A NUMERICAL STUDY ON FINDING AN OPTIMUM MODEL IN VELOCITY TOMOGRAPHY BY USING THE EXTENDED INFORMATION CRITERION

We performed numerical experiments to show that Extended Information C riterion (EIC) can be utilized for determining an optimum model in vel ocity reconstruction problems, in which a wavefront tracing method was employed for finding the minimum travel time raypath. First, travel t imes were calculated from a specified velocity model, and simulated da ta were produced by adding random error to the travel times. Velocity models were then reconstructed from the simulated data by employing a Simultaneous Iterative Reconstruction Technique (SIRT) starting from d ifferent initial models. Finally, EIC values were calculated for each reconstructed velocity model at every iteration, and were used for sel ecting an optimum velocity model. Results indicated that EIC provides an objective method for selecting an optimum solution from a suite rec onstructions obtained from different initial models.