Full-3D Waveform Tomography for Southern California

Our full-3D tomography (F3DT) uses 3D SCEC Community Velocity Model Version 4.0 (CVM4) in Southern California as initial model, a staggered-grid finite-difference code to simulate seismic wave propagation and the sensitivity (Fréchet) kernels are calculated based on the scattering integral and adjoint methods to iteratively improve the model. We use both earthquake recordings and ambient noise Green's function data, stacking of station-to-station correlations of ambient seismic noise, in our F3DT inversions. To reduce errors of earthquake sources, the epicenters and source parameters of earthquakes used in our F3DT are inverted based on full-wave method. An automatic waveform analysis algorithm that based on continuous wavelet transforms and a topological watershed method is used to pick waveforms full-wave inversions and make frequency dependent phase and amplitude measurements. Our current model shows many features that relate to the geological structures at shallow depth and contrasting velocity values across faults. The perturbations with respect to the initial model in some regions could up to 40% and relate to some structures do not exist in the initial model, such as southern Great Valley. The waveform fittings of earthquake waveforms and ambient noise Green's function data are both improved after iterations. The earthquake waveform misfit and summation of square of ambient noise Green's function group velocity delay time between observed waveforms and updated synthetic waveforms both reduced more than 50%.