中文摘要：

应用阿克苏河两条支流的年径流量和国家气候中心发布的74类大气环流月指数资料，分析了年径流量与前一年大气环流月指数之间的关系．用滑动相关法提取初始因子库，建立回归模型和集合分析，得到超级集合回归模型．托什干河年径流量监测序列与超级集合拟合（预测）序列的相关系数达0．8766，而库玛拉克河达到0．8122．用逐级反推法，确定影响两条支流年径流量年际变化的前一年环流指数因子．影响托什干河年径流量年际变化的前期环流指数都以持续1个月的因子居多；影响库玛拉克河年径流量年际变化的前期环流指数都以持续2个月的因子居多，其次是4个月、1个月．影响两条支流年径流量的前期大气环流指数因子的种类、空间分布、时间分布、持续影响时间等差异很大，这些不同与阿克苏河两条支流的流域参数、地表水来源组成、地表径流形成的气象条件等多方面的差异有关．

英文摘要：

The relationships are analyzed between the annual runoff of two embranchments of the Aksu River and all 74 circulation indexes of the former year published by the Climate Diagnostics and Prediction Division of the National Climate Center. Thirty-four sets of initial factor database are constructed separately by moving-correlation method and regression model, and two assembling- regression models are constructed by stepwise regression method. The correlation coefficient is 0. 8766 between the measured and simulated annual runoff in the Toxkan River, and 0. 8122 in the Kumarik River. There are six atmospheric circulation indexes which have notable correlation with annual runoff of the Toxkan River, one of the two embranchments of the Aksu River. The six indexes belong to six of all the 74 circulation indexes. They are mainly in middle-latitude areas, appea- ring mainly in February, March, April, and their duration is mainly one month. There are 11 atmospheric indexes which have notable correlation with annual runoff of the Kumarik Rriver, another embranchment of the Aksu River. The 11 indexes be- long to seven of all the 74 circulation indexes. They are mainly in low-middle-latitude and highlatitude areas, appearing tember and October, and mainly in August, Sew their duration is mainly 2 months. The differences of the spatial-temporal distributions and duration of prophase circulation indexes, which have notable correlation with the annual runoff between the two embranchments of the Aksu River, are obvious.