利用独立成分分析检测2004年和2012年印度洋地震的重力变化

2004年12月苏门答腊发生M_W9.3地震，造成巨大的质量重新分布.利用GRACE卫星月重力场数据计算了研究区域地面1°×1°网格点上的重力变化时间序列，采用主成分分析和独立成分分析两种方法，提取了重力变化的空间与时间特征，结果显示震中两侧区域的重力变化呈两极分布，其中东侧重力下降，西侧重力增加.相较于传统的主成分分析方法，独立成分分析能更好地从原始信号中提取地震的信息，能分解出具有显著阶跃变化的独立成分.除了2004年的苏门答腊大地震外，独立成分分析还分解得到了2012年Andaman地震的特征，与该次地震的空间特征与时间序列基本一致.相较于适合定量分析的多项式拟合方法，独立成分分析更适合大范围区域的定性分析，建议将两种方法相结合，取长补短，从而为GRACE地震监测提供一种更为客观、有效的方法.

Using independent component analysis to detect the gravity changes in the Indian Ocean Earthquakes of 2004 and 2012

A massive redistribution of mass was caused by the M_W9.3 earthquake that occurred in Sumatra in December 2004. By using the gravity data of GRACE satellite, the time series of gravity change on the ground 1°×1°grid points are calculated, and the spatial and temporal characteristics of gravity change are extracted by using Principal Component Analysis and Independent Component Analysis, and the results show that the gravitational change in both sides of the epicenter has the character of polar distribution, the gravity of the up disk in the fault decreases, while the gravity of the down disk increases. Compared with the principal component analysis method, independent component analysis can better extract seismic-related information from the GRACE gravity signal, and can decompose independent components with significant gravity changes. In addition to the 2004 earthquake in Sumatra, the independent component analysis also decomposed the characteristics of the 2012 Andaman earthquake, which is basically consistent with the spatial characteristics and time series of the earthquake. Compared with the polynomial fitting method which is suitable for quantitative analysis, independent component analysis is more suitable for qualitative analysis of large-area regions. The two methods can be combined to provide a more objective and effective method for GRACE seismic monitoring.