Swell Noise Solution

Swell Noise Solution

CGGVeritas algorithm FXEDIT.

FX domain noise attenuation routine.

User specified spatial, time & frequency gates transformed to the FX domain.

Frequency components within the gates are flagged as bad if their amplitudes are a user specified amount greater than the median amplitude of the gate.

Frequency components that are flagged are zeroed & replaced with signals constructed by an FX projection filter.

The emphasis here is that only flagged frequencies are dealt with, and that they are dealt with in the FX domain ie both frequency and amplitude are used in conjunction to identify the noise safely.

Seismic interference is caused by the recording of shots fired by other acquisition vessels on seismic data. The resulting noise packets can become complicated if they are observed over long distances as they can exhibit dispersion and multi-modes, depending upon the water depth. A number of techniques can be used to suppress seismic interference, including: If the coordinates of the noise source are known, then it can be easily modelled and subtracted.

-p domain extraction of the noise and its adaptive subtraction can work well for predominately head-to-tail or tail-to-head noise. De-spike approaches based on XWAVE and FXEDIT applied in a domain in which the noise is randomised.Noise Problem

Linear noise/direct arrivals preventing the use of longer offset information for:

Far angle stacks

General wider angle AVO attributes

Improved fold

Anisotropy parameterisation (Eta, Epsilon & Delta derivation)

General acquisition related linear noise such as screw noise.Linear Noise Solution

CGGVeritas algorithm XRLIN performs a constrained higher resolution linear transform of the input shots into the Tau-p domain. Noise and primaries are thus distinguished from one another based on their respective dips. This transformation gives a better focussed picture of events in Tau-p when compared against more typical least square or direct summation transformations. This superior focussing of events provides clearer separation between primary & noise hence leading to safer noise subtraction.

This technique is able to resist spatial aliasing issues, which would typically cause problems for more conventional FK style methods. As such it is particularly suited to direct arrival elimination within the shots.

Noise Problem

Noise reflected back, or generated from, passing vessels (non seismic vessels) generating localised non-predictable noise bursts within the shots. Noise Solution 1

Forward Wavelet transform (X-wave) to split the data into a number of overlapping frequency bands (voices)

Use amplitude discrimination techniques within specific voices ie both amplitude and frequency are used in conjunction to distinguish the noise from the primary data.

Re-construct the data using the inverse transform.

Wavelet transform splits the data in terms of time & frequency simultaneously (traditional Fourier transform splits signals in time & frequency but separately) ie frequency decomposition preserving both spatial & temporal information.

Noise Problem

Seismic interference generated by passing seismic vessels generating coherent noise trends within the shots.

Noise Problem

Noise reflected back from stationary objects (rigs) generating localised predictable noise bursts within the shots.

Noise Solution (External Noise X/Y Modelling) Build a model of the noise using the real world coordinates of the noise generator & the shot/receivers.

Subtract this model from the input data

If X/Y coordinates not known then model can be predicted from the timing & positioning of the noise on shots themselves.