Tomografia de refração sísmica para construção de modelos de velocidade rasos e correções estáticas

Abstract

This work proposes a new methodology for the construction of 2D near-surface velocity models and statics corrections of land seismic data using surface refraction information. Standard refraction traveltime tomography solutions for statics corrections do not allow irregular parameterization of the model, simultaneous inversion of velocity and depth parameters, or selection of the types of first arrival times to be inverted. An algorithm, widely used in wide-angle seismic data for crustal modeling applications, was adapted for modeling and inversion of velocity and layer structure close to the surface. This flexible tomographic solution allows the definition of complex heterogeneous models through trapezoids of irregular size to parameterize the model, and it allows the simultaneous inversion of head-waves and turning-rays, which can be defined independently for each layer interface. For construction of an initial velocity model, a direct x reverse shot method was developed to estimate the velocity of the refractors and, with of these velocities, a least squares delay time method was implemented to estimate the depth of these refractors. A new method for estimate the uncertainties of the data, called standard deviations of the difference in reciprocal times, was developed and used in the tomography algorithm. The proposed methodology was evaluated on four synthetic models built with real topography information and velocity models from geophysical and geological information. It was observed that the algorithm obtained the smallest relative errors in most parameters of the model and in the estimation of statics corrections. Three real land seismic data from 2D line located in Brazilian sedimentary basins were also used. It was observed a better stacking of the seismic reflections compared to a standard software of refraction tomography and statics corrections. In addition, for two seismic lines, the inverted model provided more reliable structural depths when compared to information from wells.