Seismic reflection data sets recorded on land are often contaminated by coherent ground-roll noise generated by the propagation of dispersive waves along the free surface. For crustal-scale investigations, this ground-roll contamination can be particularly harmful as the higher amplitude, low-frequency noise overwhelms low-frequency signals coming from deep reflectors. Consequently, conventional ground-roll suppression techniques which rely on frequency separation of ground roll from signal become ineffective for crustal studies. This paper presents the successful use of a new 2D wavelet method based on frame theory (physical wavelet frame denoising) in removing ground roll from a deep 3D reflection data set intended for the study of upper crustal Precambrian mafic sills in southwestern Alberta, Canada.

You do not currently have access to this article.