Deconvolução iterativa tipo Wiener-Levinson

The importance of the seismic resolution, and of the quality of the seismic images to the industry of petroleum, has motivated the development of new filtering techniques for processing of the seismic reflection data. In this paper we present a new filtering method to improve the resolution of the seismic sections. Unlike the conventional Wiener-Levinson approach which uses causal filter and L2 norm, the proposed method is iterative and uses causal and anticausal filters, combined with variable norm. Numerical results using synthetic seismic data demonstrates its ability to overcome some limitations of the conventional Wiener-Levinson method, providing good results even when the wavelet is not minimum-phase or the reflectivity series is not a random process. Numerical examples using poststack marine seismic reflection data, demonstrates the efficacy of the new iterative wavelet deconvolution method.