Abstract: When 2 events occur within a same focal region and the distance between 2 events is far shorter than the distance between events and the station, the ratio of observed spectral amplitudes of the direct P and S phases is linearly correlated with the ratio of their radiation pattern factors, and radiation pattern factors depend on geometrical property focal mechanisms. If the 2 closely located events are recorded by the same station, we calculate correlation coefficients using low-frequency spectral amplitude values, and the correlation coefficients gives a normalized measurement of the similar focal mechanism. Applying this method to earthquake sequence, the events with similar focal mechanism can be grouped, and temporal average correlation coefficient curve of earthquake sequences can be obtained. Based on above principle, we calculated and analyzed 2 strong earthquake sequences of 2001 Yajiang M6.0 Sichuan, 2003 Dayao M6.2, Yunnan, and 2 small earthquake sequences of 2002 Yangbi, Yunnan, 2003 Eryuan, Yunnan. The results show: First, the features of temporal average correlation coefficient curves are different between strong and small earthquake sequences. Second, some strong earthquake sequences' focal mechanisms are changed from scattering to centralized pattern at the beginning period. This suggests that focal stress intensity has restraining effect on earthquake's rupture direction.

Key words: Focal mechanism, Earthquake sequence, Body-wave spectral amplitude, Correlation coefficient, Cluster analysis