Hamidi, R., Javaherian, A. and Reza, A.M., 2013. Eigenimage wavelet transform for ground roll attenuation: a case study on an Iranian oilfield. Journal of Seismic Exploration, 22: 251-270. Coherent noises such as ground roll, guided waves, multiples and refractions are usually present in seismic records. In land surveys, the ground roll can cause serious problems since it has higher energy and stronger amplitude than reflected signals. In this case, detection of reflections becomes very difficult and sometimes impossible. Common methods in suppressing the ground roll noise are based on high-pass and frequency-wave number filters. In spite of the fact that these filters are commonly used in attenuating the ground roll noise, they have some limitations, e.g., signal distortion, data aliasing and artifacts. Therefore, we need to consider alternative methods. Wavelet transformation provides a mean to analyze signals simultaneously in both time and frequency. With this approach, we can analyze the time evolution of the signal’s frequency content. Therefore wavelet-based filters can easily differentiate between early arriving ground roll signals and late arriving signal of interest. In other words, signal components at higher scales (lower frequencies) and specific time periods consist mainly of noise. Since there is usually some overlap between the (desired) signal and noise, it is necessary to introduce a measure to determine whether a certain time-scale region is dominated by noise or signal. Here we use singular value decomposition as a criterion to separate signal from noise. Our results show that proper setting of the new filter parameters result in distinct signal and noise frequency bands, arrival times, and energies. This approach causes less signal distortion when compared with conventional 1D wavelet transform or singular value decomposition.