利用大地测量数据反演断层滑移分布的MCMC方法

Markov Chain Monte Carlo Method Used to Invert for Fault Slip from Geodetic Data

断层滑移的反演可以表达为一个不等式约束下的正则化问题。本文基于改进的MCMC方法,利用GPS位移数据反演2015年尼泊尔地震的同震滑移分布。结果表明,主要滑移集中在震中东侧靠近加德满都的区域,最大滑移约7m,最大滑移误差约为滑移量的10%,地震矩为7.67×10^(20) Nm,相当于M_W7.86,与其他手段得到的结果一致。利用反演模型预测的GPS位移与观测相吻合。震例结果表明,改进的MCMC方法适用于断层滑移分布的反演,并能够从统计角度给出滑移误差。

Fault slip inversion can be represented by a regularization problem constrained with inequali- ties. Based on the improved MCMC approach, we invert for coseismic fault slip distribution of the 2015 Nepal megathrust event from GPS observations. Results show that the slip concentrates at a re- gion eastward to the epicenter, near Kathmandu, with a maximum of 7 m. The slip errors statistically obtained by resampling model space are about 10% of slip. The seismic moment is 7.67 × 10^20 Nm, e- quivalent to Mw7.86, consistent with that from other research. The modeled GPS displacements agree well with the observations. Our results demonstrate that the improved MCMC method is appropriate for fault slip inversion from geodetic data and especially can give the slip errors statistically.