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.
I usually feel pretty confident that I can determine a good zero-phase match when I correlate a synthetic with the seismic data (with a stable decon operator, and good processing). Of course, a wavelet extraction technique, prior to the inversion, could also be used as a solution for getting back to a zero-phase input. However, that’s a bit more involved, and could be the subject for an entirely different series of postings. To keep things relatively straightforward for this posting, I kept it limited to using simple phase rotations.
After putting all of the plots together, one of the first things that I noticed was that from zero to 180 degrees, the background values of the inversion got “cooler”. That is, there were more blues and greens, or lower impedance values, overall. At 180 degrees of phase rotation (reverse polarity of the original input synthetic), there was an almost total lack of higher impedance values—only two yellow bands. Then, from 180 to zero degrees, the background impedance values increased back to the browns and yellows, or got “warmer”. Interesting……
At first, I wasn’t sure what to make of this. However, after thinking about it a bit, it makes perfect sense! A zero-phase wavelet will have the largest positive amplitude, and reverse polarity wavelet will have the most negative amplitude. The amplitudes of all wavelets in between these two, will be gradational from one to the other. Don’t you just love science! I wonder if you can determine zero phase by the amplitude of the inversion output—the maximum equals zero phase. The subject for yet another posting.
I also noticed that there were quite a few similarities of the calculated impedances, in the 45-90 degree ranges, as well as from the 270-315 degree ranges, to the zero-phase impedance. With a small time shift, some of them could possibly be mistaken for the zero phase response, especially if we were using real seismic data.
However, if you read my posting, “It’s Just a Phase They are Going Through“, I mentioned some distinctive characteristics of synthetic seismograms, such as “top loading”, or “bottom loading” peaks and troughs. I also think that these may be helpful correlation tools for fine-tuning inversions, if you’ve got a direct tie to a well with an acoustic impedance log. If not, the deep induction resitivity logs may be beneficial as a substitute because of their similarities with the impedance logs.
Next, I will look into the effects that AGC (Automatic Gain Control) window lengths, and frequency content have on inversion results. More to come……
Copyright © T/X RESOURCES, 1995-2008. All Rights Reserved.