中文摘要：

强震持时计算分析中,选取记录不同分量会导致持时的离散性。针对持时与传感器布设方向存在不合理的依赖性这一问题,引入旋转不变平均持时概念。分析了断层距小于300km的87个台站汶川地震加速度记录的旋转不变平均持时与记录持时及合成时程持时的相关性,探讨了旋转不变平均持时随断层距的空间变化特性。结果表明：旋转不变平均持时与记录持时及合成时程持时均具有强相关性,其中：旋转不变平均持时与合成时程持时的相关性强于与记录持时的相关性;旋转不变平均持时与合成时程的重要持时的相关性强于Bolt持时,90%重要持时相关性强于70%重要持时,绝对Bolt持时相关性强于相对Bolt持时。分析认为可将合成时程的90%重要持时代替不受加速度传感器布设方向影响的旋转不变平均持时。旋转不变平均重要持时和相对Bolt持时将随断层距的增大而增加,旋转不变平均绝对Bolt持时则将随断层距的增大而减小。

英文摘要：

Different components of the stations recording will lead to the discrete duration analysis. In this paper, Wenchuan seismic three components acceleration records of 87 stations with fault distance less than 300km were chosen, and the definition of significant duration and Bolt duration were delimited. The correlations between the ro- tation-invariant mean duration and the durations of the recorded, as well as the correlations between the rotation- invariant mean duration and the duration of resultant motion were analyzed. The spatial variability law of rotation- invariant mean duration with fault distance was discussed. The results indicate that there are significant correlations between rotation-invariant mean duration and the durations of the recorded and between rotation-invariant mean du- ration and the resultant motion. By comparing the different definitions of duration, there are more significant corre- lations at the definition of 90% significant duration between rotation-invariant mean duration and the durations of the recorded and between rotation-invariant mean duration and the resultant motion than that of 70% significant du- ration. Furthermore, the correlation of the absolute Bolt duration is more significant than that of relative Bolt, and the correlation of significant duration is more significant than that of Bolt. What＇ s more, there are the strongest sig- nificant correlations between the rotation-invariant mean 90% significant duration and the 90% significant duration of resultant motion, and the rotation invariance of 90% significant duration of resultant motion is equivalent with the rotation-invariant mean duration. It is suggested that rotation-invariant mean duration can be reasonably replaced by 90% significant duration of resultant motion in calculating the duration, and it can be deemed that the concept is concise and is not affected by the layout direction of the acceleration sensor. Last but not least, rotation-invariant mean significant duration and relative Bolt duration would inc