climatological profile skill to reproduce the upper layer temporal evolution have demonstrated its ability to simulate variability at seasonal time scales and reproduce the most conspicuous events observed. This has enabled us to carry out a reconstruction of the MLD variability for the last 60 years in the study area. Favourable sequence of intense mixing events explains interannual differences and cases of extraordinary deepening of winter mixed layer. The negative phase of the Eastern Atlantic pattern seems to determine important interannual variability through intense episodes of cooling and mixing as in winter 2005 in the Bay of Biscay. Low-frequency variability is also observed. A very striking and unexpected shallower winter MLD during the 1970s and 1980s than those observed from 1995 has been found. Simulation results support this counter-intuitive outcome of shallower winter mixed layers concurrent with generalized upper water warming trends reported on several occasions for the area. The long-term trends in MLD seem related with decadal variability in the North Atlantic Oscillation, being in phase and opposition with other deepening-shallowing cycles found from subtropical-to-subpolar areas in the North Atlantic.     

20世纪温度变化中自然变率和人为因素的影响:基于耦合气候模式的归因模拟

本文基于中国科学院大气物理研究所大气科学和地球流体力学国家重点实验室(LASG/IAP)发展的气候系统模式FGOALS_gl对近百年气温变化的模拟,讨论了自然变率和人为因素对20世纪全球变暖的相对贡献.数值试验结果表明,在自然和人为因子的共同强迫作用下,耦合模式能够合理再现20世纪全球平均气温随时间的演变;仅在自然因子作用下,模式不能再现1970年以后的全球变暖.自然因素对20世纪第一次变暖的作用是显著的,但温室气体是20世纪后期变暖的主要原因.在这一定性结论基础上,进一步对近百年变化中自然和人为因素的相对贡献做定量的归因分析,结果表明,除赤道中东太平洋和北大西洋外,人为因素对近百年的增暖起决定性作用.对全球、半球及大陆尺度而言,外强迫可以解释平均气温变化的70%以上,而内部变率贡献较小;但对于区域尺度而言,多数地区内部变率的贡献大于外强迫,区域尺度气温变化的机制较全球、半球尺度要复杂.对中国地区而言,20世纪早期的气温变化受自然变率影响,但20世纪后期的变暖主要是温室气体增加的结果.中国东部气温变化的空间分布表明,自然因素对近50年及近百年中国地区的变暖趋势贡献较小.在自然和人为因子共同作用下,模式能够再现近50年中国东部气温变化冬春两季增暖的特征、但没有模拟出夏季长江中下游地区及淮河流域的降温趋势;自然因子试验的结果表明,太阳活动对该区域的变冷有贡献,但模式无法再现该地区气温的季节变化特征.