Automatic robust velocity estimation by poststack stereotomography

Stereotomography, which is based on the concept of locally coherent events, appeared to be a fast and powerful method for velocity macro-model estimation. However, in presence of low signal-to-noise ratio and coherent noise, automatic event picking on prestack data does not guarantee reliable information and may lead to wrong velocity models. In this paper, we present a new implementation of Stereotomography where the picking is performed in poststack time domain. It allows a robust and reliable picking procedure. We show results obtained using "poststack Stereotomography" on synthetic and real examples and compare these results with the conventional "prestack Stereotomography".