Velocity Model Estimation Based On Common Reflection Surface Gather

In recent years, professor Hubral has put forward a new imaging technology called Common Reflection Surface,which is only dependent on the near surface velocity and foreign to the macroscopic velocity model. It describes the kinematics reflected response of non-uniform medium curved interfaces by analysis formulas.It realizes seismic imaging by optimizing emergent angleα,the wavefront curvature radius of normal incidence point wave(NIP)and the wavefront curvature radius of normal direction wave(N).Its theoretical basis is geometrical seismology.It considers the local feature of reflector and the whole reflect of the first Fresnel zone, hence it utilizes all the information of multiple coverage reflect data effectively.The determination of a suitable velocity model is one of the crucial steps for seismic depth imaging in laterally inhomogeneous media. There are two methods of velocity estimation: migration velocity estimation methods and reflection tomography . Popular migration velocity estimation methods are usually based on residual moveout analysis in common image gathers. They use the criterion that in the correct velocity model, therefore, expensive in terms of computation time. An often-used tool for determining velocity models for migration is reflection tomography. The drawback of this method is the tremendous amount of picking necessary to obtain traveltimes from the prestack data and the assumption of continuous reflectors, often across the entire section. Although many attempts to automatize the process have been made, the problem of identifying and picking reflection events remains.In many cases it is not the complexity of the subsurface that makes the construction of a velocity model for depth imaging difficult, but the presence of a high level of noise in the prestack seismic data. In this paper, a tomographic velocity estimation method is presented that uses kinematic information extracted from the prestack data in a data-driven way by applying the common-reflectionsurface (CRS) stack. Using the CRS stack technology,we can obtain a high quality ZO section and three wavefield parameter sections,by which the best velocity model can be deduced. Input data for the inversion can then be picked in the resulting simulated zero-offset section of significantly improved S/N ratio. The inversion is performed with a tomographic approach. In that method, source and receiver rays pertaining to a reflection point are considered, whereas in the approach presented here, properties calculated along normal rays by dynamic ray tracing are used to approximate locally the kinematic multioffset response of reflection points. The inversionmethod is based on the above-mentioned criterion for a correct velocity model: In a correct model, all considered NIP waves,when propagated back into the earth along the normal ray,focus at zero traveltime. This method is suit for the case of low S/N ratio and need less computer time then others.