Integrating Gradient Information with Probabilistic Traveltime Tomography Using the Hamiltonian Monte Carlo Algorithm
A. Zunino and K. Mosegaard
Event name: 80th EAGE Conference & Exhibition 2018 Workshop Programme
Session: WS02: Data Integration in Geoscience - Perspectives for Computational Methods
Publication date: 10 June 2018
Info: Extended abstract, PDF ( 483.45Kb )
Price: € 20
Seismic traveltime tomography is a popular methodology used to infer the velocity structure of the subsurface (Cerveny, 2001; Nolet, 2008) which has been used across all the scales from near surface imaging to global seismology. In near surface seismology inversion for traveltime is employed to construct velocity models used for further processing of seismic data and crucial to correctly assess deep structures. Formally, the relationship between velocity and traveltime can be described by the eikonal equation (Nolet, 2008), a nonlinear partial differential equation describing the arrival time for a given velocity model as a function of position. Solving numerically the eikonal equation (Vidale, 1988; Podvin and Lecomte, 1991; Rawlinson and Sambridge, 2004) for a given source, provides the traveltime at all grid nodes at once, hence saving a substantial amount of computational time compared to the traditional ray-tracing approach.