利用GPS观测的时间序列资料反演地壳地幔黏性结构

地球介质的黏度是地球动力学研究中的重要参数. 在考虑震后震源区介质物性变化的情况下，本文提出利用GPS观测的时间序列资料反演地壳、地幔黏度的方法. 计算中，将遗传算法与Marquardt法相结合，利用遗传算法搜索结果作为Marquardt法反演的初始模型，运用微扰法形成Jacobian矩阵. 根据余震分布及地震波速度变化确定震源区的范围及介质的物性变化情况，利用黏弹性有限单元模型（Maxwell体）及台湾集集地震后GPS观测的时间序列数据，反演了台湾地区地壳、地幔的黏性系数. 反演计算及数值实验表明，该方法反演黏度的效率高，稳定性好. 反演结果初步显示台湾地区的地壳、地幔的黏度分别为1.2×1018Pa·s和3.6×1019 Pa·s.

Inversion of viscous properties of crust and mantle from the GPS temporal series measurements

The viscous properties of Earth media are very important in geodynamics. Taking the physical property variation of focal media after earthquake into account, we propose a method to invert for the viscosities in lower crust and upper mantle from GPS temporal series measurements. Combining the genetic algorithm (GA) and Marquardt inversion, we regard the optimum model found by GA as the initial model to carry out inversion by Marquardt method, in which the Jacobian matrix is produced by small perturbation. The focal region and physical properties are decided by aftershocks and velocity changes of seismic wave. We apply viscoelastic finite element method to determine the viscosities in Taiwan with GPS measurements after Chi_Chi, Taiwan, earthquake. Inversion computation and digital experiments show that the inversion method is highly efficient and stable. The preliminary result suggests that the viscosities of lower crust and upper mantle in Taiwan are 12×1018Pa·s and 36×1019Pa·s, respectively.