Abstract: In the context of seismic data processing, the presence of multiples poses inherent challenges to the imaging and interpretation of seismic data. The effective suppression of these multiples stands as a key issue in seismic exploration. Leveraging its high efficiency, the parabolic Radon transform emerges as a widely used technique for multiple suppression. However, in field seismic data acqisition, due to the limited offset, energy diffusion and illusions reduce the effect of multiple suppression in the Radon domain. In response to this challenge, we propose a L1/2-regularized high-resolution parabolic Radon transform with sparse inversion, where the inverse problem is solved by  generalized iterated shrinkage algorithm (GISA). The L1/2 regularization chosen for its robust sparse constraint capabilities plays an important role in enhancing the solution sparsity and improving the signal-noise separation. Compared with the least square inversion and the sparse inversion method based on L1 regularization, the L1/2-regularized sparse inversion of using the high-resolution parabolic Radon transform can suppress multiples effectively and ensure the consistency between the reconstructed data and the original data.

Key words: multiple suppression, high-resolution parabolic Radon transform, L1/2 regularization