Author: Mike Cline, T/X Resources
This was initially going to be the fourth posting in my seismic inversion-related series (see the 03/08/08, 03/12/08, and 03/16/08 postings), with the title “Seismic Inversion—Frequency Sensitivity Analysis”. However, after thinking about the subject for a while, I decided to expand the scope, and shorten the title a bit, to be more general in nature. After all, a study of frequency-related seismic responses can not only be applied to inversion, but can also be used to illustrate the complications of seismic correlations between different datasets, as well as why spectral decomposition can better highlight a variety of seismic features at different frequencies.
The image below is a series of synthetic seismograms which resulted from the convolution with four different zero-phase wavelets—the wavelets and frequencies are posted at the top of the display. Since they were going to be the input for inversion examples, the synthetics are all relative amplitude (ie. no AGC, or Automatic Gain Control amplitude equalization).
See the larger Adobe Reader pdf file.
Author: Mike Cline, T/X Resources
At some point, you may find it necessary to create a seismic inversion from your existing seismic data, without having the benefit of reprocessing it. Depending on whether or not the relative amplitude processing (RAP) data is available, you may have to consider using data that has been previously gained. So that you can know the ramifications of using this non-relative amplitude data, this posting tests the sensitivity of seismic inversion to AGC (Automatic Gain Control) window lengths.
First, a little info about AGC for those who are unfamiliar with it. AGC is an ancient (technically speaking) seismic processing technique for equalizing energy absorption, but many processors still use it. Basically, it is a running-average process, and the number of samples that are averaged is controlled by the time window length. Common window lengths are 1000 ms (milli- seconds), and 500 ms, and occassionally, the processors will use different values, depending on what they are trying to do. Generally, the larger seismic amplitudes get smaller, and the smaller amplitudes get larger, with the AGC process—the amount of change depends largely on the window length.
See the larger Adobe Reader pdf file.
Author: Mike Cline, T/X Resources
As I mentioned in the previous post on seismic inversion, using zero-phase seismic data as an input for inversion, is one of the most critical elements for accurate results. However, this brought up the question of “how bad is bad”, when it comes to phase-matching errors? So, to answer this question, I had an idea to test the sensitivity of inversion results based on the phase of the input data.
Below, is a series of inversion images which were produced from the same initial synthetic seismogram. However, prior to generating the inversion, I rotated the phase of the input data in the amount indicated at the bottom of each image—that is: 0, 45, 90, 180, 270, and 315 degrees.
See the larger Adobe Reader pdf file.
Author: Mike Cline, T/X Resources
I recently wondered why I hardly ever see anyone using the Seismic Inversion tool, found in SMT’s TracePak module? Maybe you’ve thought about using it, but didn’t understand it well enough, or maybe tried it once, and the results didn’t match anything in the well(s). Like anything new, if you don’t under- stand it, it’s going to be difficult to use it properly. So, I thought that it would be helpful to explain some of the benefits and pitfalls of using inversion, for those interested.
First, for those unfamiliar with inversion, what is it, and how do we use it in our interpretation? You could think of seismic inversion as the reverse of a synthetic seismogram processing flow, and we use inversion to get some idea about rock properties. For an example, the portion of the seismic inversion in the image below, was generated from a synthetic seismogram in the well at the center of the line. Normally, you would generate the inversion from an actual seismic line, but I wanted an optimum response for this example. I’ve also posted three well logs from this well: the spontaneous potential (aka. SP log) in blue, the deep resistivity (RES log) log in magenta, and the acoustic impedance (AI log) in red.
See the larger Adobe Reader pdf file.
Author: Mike Cline, T/X Resources
Why bother using synthetic seismograms (aka. synthetics) to calibrate well info to our seismic data? Simple answer, TO REDUCE DRILLING RISK !
For example, I’ve seen prospects “evaporate” because the originator was mapping the wrong event—or just as bad, started mapping on the correct event, but ended up on the wrong event due to a character, or response change in the seismic data. This only became evident after a couple of synthetic correlations!
I also continue to see prospects that are being sold on the strength of an amplitude, or avo response, that is somehow related to a key wellbore. However, often a synthetic hasn’t been used to tie (correlate) the well to the seismic data. How could they even know for sure what was causing the anomaly, without a synthetic tie?
So, with these recent real-life examples in mind, I thought that it would be a good idea to cite some reasons why we should use synthetics, with a blog posting.
.