Fast Azimuthally Anisotropic 3D Radon Transform by Generalized Fourier Slice Theorem

TL;DR

This paper presents a fast, anisotropic 3D Radon transform algorithm that enhances seismic data processing by improving resolution and enabling better medium parameter estimation, using a generalized Fourier slice theorem and sparse constraints.

Contribution

It introduces a novel, efficient anisotropic 3D Radon transform method employing the generalized Fourier slice theorem and sparsity constraints, significantly reducing computation time.

Findings

Improved temporal resolution of seismic data.

Effective construction of azimuthally dependent NMO velocity curves.

Demonstrated success on synthetic datasets.

Abstract

Expensive computation of the conventional sparse Radon transform limits its use for effective transformation of 3D anisotropic seismic data cubes. We introduce a fast algorithm for azimuthally anisotropic 3D Radon transform with sparsity constraints, allowing effective transformation of seismic volumes corresponding to arbitrary anisotropic inhomogeneous media. In particular, a 3D data (CMP) cube of time and offset coordinates is transformed to a 3D cube of intercept time, slowness, and azimuth. The recently proposed generalized Fourier slice theorem is employed for very fast calculation of the 3D inverse transformation and its adjoint, which are subsequently used for efficient implementation of the sparse transform via a forward-backward splitting algorithm. The new anisotropic transform improves the temporal resolution of the resulting seismic data. Furthermore, the Radon transform coefficients allows constructing azimuthally dependent NMO velocity curve at any horizontal plane, which can be inverted for the medium anisotropic parameters. Numerical examples using synthetic data sets are presented showing the effectiveness of the proposed anisotropic method in improving seismic processing results compared with conventional isotropic counterpart.