固物研究生论坛 - First-arrival Travel Time Tomography
First-arrival travel time tomography is an efficient approach for imaging the near-surface structures. However, it can be very time-consuming if the input data are too large. In this study, we improve the efficiency of travel time tomography by simultaneously expanding the travel time wavefront of several sources as a group instead of a single source each time. These sources are selected with reasonable spacing through stochastic sampling scheme so that most of their first arrivals can be preserved in the calculation.
For further retaining the deep refractions and wide-angle reflections in the first arrivals, we apply random time delay to source time within a group. These efforts significantly save computation time if we employ a graph method for calculating the wavefront with the use of BUCKET algorithm without any sorting involved. The BUCKET algorithm allocates intervals for travel times in the velocity model allowing a slow linear increase of computation time with the wavefront grids. Within each interval, travel times of different grids have no causal relationship to one another so that any candidate point can be considered as the next secondary source. A series of 2D synthetic and real data tests with simultaneous sources show 50% to 90% of runtime reduction to achieve comparable results with the standard approach.