Summary
Conventional adaptive-subtraction based ground rolls attenuation approaches first predict an initial model of ground rolls and then adaptively subtract it using a stationary matching filter (MF). Because of the non-stationary property of seismic data and ground rolls, the application of a traditional stationary MF is not physically plausible. Thus, in the case of highly non-stationary seismic reflections and ground rolls, a stationary matching filter (MF) cannot obtain satisfactory results. In this paper, we apply a non-stationary matching filter (NMF) to adaptively subtract the ground rolls. The NMF can be obtained by solving a highly under-determined inversion problem using non-stationary autoregression (NAR). We apply the proposed approach to one synthetic example and two field data examples and demonstrate the much improved performance compared with the traditional MF approach.
INTRODUCTION
Ground rolls are a type of seismic noise, with a high amplitude, low frequency, and low velocity. Ground rolls are the main type of coherent noise for land seismic survey. Those ocean bottom node (OBN) based marine seismic surveys may also contain this type of noise (Chen et al., 2014). The ground rolls usually mask the shallow reflections at short offset, and deep reflections at larger offset (Claerbout, 1983; Saatilar and Canitez, 1988; Henley, 2003), and must be removed before the subsequent processing tasks. Instead of being incoherent along the spatial direction like random noise (Chen and Ma, 2014; Yang et al., 2015), the ground rolls are coherent and behave much similarly to the primary reflections, which makes the removal of them very difficult using simple signal processing methods. There have been a lot of researches about removing ground rolls done in the literature and many researchers have proposed different methods for handling the ground rolls problem (Shieh and Herrmann, 1990; Brown and Clapp, 2000). The simplest and the probably most widely used way for removing ground rolls is by bandpass filtering. Because of the low-frequency property, a simple high-pass filter is usually applied to the seismic record to attenuate the ground rolls. However, the low-bound-frequency (LBF) for the bandpass filter is not easy to choose, because higher LBF will cause much damage to primary reflections while lower LBF will fail to remove enough ground rolls. This issue exists because of the coupling of primary reflections and ground rolls in the frequency bands. In order to deal with this problem, we can either make improvement to the traditional frequency-based approach or devise new approaches from different perspectives.
