Deghosting Via Inversion
Publication Date: 2015-Oct-29
The IP.com Prior Art Database
We describe a work flow and algorithm for deghosting 2 D or 3 D seismic data via a constrained inversion approach Traditional deterministic deghosting methods are susceptible to artifacts when the deghosting operator contains errors in the modeling of the ghosting process or when noise is boosted The signal to noise level in the data and the ghost effect both vary in time and space This novel method adjusts the deghosting process to mitigate the artifacts in a time and space variant manner enabling an efficient automated commercial quality deghosting process to be implemented for marine seismic data
Page 01 of 10
Deghosting Via Inversion
We describe a work flow and algorithm for deghosting 2-D or 3-D seismic data via a constrained inversion approach. Traditional deterministic deghosting methods are susceptible to artifacts when the deghosting operator contains errors in the modeling of the ghosting process or when noise is boosted. The signal-to-noise level in the data and the ghost effect both vary in time and space. This novel method adjusts the deghosting process to mitigate the artifacts in a time- and space-variant manner enabling an efficient automated commercial-quality deghosting process to be implemented for marine seismic data.
During the last decade, deghosting has gained renewed interest in the field of seismic exploration. The principal aim of deghosting is to reverse the amplitude and phase effects of the ghosting process on the recorded data (Jovanovich et al., 1983, and Aytun, 1999), thereby recovering the un-ghosted wavefield (that we also refer to as the up-going wavefield). The measured wavefield is recorded in time and space dimensions and can be represented in discrete format as a multi-dimensional array (two- or three-dimensional depending on the acquisition geometry). After lexicographic ordering, the measurements can be conveniently ordered in vector . In the same way, the up-going wavefield can be represented by vector . The ghosting operator is described by operator . A block-diagram illustrating the ghosting process is provided in figure 1.
Figure 1 Block diagram of ghosting process
The ghosting model is described by
= G +
where represents additive random noise.
Three charactestics of this problem are that it is ill-posed, ill-conditioned, and time-space variant. Posed this way, the deghosting problem can be solved as an up-going wavefield recovery problem. Current deghosting solutions based on single-sensor measurements do not offer an optimal result in that they only deal partially with one or more of the properties of the ghosting problem. The ringing artifact and noise amplification are two most common issues.
Page 02 of 10
Figure 1. Application of a deterministic deghosting filter to marine seismic data that does not honor signal-to-noise issues in the data can lead to an undesirable "ringing artifact". The original marine streamer data is shown on the left. After a deterministic deghosting on the right, noise is amplified by the deghosting operator and rings through the processed data.
The objective in this paper is to recover the up-going wavefield from the recorded measurements by means of a constrained iterative inversion process. The key insight is that the problem can be cast into a series of sub-problems that are well-posed, well-conditioned and time-space invariant. For each sub-problem, the algorithm successively identifies the ghosting operator and recovers the up-going wavefield, thereby determining the pair that minimizes a certain objective function. This latter is chosen...