Inverting In-situ Anisotropy in Global Subduction Slabs Using Deep Earthquakes and Imaging Binary Mixtures and Fractures



Journal Title

Journal ISSN

Volume Title



A large portion of deep earthquakes including intermediate-depth (100~300km in depth) and deep-focus (>300km depth) deep earthquakes are found to radiate seismic energy differently (with non-double-couple radiation patterns). Motivated by these observations, a hypothesis is proposed that the deep earthquakes are caused by shear faulting in a tilted transversely isotropic medium, thus producing a non-double-couple radiation pattern. The radiation pattern is mathematically described by the earthquake moment tensor. I used the moment tensors from Global CMT catalog to invert for the in-situ anisotropic structure for 22 deep earthquake groups in 6 global subduction slabs. I found that the anisotropy enabled systematic production of the non-double-couple radiation patterns with simple shear faulting. In most studied regions (for both depth between 100 km~300 km and >300km), the inverted TTI symmetry axes are almost perpendicular to the local slab interface. However, for three deep-focus deep earthquake groups, I found that the inverted TTI symmetry axes are parallel to the slab interface (down-dip) and coincide with local maximum stress direction. Among all the anisotropy parameters, the SH anisotropy is best determined and has a typical value of 25% (5%-46%) which is strong. The SV anisotropy is also strong, with typical values of 60% (30%-90%).The inferred anisotropy can systematically explain the non-double-couple radiation pattern without invoking exotic source processes. Many other factors may also influence radiation patterns. I tested the effect of slab heterogeneity, of outside slab anisotropic structure, of intra-slab weakly anisotropic structure (like metastable olivine wedge), and of station coverage on the non-double-couple radiation patterns of deep earthquakes using 3-dimensional elastic finite-difference modeling and full-waveform inversion of moment tensors. I found that these investigated issues cannot cause the observed non-double-couple radiation patterns and the in-situ structure with strong S anisotropy near to the earthquake focus is the simplest way to account for the abnormal radiation patterns of deep earthquakes. The existence of highly anisotropic intra-slab structures has many implications: (i) The deep earthquakes rupture as a shear dislocation, just like the majority of the shallow earthquakes do. (ii) Since shear faulting in a tilted transversely isotropic medium can also produce an isotropic component in the moment tensor (implying a volumetric change if the medium were isotropic), it may be useful to remove the constraint that the isotropic component of moment tensor must be zero during routine inversion. (iii) The intra-slab highly anisotropic structure could have significant effect on the shear-wave-splitting measurements. These are predicted consequences and I expect people to verify them in future. The second topic in my thesis is concerning the mixing of heterogeneities of variable seismic velocities. The Earth’s mantle has small scale heterogeneities caused by incomplete mixing of different rock constituents. Among those different mixing scenarios, the binary mixing (such as harzburgite and basalt due to decompression melting) has the fewest degrees of freedom and may provide an important framework for understanding the dynamics evolution of the Earth’s mantle. Using the power spectrum of the Earth’s heterogeneities obtained in observations, I developed a novel method to invert for the volumetric proportion and velocities of the two mixing endmembers and to reconstruct the stochastic mixing structure for a binary mixing scenario. The method should find wide applications in many branches of geosciences. The third topic is on applying a novel machine learning algorithm to identify discrete fractures from the double-beam interference images. The ability to obtain discrete fracture network in space is a significant improvement over the traditional double-beam method which can only give average fracture orientation and density parameters. The new methods should find applications in geothermal fields and unconventional energy development.



deep earthquakes, radiation patterns, anisotropy, binary mixing, fracture imaging


Portions of this document appear in: Li, Jiaxuan, Hao Hu, and Yingcai Zheng. "Physics-guided machine learning identification of discrete fractures from double beam images." In SEG International Exposition and Annual Meeting. Society of Exploration Geophysicists, 2019. And in: Li, Jiaxuan, and Yingcai Zheng. "Generation of a stochastic binary field that fits a given heterogeneity power spectrum." Geophysical Journal International 217, no. 1 (2019): 294-300. And in: Li, Jiaxuan, Yingcai Zheng, Leon Thomsen, Thomas J. Lapen, and Xinding Fang. "Deep earthquakes in subducting slabs hosted in highly anisotropic rock fabric." Nature Geoscience 11, no. 9 (2018): 696-700.