中文摘要：

在总结地极移动（以下简称极移）和径流之间可能存在的相互作用机制的基础上，利用受人类活动影响较少的青藏高原雅鲁藏布江的月径流资料和极移资料以及格兰杰因果关系检验方法，从统计学角度探索了在月、季和年尺度上，极移变化与径流变化之间可能存在的联系。结果显示，在月尺度上，极移X分量变化量在滞后1～8个月和10～21个月后是径流变化量的格兰杰原因；极移Y分量变化量在滞后1～9个月和17～24个月后是径流变化量的格兰杰原因。在季尺度上，极移X分量变化量只有在滞后2个季的情况下是径流变化量的格兰杰原因；而极移Y分量变化量在滞后4～6个季的情况下是径流变化量的格兰杰原因。在年尺度上，未检测出格兰杰因果关系。从径流到极移的检测中发现，月尺度上，径流变化量在滞后3～25个月后为极移X分量变化量的格兰杰原因；径流变化量在滞后1个月和3～25个月后为极移Y分量变化量的格兰杰原因。在季尺度上，径流变化量在滞后2～8个季的情况下是极移X分量变化量的格兰杰原因；径流变化量在滞后1～8个季后是极移Y分量变化量的格兰杰原因。在年尺度上，未发现径流变化量和极移变化量存在格兰杰因果关系。在月、季和年尺度上，极移变化量和径流变化量的不同格兰杰因果关系表明，利用极移资料可能在月和季尺度上提高资料稀缺区域的水文预测精度。

英文摘要：

The evidence of the relationship between polar motion and runoff is reviewed. The relationship between the runoff of Yarlung Zangbo, a primitive river in Tibet, China and polar motion with data records from about 1846 to present is explored. First, the action path framework was structured based on geophysical principles. In order to make time series stationary, the initial time series of polar motion X component, Y component and runoff were transformed into their corresponding ifrst order difference time seriesΔX,ΔY, andΔQ. The Granger causality test betweenΔX,ΔY andΔQ was conducted on monthly, seasonal and annual time scales. It is found that on a monthly scaleΔX inlfuencesΔQ at the lag being from the 1st to the 21st month, with the 9th month being an exception.ΔY inlfuencesΔQ at the lag from the 1st to the 9th month and from the 17th to the 24th month. On a seasonal scale, the inlfuence ofΔX onΔQ can be seen in the 2nd season （i.e. from the 4th to the 6th month）. The inlfuence ofΔY onΔQ can be seen on a seasonal scale from the 4th to the 6th season （i.e. from the 10th to 18th month）. We cannot see evident Granger causality fromΔX,ΔY toΔQ, on annual scales. For the Granger inlfuence ofΔQ toΔX,ΔY, it is found that on a monthly scale the inlfuences are prominent at the lag being from the 3rd to the 25th months for theΔX, for theΔY it is the 1st and the 3rd to the 25th months. On a seasonal scale these inlfuences can be seen at the lag from the 2nd to the 8th season （corresponding to the 4th to the 24th month） forΔX and at the lag from the 1st to the 8th season （corresponding to from the 1st to the 24th month） forΔY. Again, on a annual scale no evident granger causality can be found from runoff to polar motion. The different behaviours on monthly, seasonal and annual scales suggest that using the monthly data of polar motion to obtain the monthly runoff data is more practicable than borrowing the data from polar motion for river runoff on seasonal and annual scales.