Um estudo de estratégias de inversão de dados sísmicos via inversão da forma de onda completa no domínio da frequência

Abstract

Full-Waveform Inversion is a powerful seismic imaging technique capable of recovering highresolution subsurface models from observed data. This inversion is a nonlinear problem formulated as a local optimization problem that seeks to minimize the objective function in an attempt to find values that fit the modeled and observed data. In order to minimize the nonlinearity problem, a multiscale inversion approach is used to mitigate the sensitivity of the problem to converging toward local minima. In this work, we studied the frequency domain Full-Wave Inversion strategies (frequency and small frequency band inversion) with the Marmousi model in the presence and absence of noise. Additionally, a hybrid regularization based on Cauchy and Tikhonov constraints was proposed, in order to balance smoothness and sparsity also in noise presence and absence situations. The tests showed a superiority in the use of frequency band inversion. Moreover, for Marmousi, the proposed hybrid regularization was able to satisfactorily invert the initial model and it is a good option for data inversions in the absence of noisy components while Tikhonov was better adjusted considering the noise used and the parameters of this work.