Knowledge Resource Center for Ecological Environment in Arid Area
| CMIP5气候模式对陆表水循环过程的模拟及预估研究 | |
| 其他题名 | Assessment of Simulation of the Land Surface Hydrological Cycle and Prediction by CMIP5 Climate Models |
| 杨笑宇 | |
| 出版年 | 2017 |
| 学位类型 | 博士 |
| 导师 | 林朝晖 |
| 学位授予单位 | 中国科学院大学 |
| 中文摘要 | 陆表水循环是联系气候系统五大圈层的重要纽带，对全球水分、能量收支以及生态系统有着重要调节作用，也是全球能量和水分循环试验(GEWEX)计划的重要研究内容。在全球气候变化背景下，降雨、蒸发、积雪等水循环要素的演变对于人们生产生活和经济社会可持续发展有着极其重要的影响。本论文基于CMIP5多模式的模拟结果，并结合多种观测资料，包括：由2200多个台站观测分析得到的东亚地区逐日降水资料集、美国Montana大学发布的全球逐月蒸发数据集以及美国冰雪资料中心(NSIDC)提供的卫星反演全球逐月雪水当量数据集，系统性评估了CMIP5气候模式对东亚地区降雨、蒸发、降水蒸发差（P-E）以及雪水当量等水循环要素的综合模拟能力，详细分析了CMIP5模式对我国夏季不同强度降水的模拟性能；最后分析了CMIP5中对水循环分量模拟能力较强的模式在不同排放情景下对各个水循环要素的预估结果。本论文的主要结论如下：1、分析了观测和CMIP5模式模拟的1962-2005年中国东部夏季不同强度降水的累积降水量分布特征。观测结果表明，中国东部夏季总降水在空间上由南向北逐渐递减，年均量值从华南沿海的900mm以上过渡到内蒙古地区的200mm；小雨量值空间上自北向南呈现“东北多—华北少—江南多”三级结构；中雨量值呈现由南向北逐渐递减趋势；大雨集中在东南季风区，长江流域下游及江南地区可达180-200mm，华南量值高于200mm；暴雨的空间形态及累积降水量量值与大雨类似。从CMIP5模式模拟结果分析，多数模式能够抓住不同强度降水量值空间形态的主要特征，其中对总降水分布特征的模拟优于对不同强度降水的模拟；多数模式高估了整个东部区域的小雨、中雨量值，并低估了大雨、暴雨在华北以南地区的量值。2、不同强度降水占总降水的比重在中国不同区域的分布特征差异较大，小雨占总雨量比例在内蒙地区较高，可超过50%，在黄淮以南比值低于30%；中雨占总雨量比重东北最大，可达40%以上，在江淮流域和华中地区略低，约25%-30%，其他地区在30%-40%之间；大雨占总降水量的比重在中国东部地区大都在25%左右，但在黄淮—江淮部分地区略高，可占30%以上；对于暴雨而言，其比重的空间分布与大雨类似，只是在东北地区暴雨的比重略低。CMIP5多数模式可以模拟出小雨、大雨和暴雨比重的空间分布的主要特征，但是各地小雨比重被高估5%以上，江淮流域的大雨和暴雨比重被低估了约5%，中雨比重空间变率很低，没有明显的地区差异。3、中国东部不同地区降水的年代际变化差异较大，从70年代到80年代观测总降水量值在东北和江淮流域有所增加，华北和江南地区有所减少；90年代以后，观测总降水量开始反向转变，东北和江淮流域的雨量相对80年代减少，华北和江南增加。不同强度降水量的年代际变化趋势遵循总降水演变规律，并且各自对总降水量值变化的贡献相当。CMIP5模式对中国东部不同强度降水量值年代际变化的模拟欠缺整体把握能力，只能刻画东北平原、华北地区、江淮流域和华南地区等局地变化特征。定量化模拟指数S显示BCC、CCSM4和MPI-ESM-P对总降水及不同强度降水量值年代际变化的模拟技巧相对最高。4、观测和CMIP5模式模拟的东亚夏季不同强度降水日数的空间分布及其年代际变化的分析结果表明，中国东部降雨日数在东北和江南雨日较多，年均60-70天，黄淮地区降水日数不足50天。总降水日数主要由小雨日数和中雨日数贡献，其中，小雨日数空间形态和总降雨日相似，自南向北呈现“多—少—多”的三级结构；中雨日数则由南向北逐步递减；大雨、暴雨日数空间上南高北低，对总降水日数的贡献很小。分析降水日数的年代际变化发现，80年代较70年代东北和江淮总降水日数增加、华北和江南总降水日数减少。同时期不同强度降水日数变化也是如此。90年代以后，总降水日数和小雨日数在华北和江南有所增加、东北和江淮地区有所减少；中雨和大雨日数在长江以北微弱减少、长江以南显著增加。CMIP5多数模式可以体现小雨和中雨日数的空间主要特征。模式集合平均（MME）结果体现了大多数模式的共同特点，其与观测资料的主要差异在于，MME高估东部不同区域小雨日数8-10天/年，高估东部不同区域中雨日数2天/年左右。并且MME模拟的南方地区大雨、暴雨日数不及观测的一半，南北差异被严重减弱，体现不出强降水日数分布的南北空间差异。CMIP5模式对不同强度降水日数年代际变化的模拟能力同样较低，其中BCC、CCSM4和MPI-ESM-P的模拟效果相对最好，与雨量的模拟结果相类似。5、基于多年平均陆地水量平衡方程（P=R+E），评估CMIP5模式对1983-2005年中国东部夏季陆气水循环分量——蒸发和径流的模拟效果，其中在多年尺度上中径流可以有降水与蒸发差（P-E）进行估计，并利用经验正交函数（EOF）方法分析蒸发、降水与蒸发差（P-E）的时空演变特征，考察模式对陆气水循环分量的时空演变规律的刻画能力。观测结果表明，中国东部地区夏季蒸发集中在南方地区，江淮流域、东南沿海年均蒸发量在80mm-100mm之间；北方地区蒸发量一般在50-70mm之间。各地降水普遍高于蒸发，径流量从南向北依次递减，江南地区的年均径流量高达100-150mm；中北地区、内蒙古等干旱区的径流量只有30-60mm。通过分析蒸发和降水与蒸发差（P-E）的时空演变特征发现：中国东部夏季蒸发量的EOF第一模态在1983-1996年期间的量值普遍偏高、1997-2005年则普遍偏低；第二模态呈现年际变化特征，东北和江淮流域变化趋势一致，华北和江南与之相反。降水与蒸发差（P-E）第一模态在1992年之前江淮流域量值较平均态偏大、江南偏小，1992年之后情况相反；第二模态空间变化集中在长江流域，华南地区与之反位相。CMIP5多数模式能够体现中国东部夏季蒸发值南高北低的观测特征，但中国南方地区蒸发值被高估40%-80%。模式对长江以北径流量分布形态模拟较好，但在华南低估了至少50mm/yr，对于个别模式，如FGOALS、IPSL-CM5B-LR、MIROC5和MRI-CGCM3，在华南、江淮等地出现局地蒸发大于降水的情况，不符合观测事实。另外，评估模拟的水循环分量的时空演变特征，发现CMIP5模式对蒸发和降水与蒸发差（P-E）时空演变特征的模拟技巧很低，少数模式体现了第一模态空间变化特征。 6、评估了CMIP5模式对1981-2005年欧亚大陆冬季雪水当量的分布和时空演变特征，结果表明，欧亚大陆多年平均的冬季雪水当量由南向北逐渐增加，年均量值从30mm增加到200mm；青藏高原积雪多于同纬度其他地区，大多数CMIP5模式可以反映出这些特征。但是，模式模拟相对于观测资料也存在如下不足：模式对雪水当量的大值中心模拟存在偏差，遥感资料大值中心位于西伯利亚东部，而模式大值中心一般位于西伯利亚中西部；模拟的积雪边缘与观测存在偏差，遥感资料的积雪南侧边缘不越过30°N，而模式积雪南伸至沿海区域；在量值大小方面，多数模式雪水当量较观测偏低，其中在西伯利亚东部地区的负偏差最为显著，较观测偏低50%-80%。通过分析1981~2005年冬季雪水当量的时空演变特征发现：观测资料EOF第一模态在1988年之前，欧亚大陆雪水当量比平均态偏多、之后则偏少；第二模态中西伯利亚东西部雪水当量变化趋势相反。模式对大陆尺度雪水当量时空演变特征模拟技巧很低，多模式集合平均的模拟能力甚至不如单个模式。7、基于前文CMIP5模式的模拟结果，分别筛选出对极端降水和雪水当量模拟技巧较高的模式，利用这些模式的集合平均结果分析不同排放情景下极端降水和雪水当量的预估情况。在极端降水方面，模式集合平均结果显示，在不同排放情景下21世纪中国东部极端降水(日降水量＞25mm)量值和日数总体呈现增长趋势，其中东北地区增加较为明显；华南地区有所减少。RCP8.5高排放情景下极端降水量值及日数的增幅超过RCP4.5低排放情景下的增幅，这主要是由21世纪后期(2076-2100年)的差异造成的。在雪水当量方面，模式集合平均结果显示：21世纪欧亚大陆雪水当量主要呈现减少趋势，西伯利亚地区西部和青藏高原减幅超过了-1.0mm/10a，西伯利亚中东部地区和中国南方则有小幅度增加现象；在高排放情景下，西西伯利亚区域的减幅大值区出现大范围的南扩，由低排放情景下的55°N以北延伸至50°N以南；贝加尔湖西北地区雪水当量随着排放量的增加呈现加强趋势。 |
| 英文摘要 | As an important part of the Global Energy and Water Cycle Experiment (GEWEX), land surface hydrological cycle provides vital links among the components of climate system and contributes significantly to the exchange of moisture and energy. The potential changes in the Earth’s water cycle due to climate change have important effect on daily life and economic development.This study evaluates the variability of Coupled Model Intercomparison Project Phase 5 (CMIP5) historical simulations of hydrological cycle components such as precipitation and evaporation, using observations from multi-source data. The datasets include daily precipitation station data over East Asia, global monthly evaporation data by the University of Montana USA and global monthly satellite retrieval data of snow water equivalent (SWE) obtained from National Snow and ICE Data Center (NSIDC). More so, this study focuses on evaluating the ability of these models in reproducing summer precipitation of different intensities over Eastern China. Then, models with higher skill in simulating water cycle components are selected to investigate model projections under different emission scenarios. The main conclusions are as follows: (1) Analyses of the distribution of summer rainfall with different intensities over Eastern China in CMIP5 simulations reveal that, total summer rainfall in the Eastern China gradually decreases from southeast to northwest. The value decreases from 900mm/yr over South China coast to 200mm/yr over Inner Mongolia. The amount of light rainfall is high over Northeast China and south of the Yangtze River, and low in North China. The amount of medium rainfall decreases by degrees from south to north. Large and heavy rainfall mainly occurred in Monsoon regions, with the spatial average values of above 180 mm/yr over Yangtze River basin and south of the Yangtze River and of above 200 mm/yr over south China. Most of the CMIP5 models reproduce well the distributions of different intensities of rainfall, with a better performance in simulating the total rainfall. Most models overestimated (underestimated) light and medium (large and heavy) rainfall values over the southern part of north China. (2) The ratios of the amounts of rainfall with different intensities to the amount of total rainfall are different in various regions of China. The ratio of the amounts of light rainfall to the amount of total rainfall is more than 50% over Inner Mongolia area, and is less than 30% in the south of Huanghuai area. The ratios of the amounts of medium rainfall to the amount of total rainfall are 25-30% in Yangtze-Huaihe River basin and Central China, it is more than 40% in the Northeast of China, to the rest of China, the ratios are between 30% and 40%. The ratio of the amounts of large rainfall to the amount of total rainfall is about 25% in Eastern China, and more than 30% in some areas of Huanghuai and Jianghuai plains. The pattern of ratios of the amounts of heavy rainfall to the amount of total rainfall is similar to the pattern of the ratios of the amounts of large rainfall to the amount of total rainfall, only lower in Northeast China. Most CMIP5 simulations can reproduce distribution patterns of the ratios of the amounts of light, large and heavy rainfall to the amount of total rainfall, simulated ratio of the amount of light rainfall to the amount of total rainfall is 5 percent overestimated in the whole China region, and simulated ratios of the amounts of large and heavy rainfall to the amount of total rainfall are 5 percent underestimated in Yangtze-Huaihe River basin. The spatial variability of the ratio of the amounts of medium rainfall to the amount of total rainfall in multimodel ensemble mean (MME) of CMIP5 models is very low, this indicates CMIP5 simulations.are inconsistent with observations(3) The interdecadal variabilities of summer precipitation are variant in different regions of China. The amount of observed precipitation is increasing in Northeast China and Yangtze-Huaihe River basins, and decreasing in North China and South of Yangtze River during the period of 1970-1989. Since 1990s, the trends of precipitation in those area have been reversed. The trends of rainfall with different intensities have the same changing trend as total rainfall. The contributions of rainfall with different intensities to total precipitation are similar. No model in CMIP5 can produce identical trends of rainfall with different intensities in every area of China. Every model can only give similar trend in some areas or for some precipitation intensities. The S score, which quantifies the similarity of the distribution and amplitude of a spatial pattern to the observations, is used to evaluate the interdecadal variabilities of summer precipitation in CMIP5 models, the results indicate that BCC, CCSM4 and MPI-ESM-P shows the best simulating capability.(4) The spatial and temporal characteristics of the number of days of summer precipitation in China with different intensities are also studied. The spatial pattern of total rainfall days is (more-less-more) tendency in the meridional direction. The maximum numbers of days of total precipitation are 60-70 days/yr in North China and South of Yangtze River. And the numbers of days of total precipitation is less than 50 days/yr in Huanghuai area. Most number of days of total precipitation consists of number of days of light and medium precipitation. The spatial pattern of light rainfall days is similar to the spatial pattern of light rainfall days; the number of medium rainfall days decreased from north to south; the numbers of large and heavy rainfall days, higher in lower latitude, has a low proportion in the number of days of summer precipitation. The interdecadal variability of the number of days of summer precipitation is obvious. The number of days of summer precipitation in the 1980s is higher than in the prior decade in Northeast China and Yangtze-Huaihe River basin, and opposite tendency is in North China and South of Yangtze River. Since 1990s, the number of days of total and light precipitation have increasing trend in North China and South of Yangtze River and decreasing trend in Northeast China and Yangtze-Huaihe River basin. The trends of the number of days of medium and large precipitation are significantly increasing in south of Yangtze River, but in the north of Yangtze River, decreasing tendency are not significant. Most CMIP5 models can produce similar main features of spatial pattern of the number of days of light and medium rainfall. The MME of CMIP5 models, representing common feature of most models, overestimated 8-10 days/yr and 2 days/yr in some areas of Eastern China for the number of days of light and medium precipitation, respectively. The MME of CMIP5 models cannot reproduce the spatial variability of the number of strong precipitation, simulated number of days of large and heavy rainfall is less than half in South China. No model in CMIP5 can reproduce identical decadal variability of rainfall with different intensities in every area of China. Relatively, BCC, CCSM4 and MPI-ESM-P shows the better simulating capability, in good agreement with those for amount of precipitation.(5) Based on the land surface water balance equation (P = R + E) to assess the summer evaporation and runoff simulation results in China Eastern from 1983 – 2005. Then analyzing the temporal and spatial patterns of evaporation and P-E with empirical orthogonal function (EOF) analysis method. Observations suggest that the evaporation in eastern China are mostly found in the south, and the results are as follows. Evaporation over the Yangtze-Huaihe River basin, southeast coastal areas and northeast plain is about 80-100mm/yr. Relatively, the amount is 50-70 mm/yr in northern area. Over the east part of China, the values of summer precipitation are higher than those of summer evaporation. P-E decreases from south to north. The value is about 100-150 mm/yr in Jiangnan district and only 30-60 mm/yr over arid region such as Inner Mongolia. More so, the first mode in the observational EOF shows that evaporation during 1983-1996 is more than average value during 1983-2005 over the entire eastern China. And evaporation during 1997-2005 is less than the average value. The second mode shows that evaporation over Northeast and Yangtze-Huaihe River basin vary in the same trend, while North China and Jiangnan area vary in opposite way. The corresponding time series is characterized by interannual variability. Likewise, the first mode of observational P-E shows that P-E before 1992 over Yangtze-Huaihe River basin is more than the average value during 1983-2005 and over Jiangnan district is less than the average value. The mode after 1992 is contrary to that before 1992. The second mode shows that the change of P-E mainly occurs over Yangtze River basin, since it is contrary over South China.Most of the models could capture the main features of summer evaporation. But evaporation value in southern China is overvalued by 40% -80%. Models show good ability on simulating the patterns of runoff in north of the Yangtze River. But the values are underestimated about 50mm/yr in South China. Some models, such as FGOALS, IPSL-CM5B-LR, MIROC5 and MRI-CGCM3, show large evaporation values over South China and Yangtze-Huaihe River basin which are greater than rainfall value. It does not comply with the observation. And CMIP5 models show little ability in simulating the EOF modes of evaporation and P-E. Only a few models can reproduce the spatial pattern of the first mode. (6) The spatial and temporal characteristics of winter SWE in the Eurasia are analyzed. The results show that annual average of SWE increases from 30mm/yr to 200mm/yr from south to north; SWE in Tibetan Plateau is much higher than those in other parts of the same latitude. Most CMIP5 models can produce all these features, but some errors still exist in the models. The maximum SWE exist in midwestern Siberia in most CMIP5 models, it is different with observation which maximum SWE in eastern Siberia. Compared with observed southern margin of snow cover locating in the north of 30°N, the simulated snow cover stretches southward to coastal areas that is a great error in some CMIP5 models. Most CMIP5 models underestimated SWE, especially in eastern Siberia underestimating by 50-80% than the observation. The spatial and temporal characteristics of winter SWE in the Eurasia during 1981-2005 are analyzed by using EOF analysis. The first eigenvector reflects that before 1988, SWE is higher than the average, and less than the average during following period. The trend of second mode is opposite in eastern Siberia. All models are unable to produce main feature of the spatial and temporal characteristics of winter SWE.(7) Based on the simulation results of CMIP5 models, picking out the models which have high skills on simulating extreme precipitation and SWE. Then analyzing the patterns of extreme precipitation and SWE under different emission scenarios with ensemble mean of these models. Ensemble mean model results show that the magnitude and frequency of extreme precipitation (daily precipitation> 25mm) in Eastern China have a rising trend in the 21st century. This trend is obvious in Northeast China. Contrarily, there is a downtrend in South China. The magnitude and frequency of extreme precipitation increase faster in RCP8.5 (high emission scenario) than that in RCP4.5 (low emission scenario). It is caused by the difference from late 21st century (2076-2100 year). Besides, ensemble mean model results show that the value of SWE in Eurasia has a decreasing trend in the 21st Century. Reduction is over -1.0mm/10a in Western Siberia and the Tibetan Plateau. However, there is a small increase in eastern Siberia and South China. Large reduction area in West Siberian region has a southern expansion under high emission scenarios. It extends from the north of 55 ° N (under low emission scenario) to the south of 50 ° N. And the value of SWE in Lake Baikal northwest is increasing with the emission scenario. |
| 中文关键词 | 区域水循环 ; 降水等级 ; 蒸发 ; 雪水当量 ; CMIP5 ; 模式模拟 ; 气候预估 |
| 英文关键词 | Regional Hydrological Cycle Precipitation Categories Evaporation Snow Water Equivalent CMIP5 Models Simulation Climate Projection |
| 语种 | 中文 |
| 国家 | 中国 |
| 来源学科分类 | 气象学 |
| 来源机构 | 中国科学院大气物理研究所 |
| 资源类型 | 学位论文 |
| 条目标识符 | http://119.78.100.177/qdio/handle/2XILL650/287828 |
| 推荐引用方式 GB/T 7714 | 杨笑宇. CMIP5气候模式对陆表水循环过程的模拟及预估研究[D]. 中国科学院大学,2017. |
| 条目包含的文件 | 条目无相关文件。 | |||||
| 个性服务 |
| 推荐该条目 |
| 保存到收藏夹 |
| 导出为Endnote文件 |
| 谷歌学术 |
| 谷歌学术中相似的文章 |
| [杨笑宇]的文章 |
| 百度学术 |
| 百度学术中相似的文章 |
| [杨笑宇]的文章 |
| 必应学术 |
| 必应学术中相似的文章 |
| [杨笑宇]的文章 |
| 相关权益政策 |
| 暂无数据 |
| 收藏/分享 |
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。
