Data space reflectivity full waveform inversion

The full waveform inversion of seismic data aroused the hope to perform simultaneously and in automated way tomography and imaging by solving non-linear least-squares optimization problem. As it has been recognized early, brute force minimization by classical methods is hopeless if low time frequencies are absent in the data. The paper develops a reliable numerical technique for smooth velocity reconstruction via model space decomposition. We present realistic synthetic examples for validating presented algorithm.