1. A history of misguided pre-stack processing.
A repeat of the original show on this subject to provide reference for a new example. Click on oval to bypass -
Below you see two gather sets before and after pre-processing. On the left is the raw data, whereas NMO and pre-stack migration have been applied at the right. The yellow oval encircles a critical angle crossing of the North Sea chalk horizon. This phenomenon caused the reflection event to clone into a refraction, interrupting the downward flow of energy and causing more deep refractions. The energy explosion on the right panel is apparently caused by the pre-stack migration not being able to handle the generated noise. The raw data came at the end of the project. The next slide shows how much was lost and the rest of the presentation shows the recovery. Click to continue.

2. What was clobbered. The raw data -
By the time any interpreter sees the data, most coherent noise patterns have been lost in the heavy mixing used by pre-stack migration. This is a prime example of bad priorities, since there is no hope of improving location accuracy in the presence of the heavy noise we will shortly point out.
While the myriad of noise events I see on the raw input might be easily visible to you, it generally requires a practiced eye. Please bear with this thought concentrating approach for the sake of the neophytes. I learn a little more each time I pass through.
By over – zealous “pre” processing.
A. Here you see the effect of a critical angle crossover. The reflection is cloning into a refraction. The effect is continuing through B.
Close observation of this phenomenon is vital to the understanding of two sources of very common refraction noise. The cloning of a strong reflection event into the initial strong refraction is clear. What is not so obvious is the fact that all downward traveling source energy has been interrupted., creating a void that gets filled with deeper refractions that confuse targets from here on down.
Refractions are generated by abrupt changes in the reflection equation. They can be caused by faults that terminate strong events, and, in the case here, by these sudden voids in the downwave. This takes a little thought, but I am sure you are up to it. Click when ready C A
C. The first thing to observe here is how this event walks across the weaker ones. Obviously the total energy at any point is a summation of all events that independently cross paths. In any case, the dip and the alignment shows that is is a lower velocity refraction that is coming from the shallow section. Because bothersome preliminary logic (such as pre-stack migration) is not screwing up the picture, we could get at this event and lift it gently off, thus clarifying the remaining problems. I got access to this piece of pre-migration data at the end of my efforts but did not have enough data (or time) to go farther with it. Click when ready. B D
D. And this brings us to the “Ground Roll” myth. Obviously this energy is traveling horizontally (refractions), but it is not moving on the surface. Instead it is an over-lapped series of shallow refractions coming from event cross-over, the faster ones catching up with and over-riding the earlier ones. The important thing here is that we can predict and lift off these segments if we can get to them before they have been clobbered.
Now that you are used to looking at these events go back to A and B. This critical angle crossover phenomenon is all important, and it has been largely ignored by the industry – so click for next slide..
The raw data -

3. Given my analysis of the critical angle problem, again shown at the left, my best option was to use a very deep mute. To continue, click here: A B C
A is the North Sea Chalk.
B is the critical angle position, where the reflection begins cloning to a refraction.
C .is where the pre-stack migration loses control, causing an energy explosion.
The yellow line represents the deep mute I had to use to get any results.
TAKE THE TIME TO EXAMINE THE EXPLOSION OF ENERGY, NOTING THAT IT IS COMPLETELY FOREIGN TO WHAT YOU HAVE SEEN ON THE INPUT.
The trace at the far right is the stack of this set of comprised gathers, proving you can get some sort of a stack out of heavily comprised data.
Click to go on to the resolution of the processing problem.

4. I repeated this North Sea example to show the similarity in gather explosions
With one from a prospect (offshore Louisiana), where the un-garbled gathers were unfortunately not available.
In lieu of the pure gather I instead point to the gather location (a probable gas sand).
This arrow is pointing to a simulated sonic log section, (the product of my inversion and integration systems). I show it again later, with an abbreviated explanation of the non-linear logic. Next we look at a progression of individual gather sets ending at that energy explosion.

5. And here is the problem – A series of every 25th depth point gather sets (ending at the one pointed to on the last slide). The gray line conceptually separates clobbered data from usable. (I approximated this with a deep mute, which helped a lot as you will see later). There are shallower cross-overs, but for some reason they did not result in energy explosions. If we had the pure, un-clobbered gather data we should at least be able to predict and remove the artificially generated noise that resulted from the premature application of migration logic.

6. So what am I trying to say. A
So what am I trying to say? A. That another great well match proves my approach.
B. Repeat that critical angle cross-over creates refraction noise that pre-stack migration can’t handle, more serious noise is created that needs deep muting. A comparison is provided to prove that throwing out up to half of the gathers is the right thing.
C That simulation of lithology improves the stratigraphic picture.
D. That a matrix of almost invisible strike slip faults blankets the prospect. since all indications are that they trap and thus control the reservoirs, bringing out the patterns is a prime goal.

7. Let us look at the interpretation problem
Let us look at the interpretation problem. Later in the presentation I show several intersections of in-lines and cross lines. Each has been run independently, and there is a near perfect match at all of the events. This means that what we are seeing is not a processing accident, but is really there. However, since coherent noise acts just like reflections in space,it does not mean that what we see is free of the effects of strong noise forcing itself through.
I see strong evidence of deep faulting which I believe is strike slip. Tracking it runs into continuity problems that I think are related to unresolved noise. I show a couple for emphasis. Having worked other prospects where my noise removal made an amazing improvement in geological believability, I think such improvement might be possible here. I might say this type of interpretive thinking has almost become a lost art! – so click to go.
The green arrow points to the event that generated the major amplitude explosion on the last slide. An obvious conclusion is that this explosion would have triggerd a hit on an AVO search, thus pointing to a very probable reservoir by accident.
I have long argued that strong noise on most gather sets will over-ride any actual angle of incidence amplitude differences, thus minimizing the value.of such searches.

8. Continental drift is an accepted phenomenon
Continental drift is an accepted phenomenon. An obvious effect is the tearing of shallow beds to accommodate the deep plate (horizontal) movement, forming strike slip faults. This should be considered a logical fact, rather than a theory. Picking normal and reverse faults is made simple by the presence of vertical throw, and the continuation of lithologic visibility across the faults. Strike slip faults often show no vertical throw, and great horizontal movement can cause distinct changes in stratigraphy.
The heavy mixing used by pre-stack migration blurs the fault breaks, really compounding the picking problem.
Since it is clear to me that these faults are trapping, Nothing is more important than improving our ability to map them. Try not to get too picky about my attempts, keeping your eye on positive examples of their presence. This interpretation challenge cries out for the need to get at the data before it is clobbered by mis-guided processors. One of the secrets yet to be uncovered is whether the faults always continue with depth, or whether the movement is shared between many breaks.
As usual I add a plea for help in getting that kind of data to go on with my studies. I will add more fault picks from time to time.
The green arrow points to the event that generated the major amplitude explosion on the last slide. An obvious conclusion is that this explosion would have triggerd a hit on an AVO search, thus pointing to a very probable reservoir by accident.
I have long argued that strong noise on most gather sets will over-ride any actual angle of incidence amplitude differences, thus minimizing the value.of such searches.

9. Direct reservoir spotting (with no well control) -
All agree that the presence of hydrocarbon affects bed velocity. In turn, there is no doubt that reflections are caused by vertical velocity changes.
Amplitude peaks are, therefore of prime interest.
Trouble is They’re meaningless on raw seismic sections, since they relate to individual contacts. rather than the body of the bed.
Before explaining how we handle this problem I point to one of the proven reservoirs in this project (yellow oval) as an example of what we could have pointed out before drilling began. Obviously the phase amplitudes had to be put into perspective to make this picture meaningful.
While the resolution seems good I am convinced there is still a lot of unresolved noise here, and feel the results could be much better.
The green arrow points to the event that generated the major amplitude explosion on the last slide. An obvious conclusion is that this explosion would have triggerd a hit on an AVO search, thus pointing to a very probable reservoir by accident.
I have long argued that strong noise on most gather sets will over-ride any actual angle of incidence amplitude differences, thus minimizing the value.of such searches.

10. Non-linear inversion –
This simulated sonic log section is the integration of the sliding convolution of the interface solutions and the evolving wavelet-thape. (Rocky Mountain data). And the integration –
The two sections have been carefully lined up, and the sonic log overlay has been accurately placed on both. The input stack contains events from the tops and the bottoms of all the beds.
The system has resolved this complex to provide the lithology simulation. The amazing accuracy of the match on the results provides a strong logical proof of the inversion itself, along with the obvious verification of the integration. Produce the simulation.

11. So on this slide we go back to the full stack, explosion and all, with no deep mute.
Please toggle, (using the right arrow) paying attention to what happens at A. A

12. And this is my final product, using the deep mute.
You might find this comparison unbelievable but let me assure you that I have routinely run such before and after tests with similar results. Please toggle here until satisfied, (using the left and right arrows on your keyboard). A

13. One more comparison, this time adding the optimized stack (that uses a deep mute that excludes the artificial explosion traces).
Before going on, review the integration logic responsible for removing over 1/3rd of the events on the input stack. Then focus on point A (shown on this preview of the final inverted * integrated result at the right).
Final result
Final ADAPS result A

14. Unadulterated stack A
Notice that the worst distortion on this untouched input stack appears close to the origin of that “critical angle cross-over” (that generated these energy explosions).

15. Optimized stack input
Here is the intermediate (optimized) stack that provides the input to the inversion logic. A deep mute was used to throw out the artificial explosion of energy caused by the migration logic’s attempt to pull all the wayward energy in. A

16. Final result
Final ADAPS result
Please toggle back and forth over the previous slides to see the re-alignment of the events. Remember that the process is proven by the well matches. A

17. These last five slides show 6 pairs of in-line and cross line pairings
All 12 sections were done independently, with identical driving parameters. The files were spliced later, so seismic mixing was impossible.
The remarkable pin point matches at all joinings constitutes a powerful QC of the integration logic.
The differences that do exist are probably caused by remaining coherent noise.
In-line 1526 and cross-line 2900

18. In-line 1226 connected to cross-line 2500

19. In-line 1386 connected to cross-line 2700

20. In-line 1426 and cross-line 2900

21. In-line 1506 and cross line 2400
In-line 1386 and cross-line 2300

22. Click to start over.
Click for ADAPS base -