The video shows the use of the Wiener filter to study the spatial variability of a seismic section, relatively to a chosen trace of the section and taken as a reference.
The difference between the initial section and the predicted section by the Wiener filter clearly highlights the spatial variability. In this case, the Wiener operator is different for each trace of the section. The difference section between the initial section and the section predicted by the Wiener filter is also called error section or residual section. The trace chosen as a reference is selected. The trace #3 is a good model for this deconvolution (at the opposite of trace16). It is possible to select one of the following seismic sections:

• Initial section (Input);
• Predicted section by the Wiener filter (predicted);
• Difference section between initial section and predicted section by the Wiener filter (Error).

It is possible to modify the following parameters:

• Correlation length on the seismic trace;
• Length of Wiener's operator.

The Wiener filter which is used here is non-causal filter. The Wiener filter computed on a time window from a seismic trace called correlation window allows to define an operator which be applied on whole seismic trace. This operator minimizes the difference between the input trace and the reference on the correlation window. The error trace associated with the reference will be a null trace. In this example the main energetic wave is located between 120-160 ms. The error section shows very weak residual waves in this time interval. A short length operator (50 ms) is efficient to extract the dominant wavefield. The error section shows small ringing traces. Here, the Wiener filter is used to study the spatial variability of seismic layers versus layers located in 120-160 ms. It also shows that the seismic horizon situated approximately at 550 ms presents spatial variations in amplitude which are quite significant.