View image
How to open the spectral decomposition dialog box
Spectral decomposition produces tuning cubes for analyzing reservoir thickness.
This new type of seismic trace attribute analyzes the frequencies in the seismic trace over a small time window centered on your zone of interest. The reflections off the top and bottom of your reservoir constructively or destructively interfere depending upon reservoir thickness and wavelet frequency. Each frequency of the wavelet ‘tunes’ at a specific reservoir thickness. Correlate your tuning cube results with a model study to precisely determine reservoir thickness distribution.
The Spectral Decomposition dialog box enables the user to select data sets
and to specify the computation parameters. View image
The Time Window controls are used to determine which time samples on each of the selected seismic traces are used for the spectral decomposition calculation.
Do not expect to resolve low frequencies if your time window is shorter
than a wavelength. For
instance, a 10Hz wave has a period of 100ms, and you will need a time
gate of 100ms or more to adequately image frequencies near 10Hz. View image
Normally your analysis will be centred off a picked horizon. You can use both positive and negative times for Time
Above and Time Below. The
times you enter are with respect to the reference horizon. View image
The following table shows examples of how to specify a variety of time windows with respect to a reference.
| Desired time Gate with Respect to the reference Horizon | Gate Length (ms) | Time Above (ms) | Time Below (ms) |
|---|---|---|---|
|
Centered on Reference |
100 |
50 |
50 |
|
Start at reference, end 100ms below reference |
100 |
0 |
100 |
|
Start 50ms below the reference, end 150ms below the reference |
100 |
-50 |
150 |
|
Start 100ms Above the reference, end at the reference |
100 |
100 |
0 |
|
Start 150ms above the reference, end 50ms above the reference |
100 |
150 |
-50 |
Use the time gate controls to choose either a Hanning
or Hamming taper style. View image
Suppress tapering by selecting the rectangular option. Tapering the time series prior to applying the FFT is highly recommended. If your time series terminates on a peak or trough and you do not taper, broadband noise will be created by the abrupt termination.
Specify the output frequency range and frequency interval. View image
Normally you will not change these numbers unless output file size is a consideration. Just keep in mind that your time window length introduces low frequency resolution limits.
The spectral decomposition results can be displayed in either a linear
or logarithmic scale. View image
The output defaults to a linear scaling.
By default the individual frequency slices are not normalized. In
order to remove the source wavelet, you may want to normalize the values
of each frequency slice. View image
You can choose to normalize, the mean, RMS, or standard deviation of each frequency slice.
The Phase I spectral decomposition results are treated as new 2D lines and 3D areas in your project. These new data sets are located at the exact same location as the source data. You must use the ‘b = send to back’ hotkey to activate the data you are interested in displaying.
The Phase I program creates new SEG-Y files containing the F-X displays and new time slice files that contain the tuning cube data. These new files are treated as additional surveys by WinPICS. Currently no special display behavior is implemented and when displaying these data, the frequency axis will be labeled as if it were a time axis.
There are no negative values when displaying the linearly scaled spectral decomposition results. Choose an appropriate colour palette and bias it to one-sided positive values for optimal colour bar saturation.
Limited operational testing has shown that the results are sensitive to the time gate length and contents. Make your gate long enough for the lowest frequency wavelength of interest, and make your gate short so that it contains only the stratigraphic sequences whose thickness and character you are analyzing.
The Phase I spectral decomposition program creates new SEG-Y files containing the F-X displays and new time slice files that contain the tuning cube data.
These new files are treated as additional surveys by WinPICS. Currently no special display behavior is implemented and when displaying these data, the frequency axis will be labeled as if it were a time axis.
A list of the files created during spectral decomposition is shown below:
Original_NameAmp = linearly scaled F-X data
Original_Name_dB = logarithmically scaled F-X data (optionally created)
Original_NameAmp = linearly scaled tuning cube
Original_Name_dB = logarithmically scaled tuning cube (optionally created)
Original_NameAmp.txt = linearly scaled trace statistics
Original_Name_dB.txt = logarithmically scaled trace statistics
1) L. Peyton, R. Bottjer, G. Partyka, 1998, Interpretation of incised valleys using new 3-D seismic techniques: A case history using spectral decomposition and coherency: The Leading Edge, Sept. 1998, 1294-1298
2) G. Partyka, J. Gridley, J. Lopez, 1999, Interpretational applications of spectral decomposition in reservoir characterization: The Leading Edge, Mar. 1999, 353-360
