世界上主要的人类饮用水源和自然生态系统水源均来自于河流，定量确定河流流量并掌握其时空分布对人类福祉具有重大意义。当前最重要的观测流量的方法是通过地面水文测站直接观测，但是对于世界上的大多数地方，由于地理位置偏远、人烟稀少和政治不稳定等因素，缺少相关的水文观测数据。遥感技术的发展为全球河流流量的观测提供了一种新的手段，可以弥补地面流量观测的不足。梯田是黄河中游重要的水土保持措施，具有明显的减水效应。定量评价梯田减水作用对于分析黄河中游径流变化原因具有重要意义。目前梯田减水计算的三种方法均无法有效地对梯田减水进行定量定位的计算。利用遥感估算前后两期梯田坡面出口处的河道流量变化，可以实现梯田减水的定量定位计算。本文利用C/M时间序列法和多站水力几何法（AMHG）遥感估算河流流量，并将估算流量的方法应用于黄河中游梯田减水计算。主要研究内容和结论如下：（1）利用C/M时间序列法遥感估算河流流量，并扩展其时空尺度和应用区域对雅江羊村、雅江奴下、泾河板桥三个流域分别构建了C/M-MODIS时间序列矩阵，并拟合了C/M-MODIS最优化流量回归关系，流量模拟的效率系数分别为0.61、0.80、0.15，雅江羊村和雅江奴下的结果较好，泾河板桥结果反映出MODIS自身空间分辨率不高并且无法全天实时获取地面水体变化的缺陷。对孤山川高石崖、秃尾河高家堡两个流域分别构建了C/M-Landsat时间序列矩阵，并拟合了C/M-Landsat最优化流量回归关系，流量模拟的效率系数分别为0.57、0.62，模拟结果较好。C/M时间序列法对于新的时空尺度和应用区域具有较好的适应性。（2）利用多站水力几何法遥感估算河流流量，并扩展其时空尺度和应用区域根据Landsat提取了河流断面宽度，利用遗传算法驱动AMHG-Landsat估算出雅江奴下的河流流量，流量模拟的效率系数为0.97，模拟结果较好。对雅江奴下AMHG-Landsat进行了参数敏感性分析，发现b-high最敏感，取值在0.29-0.35之间时模拟效果较好。根据Google Earth提取了河流断面宽度，利用遗传算法驱动AMHG-Google Earth估算出延河延安的河流流量，流量模拟的效率系数为0.89，模拟结果较好。多站水力几何法（AMHG）对于新的时空尺度和应用区域具有较好的适应性。（3）遥感估算河流流量在黄河中游梯田减水计算中的应用利用DEM数据构建出黄河中游地区流域数字河流网络，并结合梯田空间位置数据，计算统计出黄河中游流域内部的梯田集水区。祖厉河流域梯田拦截比例在0.44-0.70之间；渭河流域梯田拦截比例在0.37-0.54之间；泾河流域梯田拦截比例在0.26-1.39之间；北洛河流域梯田拦截比例在0.55-0.98之间；延河流域梯田拦截比例在1.08-1.59之间；无定河流域梯田拦截比例在1.18-2.37之间；窟野河流域梯田拦截比例在2.25-3.87之间；其它黄河河口-龙门区间西侧流域梯田拦截比例在1.07-4.35之间；其它黄河河口-龙门区间东侧流域梯田拦截比例在0.65-1.90之间。利用1967年Key Hole影像和2010年Google Earth影像结合AMHG估算出梯田改造前后坡面出口处的河道流量，分别提取出五个区域的梯田集水区，进而对梯田减水进行了定量定位计算。在梯田和林草植被减水共同作用的区域中，梯田区域-1全区域1967-2010年削减洪峰流量为0.85m3/s，削减百分比为34.4%，梯田削减洪峰流量为0.09m3/s；梯田区域-2全区域1967-2010年削减洪峰流量为0.38m3/s，削减百分比为44.7%，梯田削减洪峰流量为0.08m3/s；梯田区域-3全区域1967-2010年削减洪峰流量为0.65m3/s，削减百分比为28.0%，梯田削减洪峰流量为0.11m3/s。在梯田减水占主导地位的区域中，梯田区域-4全区域1967-2010年削减洪峰流量为0.31m3/s，削减百分比为77.5%，梯田削减洪峰流量为0.15m3/s；梯田区域-5全区域1967-2010年削减洪峰流量为0.23m3/s，削减百分比为100.0%，梯田削减洪峰流量为0.19m3/s。

Rivers supply the major drinking water sources of human beings and water sources of natural ecological system. It has great significance for human beings to quantify the river discharge and its space-time distribution. Currently, the most important method for observating river discharge is observating through ground hydrological stations. However, because of the remote geographical positions and some political factors, most parts of the world are lacking of hydrological data. The development of remote sensing technology can supply a new method for observation for river discharge and make up for the inadequacy of the ground flow observation.Agricultural terraces having the obvious effection of discharge reduction are the important soil and water conservation measures in the middle reaches of the Yellow River Basin. It has a great significance for analyzing runoff chang reasons in the middle reaches of Yellow River Basin through quantitative evaluation of discharge reduction caused by agricultural terraces. The three main methods for calculating discharge reduction caused by agricultural terraces can not effectively make quantitative computations for discharge reduction at the accurate position.This paper estimates the river discharge from remote sensing images using C/M time series method and at-many-stations hydraulic geometry (AMHG) method and applies the method to calculate discharge reduction caused by agricultural terraces in the middle reaches of Yellow River Basin. The main research contents and conclusions are as follows:（1）Estimating river discharge from remote sensing images using C/M time series method and expanding spatial-temporal scale and application areas of this methodC/M-MOIDS time series matrixes are established in Yangcun hydrological station of Yarlung Zangbo River Basin, Nuxia hydrological station of Yarlung Zangbo River Basin and Banqaio hydrological station of Jinghe River Basin. The flow simulation efficiency coefficients are 0.61, 0.80, 0.15 through fitting the optimal relationship between flow series and C/M-MODIS series. The results are good in Yangcun hydrological station and Nuxia hydrological station of Yarlung Zangbo River Basin. The result of Banqiao reflects defects of MODIS images in the low spatial resolution and disability of access to the water changes throughout the whole day. Then C/M-Landsat time series matrixes are established in Gaoshiya hydrological station of Gushanchuan River Basin and Gaojiabao hydrological station of Tuweihe River Basin. The flow simulation efficiency coefficients are 0.57, 0.62 through fitting the optimal relationship between flow series and C/M-Landsat series. The results are good in Gaoshiya hydrological station of Gushanchuan River Basin and Gaojiabao hydrological station of Tuweihe River Basin. C/M time series method has a good adaptability for the new spatial-temporal scale and application areas.（2）Estimating river discharge from remote sensing images using at-many-stations hydraulic geometry method and expanding spatial-temporal scale and application areas of this methodRiver cross section widths are extracted based on Landsat images in Nuxia hydrological station of Yarlung Zangbo River Basin. River discharge is estimated using AMHG-Landsat driven by Genetic Algorithm. The simulation results are good and the flow simulation efficiency coefficient is 0.97. Parameter b-high is the most sensitive parameter of AMHG-Landsat in Nuxia hydrological station of Yarlung Zangbo River Basin. Simulation effect is better when the value of b-high is between 0.29-0.35. Then river cross section widths are extracted based on Google Earth images in Yan’an hydrological station of Yanhe River Basin. River discharge is estimated using AMHG-Google Earth driven by Genetic Algorithm. The simulation results are good and the flow simulation efficiency coefficient is 0.89. At-many-stations hydraulic geometry (AMHG) method has a good adaptability for the new spatial-temporal scale and application areas.（3）Application of estimating river discharge from remote sensing images in calculating discharge reduction caused by agricultural terraces in the middle reaches of Yellow River BasinDigital river network is constructed based on DEM in the middle reaches of Yellow River Basin. Terrace catchment areas are calculated in the middle reaches of Yellow River Basin combining the spatial location data of agricultural terraces. The terrace interception ratio is between 0.44-0.70 in Zulihe River Basin, between 0.37-0.54 in Weihe River Basin, between 0.26-1.39 in Jinghe River Basin, between 0.55-0.98 in Beiluohe River Basin, between 1.08-1.59 in Yanhe River Basin, between 1.18-2.37 in Wudinghe River Basin, between 2.25-3.87 in Kuyehe River Basin, between 1.07-4.35 in other river basins on the west side of Hekou-Longmen range in the middle reaches of Yellow River Basin, between 0.65-1.90 in other river basins on the east side of Hekou-Longmen range in the middle reaches of Yellow River Basin.River discharge is estimated at the exit of the hillslope from Key Hole images in 1967 before the terrace renovation and Google Earth images in 2010 after the terrace renovation. Terrace catchment areas are extracted from five regions and then quantitative computations for discharge reduction at the accurate position are made. Three of these regions are the regions where both agricultural terraces and vegetations have effects of discharge reduction. The flood peak flow reduction of the whole region is 0.85m3/s, reduction percentage is 34.4% and the flood pear flow reduction of agricultural terraces is 0.09m3/s from 1967 to 2010 in terrace area-1. The flood peak flow reduction of the whole region is 0.38m3/s, reduction percentage is 44.7% and the flood pear flow reduction of agricultural terraces is 0.08m3/s from 1967 to 2010 in terrace area-2. The flood peak flow reduction of the whole region is 0.65m3/s, reduction percentage is 28.0% and the flood pear flow reduction of agricultural terraces is 0.11m3/s from 1967 to 2010 in terrace area-3. Two of these regions are the regions where agricultural terraces have the main effects of discharge reduction. The flood peak flow reduction of the whole region is 0.31m3/s, reduction percentage is 77.5% and the flood pear flow reduction of agricultural terraces is 0.15m3/s from 1967 to 2010 in terrace area-4. The flood peak flow reduction of the whole region is 0.23m3/s, reduction percentage is 100.0% and the flood pear flow reduction of agricultural terraces is 0.19m3/s from 1967 to 2010 in terrace area-5.