Efficient Seismic Forward Modeling using Simultaneous Random Sources and Sparsity

This paper proposes an approach to speed up seismic forward modeling when the Green’s function of a given model is structured in the sense that the Green’s function has a sparse representation in some known transform domain. The first step of our approach runs a forward finite-difference (FD) simulation for a duration longer than each conventional run, but with all sources activated simultaneously using different long random noise waveforms. The cumulative responses to the simultaneous sources are measured at all receivers. The second step separates the interfering Green’s functions from the receiver measurements by exploiting prior knowledge of the random waveforms and the sparsity of the Green’s function in a suitable domain. Simulation results demonstrate such a simultaneous source approach is indeed promising.