Innovative RDWT: a new DWT-based method with applications for seismic ground roll attenuation

Abstract

The presence of noise in seismic data is inevitable. In land seismic data acquisition, ground roll noise masks reflection events so that observation of reflection events is not usually easy to interpret. It is the exploration seismologist's task to attenuate ground roll to improve the data quality and to enhance the signal-to-noise ratio. Investigations have suggested that the wavelet transform is an efficient tool for such a purpose. In this study, a new type of discrete wavelet transform, known as the rational-dilation wavelet transform (RDWT), is used to attenuate ground roll. Compared with the common DWTs, the RDWT offers a wide range of redundancies and Q-factors (wavelet centre frequency/bandwidth), to help the user choose an appropriate Q-factor, and hence provides more satisfactory results in ground roll attenuation while better preserving the signal. In this transform, the Q-factor is determined by selecting a number of parameters. True parameter selection results in better performance of ground roll attenuation. Depending on the nature of the ground roll, the parameters may vary in each shot-gather. Due to the over-completeness of the transformation, aliasing is less problematic compared to other DWTs. This paper discusses and indicates the advantages and capability of RDWT, by applying it to synthetic and real shot-gathered data with the purpose of ground roll attenuation, and compares the results with the application of f–k and band-pass filters.