Stereotomography inversion method: Theory and application testing

Stereotomography is initially proposed to cope with difficulties associated with a highly interpretive picking in traditional reflection tomography. This method redefines data and model components in tomographic inversion, making the procedure of tracking events along continuous reflection layer unnecessary. In addition to two-way traveltime of seismic waves, the source and receiver positions combined with the local propagation directions of rays in source and receiver points are used to constrain the velocity field. Furthermore, the forward modeling can be completed only by a simple initial value ray tracing to map model component to data component. All this features enhance the robustness of inversion process. This paper simplifies data and model components of stereotomography to reduce the scale of inversion calculation. Considering the compact form of ray perturbation theory in ray centered coordinate, Fréchet derivatives of data to model are computed in this coordinate. Three 2D numerical models tests are presented to validate the proposed algorithm. The first one is demonstrated with ideally noisy free data directly calculated with ray tracing, which verified the validity of the presented simplified method. The second test is conducted with noisy data to certify the robustness and the adaptability to noise of the presented inversion processing. The third test adopts the picked kinematic properties as input data of the inversion, which sets the foundation for practical application of the presented method next step.