Noise Attenuation

TGS has several powerful noise attenuation algorithms and below we describe some of our favorite approaches.


ICone™Intelligent Coherent Noise Elimination

ICone™ is a time-space algorithm designed for shot domain coherent noise rejection. ICone™ evaluates event coherency across a range of trial linear trajectories (i.e., velocities) in order to identify and remove organized noise trains. The algorithm supports both 2D and 3D implementations, and an interactive graphical tool facilitates optimal parameter selection and testing.

Click below to view our ICone™ interactive demo. 


The figure below shows ICone™ at work on a 2D shot record.


STONE™—Signal TO Noise Enhancement

STone™  is a dip-scan algorithm designed for random noise attenuation. The algorithm, which is typically run after application of statics and velocities, supports both 2D and 3D implementations and it can be run in any prestack domain --shot, receiver, common-offset, common-offset-vector, cross spread--or even post stack.  STone™ is a data driven technique which has been shown to preserve relative amplitudes and to be robust in areas of conflicting dips.

The figure immediately below shows the cascaded application of STone™ in the shot domain followed by the receiver domain on a 2D land data set. Note the dramatic improvement in signal-to-noise ratio.

Before Stone After Stone


Domain of implementation of noise rejection filters: mini-case-study

The domain of implementation of prestack noise rejection filters can have a profound influence on final image quality.  In the short case study below, we explore the relative pros and cons of running industry-standard fxy filtering  in various data domains for the Groundbirch 3D, a speculative data set owned by TGS in partnership with Olympic Seismic. In the figure below, the top left pane shows the input data, and columns (a), (b), and (c) show the results after applying fxy decon in common offset, common offset vector (also known as “offset vector tile”), and cross-spread domains, respectively. In each of these three columns, the top pane shows the output data after CMP stacking, and the middle and bottom panes show the offset and azimuth distributions for a single representative ensemble.


Input Noise attenuation


The three stacked data displays (i.e., top pane in each column) exhibit differences in both noise attenuation efficacy and preservation of subtle reflector features. Such differences are related to the relative pros and cons of the various domains of implementations, and specifically to the degree of localization of offsets and azimuths across neighboring cmp’s. While a full discussion of these pros and cons is beyond the scope of this mini-case-study, suffice it to say that TGS’ processing group has a strong grasp of the tradeoffs surrounding the approaches!


Back To Top