Seismic random noise attenuation analysis with sparse representation from shallow sandstone-hosted uranium deposits

To attenuate random noise while protecting effective signals is one of the key issues in the seismic exploration of shallow-buried uranium deposits in sedimentary basins. This work focuses on the sparse representation of seismic data and analyzing different methods in terms of their denoising performances such as median filtering, Fourier transform, wavelet transform, and curvelet transform. Both synthetic and real data examples demonstrate that seismic data in the curvelet domain has the best sparsity and curvelet transform is most suitable for seismic wave fronts characterization and random noise attenuation in the field of sandstone-hosted uranium exploration.