Recuperação de propriedades acústicas da subsuperfície através da inversão sísmica por mínimos quadrados linear e não linear

Abstract

In this work, the inverse problem of exploration geophysics is solved through two techniques based on the minimization of the least squares objective function. The nonlinear problem that estimates the acoustic velocity of the medium is solved by seismic full waveform inversion. For this, two multiscale strategies were considered in order to mitigate the nonlinearity of the problem. In addition, a preconditioner term that approximates the inverse of the Hessian matrix is contemplated to assist in the imaging of the deeper parts of the model and improve the convergence of the optimization algorithm. On the other hand, the linear inverse problem that recovers the true subsurface reflectivity model is solved by the least squares reverse time migration. The numerical examples performed here demonstrate that the inverse operator of the migration problem (implicitly estimated by an iterative solver), has deconvolutive properties, and thus, it can be used to obtain improved seismic sections with better resolution and reflectors with more balanced amplitudes.