变化环境下的水循环研究是国际水文学与水资源、地理科学以及生态科学等众多学科交叉的前沿性问题。在全球变化背景下，青藏高原气候条件、水文生态状况等的演化，直接关系到区域生态安全屏障功能，对整个西部开发的战略决策和国际河流水事谈判也具有至关重要的意义。本文以青藏高原最大的河流——雅鲁藏布江流域为研究区，分析了流域气候水文状况的演变及地表（植被NDVI和土地覆盖）特征的时空变化；在流域有观测资料的8个子流域开展了缺资料地区径流模拟研究，并获取了全流域水文模型率定参数，并将之用于气候变化对径流的影响研究；用daily scaling统计降尺度方法，基于20个GCMs构建了全球地表平均气温增加1℃下的未来气候变化情景，分析了雅鲁藏布江流域径流对未来气候变化的响应。本文主要研究内容与创新点如下： 气温呈显著增加趋势，而降水的增加趋势不显著，径流变化不明显利用Mann-Kendall非参数检验方法，分析了雅鲁藏布江流域内和周边13个气象站点的气候要素和径流的长期变化趋势。流域多数站点的年降水呈不显著的增加趋势（未通过显著性检验）；而绝大多数站点的年平均气温、最高和最低气温呈显著的升高趋势。多模式预估结果显示，流域2010-2100年不同情景下（A1B、A2和B1）的年降水量将显著增加，且夏季降水量增幅最大；未来年平均气温将显著升高，且冬季增温幅度最大。雅鲁藏布江流域多数水文站点年径流无明显变化趋势，仅日喀则、拉萨和旁多站年径流存在微弱的增加趋势。雅鲁藏布江流域历史上存在较为显著的气候变暖趋势，但尚未对年与季径流产生明显的影响；但是，流域未来降水和气温都将显著增加，其必将对流域水循环过程产生更为显著的影响，因此开展未来气候变化对流域径流的影响研究将十分迫切。 植被NDVI呈微弱增加趋势，土地覆盖变化并不剧烈雅鲁藏布江流域植被NDVI呈总体增加趋势，但增加幅度较为微弱；年NDVI与降水和气温存在一定的相关性，与降水的相关系数为0.49（0.05显著性水平），而与气温的相关系数达0.64（0.01显著性水平）。雅鲁藏布江流域1985-2005年间仅有1%左右的土地面积发生了类型变化，土地覆盖变化并不剧烈；其中，面积增加最多的是林地，而变化幅度最大的类型是耕地，以百分比表示变化最快的类型是建设用地。雅鲁藏布江流域地表特征在1985-2005年间并未发生重大改变，仍可视为未受人类活动干扰的近自然流域，是开展PUB研究和气候变化对径流影响研究的理想区域。 首次在该区域采用降雨-径流模型进行大范围缺资料地区的径流模拟青藏高原区域PUB研究尚不多见。本文应用两种区域化方法——空间相近法和属性相似法，两个水文模型——SIMHYD和GR4J模型，两种驱动数据集——站点数据和格点数据，在雅鲁藏布江8个有观测资料的子流域首次进行了区域化研究。结果显示：空间相近法优于属性相似法，二者又都优于随机选取法；这一结论与欧洲和澳大利亚区域化研究的已有结论是一致的，本文的研究结论首次将中小湿润流域的区域化结论扩展到青藏高原雅鲁藏布江高寒大流域。雅鲁藏布江流域空间分布的格点数据集可以显著改善水文模型径流模拟和区域化效果；但是，增加融雪模块对模型模拟效果和区域化结果的改善有限，这可能与该区域降雪占降水量比例小、对月水文过程影响小有关。本文利用空间相近法将8个有观测资料流域的模型率定参数展布到雅鲁藏布江全流域0.5分辨率网格上，进而运行水文模型开展气候变化对径流的影响研究。 采用20个GCM的未来气候情景模拟显示气候变化将使该区域径流增多利用Daily scaling统计降尺度方法，考虑气候序列季节格局和日降水频率分布，将20个GCMs模拟输出降到流域0.5°分辨率网格上，得到全球地表平均气温增加1℃下的未来气候情景。利用历史气候序列和未来气候情景驱动SIMHYD和GR4J模型，对比分析后得到气候变化对流域径流的影响情况。20个GCMs预估雅鲁藏布江流域未来年降水量变化的中值和第10、90百分位数值分别为+7%和-15%、+16%；20个GCMs预估流域未来年径流变化的中值和第10、90百分位数值分别为+13%和-24%、+29%；20个GCMs预估一江两河和大拐弯地区未来年降水量变化的中值分别为+6.8%和+5.2%，而预估年径流变化的中值分别为+11.7%和+8.7%。4个模拟流域降水较好的GCMs（MIUB、MIRO-H、CSIRO-MK3.0和GIS-AOM），其集合预估结果与20个GCMs的预估中值较为接近，表现出一定的预估精度。

Studies on hydrological cycle under changing environment are focued on earth biosphere, global change and impacts of human activities on water resources, and it is a multidisciplinary frontier subject in the field of hydrology, water resources, geography, and ecology, etc. The Tibetan Plateau (TP) is the Asian “water tower” and it exerts a huge influence on both hydrological cycle and climate for southern and eastern Asia. In the background of global change, evolution of climate, hydrology, ecology, snow and ice, permafrost, etc. on the TP would directly influence the strategies of eco-environment management, construction of hydraulic structures and infrastructure, and the development of economics in western region of China. It is critical to investigate the impact of climate change on runoff for better management and negotiation of water resources in the international river on the TP.This dissertation will be maily focused on the runoff simulation in ungauged catchments and the impacts of climate change on runoff in the Yarlung Tsangpo River (YTR) basin. Main works and innovations are as follows. Air temperature shows significant increase tendency while precipitation shows no significant variation, runoff has no obvious changeLong-term trends of major climate variables are detected by using Mann-Kendall non-parameter test method based on the data from 13 meteorological stations in and around the YTR basin. Annual precipitation in majority of stations shows increasing trend, but most of their null hypothesis are accept at a significance level of 0.10. There is significant rising trend of mean, maximum and minimum air temperature in majority of stations, and most of their null hypothesis are rejected at a significance level of 0.10. Ensemble projection results show that annual precipitation in the YTR basin will increase significantly during the period of 2010-2100 under A1B, A2 and B1 scenarios, and precipitation in summer shows the greatest increase. Annual and seasonal mean air temperature will increase intensively, and mean temperature in winter shows the greatest increase.Long-term trend of runoff are tested by using Mann-Kendall non-parameter test method as well. There is no obvious trend for annual runoff in most of stations while annual runoff in Shigatse, Lhasa and Pondo shows slight increase tendency.There is obvious historical climate warming in the YTR basin, but climate warming has not yet affect annual and seasonal runoff significantly. However, there will be huge increase of precipitation and air temperature in future, runoff will be affect more intensively. Therefore, it is urgent to study the impact of climate change on runoff in the YTR basin. NDVI shows obscure increase tendency, and land cover shows little changeThere is obscure increase tendency for annual NDVI averaged in the YTR basin. Furthermore, there is positive correlation between annual NDVI and annual precipitation, air temperature averaged in the YTR basin. Correlation coefficient between NDVI and precipitation is 0.49, with rejection of a null hypothesis at the significance level of 0.05, and it is 0.64 between NDVI and air temperature with rejection of a null hypothesis at the significance level of 0.01.There is no intensive land cover transformation in the YTR basin from 1985 to 2005, and only about 1% of the land cover changed during this period. Forest shows the greatest expansion, and farmland shows the greatest variation which was transformed to built-up, water and wetland. However, built-up shows the rapidest change.It can be concluded that there is not significant variation of vegetation and land cover in the YTR basin from 1985 to 2005. The YTR basin is still a natural watershed without intensively influenced by human activities. Therefore, it is a suitable watershed for studies on PUB and the impact of climate change on runoff. Runoff simulation in ungauged catchments using rainfall-runoff models for the first time in the YTR basinThere were few studies focued on runoff prediction in ungauged catchments in the TP. In order to fill in this gap, this study evaluates two regionalization methods, spatial proximity and physical similarity, for the simulation of runoff using two rainfall-runoff models (SIMHYD and GR4J) driven by two kinds of input datasets from eight large non-nested catchments (4000 km2 to 50000 km2) in the YTR basin.The regionalization results obtained from the SIMHYD and GR4J models show that the spatial proximity approach marginally outperforms the physical similarity approach and both are much better than the random selection of the donor catchment. This is consistent with recent regionalization studies carried out in Europe and Australia. Therefore, results from this study extend the general regionalization from middle or small catchments to high-cold and large-area ungauged catchments in the TP.The gridded forcing data can noticeably improve runoff simulation compared to the site forcing data. This is mainly because that the gridded forcing data considers the complex topography in southeast TP and makes the spatial pattern of forcing data more realistic. However, incorporation of snowfall/snowmelt processes into rainfall-runoff models cannot noticeably improve monthly runoff simulation in the study area because of the small percentage of snowfall to annual precipitation and the dominant summer precipitation.Based on spatial proximity approach, validated parameters of hydrological models are get in the whole YTR basin, and they are used for study on the impact of climate change on runoff in the following section. Future runoff in the YTR basin will increase projected by future climate change scenarios from 20 GCMsTwo rainfall-runoff models (SIMHYD and GR4J) are used to simulate monthly and annual runoff across the YTR basin in south-east TP under historical (1962-2002) and future (up to about 2030) climate conditions in this study. The future climate series are obtained by using 20 Global Climate Models (GCMs) outputs in the Fourth Assessment Report (AR4) of the Intergovernmental Panel on Climate Change (IPCC), to reflect a 1℃ of increase in global average surface air temperature.Mean annual precipitation and runoff across the region are projected to increase by majority of the modelling results. The changes in mean annual precipitation obtained from 20 GCMs are -15%, 7% and 16% for 10th, median and 90th percentiles of GCM outputs and the corresponding changes in simulated mean annual runoffs are -24%, 13% and 29% for 10th, median and 90th percentiles of SIMHYD model outputs. The changes in mean annual precipitation for the median of 20 GCMs outputs in middle reaches of the YTR and its two tributaries (the Lhasa River and Nyangqu River) and the great canyon region are +6.8% and +5.2% respectively, and the corresponding changes in mean annual runoff are +11.7% and +8.7% respectively. The ensemble results from four best GCMs (MIUB, MIRO-H, CSIRO-MK3.0 and GIS-AOM) are similar to the median results obtained from 20 GCMs outputs. This is the first comprehensive study on the hydrological response to climate change covering the entire upstream of the YTR, and the results found in this study is not only helpful for local water resources management, but also important for water resources management in the lower reaches of the Brahmaputra.