Evaluation of an Hybrid Refraction Tomography Code

Seismic refraction tomography is an attractive strategy for modeling near surface presenting lateral and vertical gradients and also producing static corrections, where traditional seismic refraction fails. We evaluate a hybrid inversion method that combines a global search with a local optimization method to build the velocity model from first break pickings without ray tracing. In the initial stage of the inversion, the Monte Carlo technique searches for the region of the model space where the solution is. This solution is then further improved by using a simultaneous iterative reconstruction technique (SIRT). The travel times and ray paths required for the inversion are calculated by a highly efficient grid ray tracing technique based on the Fresnel volume approach. First, we evaluate the method with synthetic data and show that the hybrid optimization method satisfactorily reconstructs the model shape and the velocity range. Then, we analyse the method with real seismic data to verify that, in both cases, the code performs at a level similar to the commercial software RAYFRACT