Joint Inversion of Body-Wave Arrival Times and Surface-Wave Dispersion Data in the Wavelet Domain Constrained by Sparsity Regularization

Tuesday, 16 December 2014: 9:15 AM
Haijiang Zhang1, Hongjian Fang1, Huajian Yao1, Monica Maceira2 and Robert D van der Hilst3, (1)USTC University of Science and Technology of China, Laboratory of Seismology and Physics of Earth's Interior, Hefei, China, (2)Los Alamos National Laboratory, Los Alamos, NM, United States, (3)Massachusetts Institute of Technology, Cambridge, MA, United States
Recently, Zhang et al. (2014, Pure and Appiled Geophysics) have developed a joint inversion code incorporating body-wave arrival times and surface-wave dispersion data. The joint inversion code was based on the regional-scale version of the double-difference tomography algorithm tomoDD. The surface-wave inversion part uses the propagator matrix solver in the algorithm DISPER80 (Saito, 1988) for forward calculation of dispersion curves from layered velocity models and the related sensitivities. The application of the joint inversion code to the SAFOD site in central California shows that the fault structure is better imaged in the new model, which is able to fit both the body-wave and surface-wave observations adequately.

Here we present a new joint inversion method that solves the model in the wavelet domain constrained by sparsity regularization. Compared to the previous method, it has the following advantages:

(1) The method is both data- and model-adaptive. For the velocity model, it can be represented by different wavelet coefficients at different scales, which are generally sparse. By constraining the model wavelet coefficients to be sparse, the inversion in the wavelet domain can inherently adapt to the data distribution so that the model has higher spatial resolution in the good data coverage zone. Fang and Zhang (2014, Geophysical Journal International) have showed the superior performance of the wavelet-based double-difference seismic tomography method compared to the conventional method.

(2) For the surface wave inversion, the joint inversion code takes advantage of the recent development of direct inversion of surface wave dispersion data for 3-D variations of shear wave velocity without the intermediate step of phase or group velocity maps (Fang et al., 2014, Geophysical Journal International). A fast marching method is used to compute, at each period, surface wave traveltimes and ray paths between sources and receivers.

We will test the new joint inversion code at the SAFOD site to compare its performance over the previous code. We will also select another fault zone such as the San Jacinto Fault Zone to better image its structure.