长江流域是我国第一大流域，随着人类活动和气候变化对流域水循环的影响加剧，近20年来长江流域径流和水文情势发生深刻变化并引起广泛关注。论文以长江流域为研究对象，采用陆气耦合模拟方法，分析了从过去到未来的径流变化特征和原因。论文首先构建了长江流域的分布式水文模型，在此基础上结合全球气候模式的动力降尺度方法和统计降尺度方法及流域中尺度数值预报模式，建立了长江流域陆面水文模型与大气模型耦合的模拟系统，用于模拟流域过去的径流过程及预报和预测流域未来的径流过程。针对长江流域过去50年的径流和水文情势变化，论文利用所建立的分布式流域水文模型，模拟了1961~2009年流域长期的逐日径流过程。在此基础上，结合生态流量指标（生态盈余量和生态不足量）和IHA指标，建立了长江流域水文情势变化的综合评价方法，分析了1961~2009年期间长江流域的径流和水文情势变化特征，并定量评价了三峡水库运行对长江中下游水文情势变化的影响。结果表明，过去50年长江流域年径流呈减少趋势，而秋季径流减少趋势最为显著，这一趋势与流域秋季降水减少趋势一致；1990年代以来，水库蓄水增加进一步加剧了秋季径流的减少，而水库泄水增加了冬季的径流量。针对长江流域未来气象条件下的短期径流预报，论文以长江上游为研究对象，以三峡入库洪水预报为研究目标，采用中尺度数值天气预报模式与分布式水文模型耦合的模拟方法，开展了陆气耦合洪水预报研究。结果表明，数值天气预报模式对长江流域面平均降水量、降水过程及降水落区有一定预报能力，陆气耦合洪水预报方法具有一定精度，并显著增长了洪水预报的预见期。针对长江流域未来径流的预测，论文比较了动力降尺度和统计降尺度方法对长江流域不同区域气候特征的模拟效果，结果表明，动力降尺度方法对强降水过程和降水持续时间的模拟效果优于统计降尺度方法；利用陆气耦合模拟系统模拟了未来气候情景下长江流域的径流变化，结果表明，对未来气候情景下径流变化的预估，GCM模式的不确定性大于不同降尺度方法的不确定性；基于4个气候模式模拟得到的平均结果表明未来气候情景下各季节长江流域降水和径流均呈增加趋势，宜昌站年径流增加14.4%，而大通站的年径流增加10.8%。
Yangze River Basin is the largest catchment in China. Due to the impact of human activities and climate change, river discharge and flow regime of the Yangtze River Basin significantly changed in recent twenty years and this change has drawn great attention worldwide. In this study, the characteristic and the reasons for streamflow changes in the Yangtze River Basin from past to the future were investigated using a land-atmosphere coupling simulation method.First, a distributed hydrological model was developed for the whole Yangtze River Basin. On the basis of this hydrological model, a land-atmosphere coupling simulation system was developed, which includes the hydrological model, the dynamical downscaling and statistical downscaling method of GCM model and a meso-scale numerical weather forecast model. This system was applied to the streamflow simulation in the past, short time streamflow forecast and projection of streamflow changes in the future climate scenario. Regarding the streamflow and flow regime changes in the past 50 years, the daily river discharges in the period of 1961~2009 were simulated by the distributed hydrological model. A comprehensive analysis method was developed using the eco-flow metrics (ecosurplus and ecodeficit), the IHA metrics and the model simulation. Based on this method, the characteristic and the reasons for streamflow and flow regime changes in the Yangtze River Basin in the period of 1961~2009 were analyzed and the impact of the Three Gorges Reservoir on flow regime changes in the middle and lower Yangtze River was evaluated. The results showed that annual streamflow in the Yangtze River decreased in the past 50 years and the autumn streamflow significantly decreased which was consistent with the decreasing trend of precipitation. Since the 1990s, increase of reservoir water storing intensified the reduction of autumn streamflow. At the same time, winter streamflow increased which was attributed to the reservoir water release since the 1990s. Regarding the short time streamflow forecast, the Upper Yangtze River Basin was selected as the study area and the objective of this study was to improve the flood prediction of the Three Gorges Reservoir. A land-atmosphere coupling method was studied using the meso-scale numerical weather forecast model and the distributed hydrological model. The results of the precipitation forecast experiment illustrated that the numerical weather forecast model showed skills in predicting the basin-averaged rainfall amount, the rainfall temporal processes and the spatial distribution of rainfall. The land-atmosphere coupling method extended the leading time for the flood forecast of the Three Gorges Reservoir and showed acceptable accuracy. Regarding the projection of streamflow changes in the future climate scenario, the skills of the dynamical downscaling and statistical downscaling method in reproducing regional climate characteristics were compared. The results showed that dynamical downscaling method showed higher skills than the statistical downscaling method in reproducing the heavy precipitation processes and the wet spells. The streamflow changes in the future climate scenario were simulated by the land-atmosphere coupling simulation system. The results showed that for the projections of streamflow changes in the future climate scenario, the uncertainty from the GCM is larger than the uncertainty from the downscaling method. The results of the mean of four GCM models showed that the precipitation and streamflow of the Yangtze River Basin will increase in all the seasons in the future climate scenario. The annual streamflow will increase by 14.4% and 10.8% at the Yichang and Datong gauge, respectively.
