ADJOINT-WAVEFIELD TOMOGRAPHY OF SOUTHCENTRAL MEXICO

Waveform inversion is a newly-popular method for Earth structure imaging on local to global scales. In this study, we successfully applied the Adjoint Wavefield formulation of Full Waveform Tomography to improve a 3D crustal velocity model of southcentral Mexico. We constructed our initial model by gathering previous tomographic studies and recent research into ambient noise cross-correlations and surface wave tomography. We then computed Fréchet kernels using octree-based finite element numerical solutions of the 3D elastodynamic wave equation (Tu et al., 2006). Model updates were solved using the Conjugate Gradient Method (Tromp et al., 2005). We made more than 2,000,000 misfit observations using the Generalized Seismological Data Functional (Gee & Jordan, 1992) for different phases on records of 77 (4.5<Mw<5.5) earthquakes recorded from 2005 to 2015 at 240 three-component stations in southcentral Mexico. Additionally, we included observations from more than 250 Green’s Functions constructed from earthquake-record cross-correlations between pairs of stations. Synthetic seismograms computed using our updated models showed better fits over observed seismograms on periods larger than 5 seconds. Furthermore, our model correlated well with geological structures such as the Mexican Volcanic Belt and revealed other heterogeneities in shallow crustal depths that were not recognized previously using other seismic tomography techniques.