Seismic noise is omnipresent The next slide goes back to earlier work where the shear problem became apparent. It curls around our reflection events, Enhancing them when it is in phase, canceling them when it is perfectly out of phase, or just showing up when there is no overlap. And it does this is such a subtle way the average user does not even know it exists, making it necessary to specifically look for it by moveout pattern. When it has a moveout pattern we call it coherent, and this is the kind that gets in our way. Patterns are either reflection (spherical) or refraction (linear). The system detects them by iterating through a range of total moveouts, stacking and testing each for maximum amplitude and statistical quality. In the spherical case, knowledge of approximate reflection velocity provides selection bounds. This not only applies to inter-bed multiples but to the more serious vertically traveling shear waves described in the gulf coast.noise removal topic. Refraction noise is the killer in most areas. Primary refractions are caused when the down wave hits the critical reflection angle. Downward movements ceases and all energy travels horizontally. This causes a void in the down wave, and since nature abhors a vacuum, secondary refractions occur from this truncation on all deeper beds.. This is a difficult thing to picture, but it certainly exists.

Raw shot record from original (Scottish coal seam) processing report. The red arrow points to the central noise cone problem I discuss below. Deep hole dynamite Vibroseis Ground roll (Raleigh) or vertical (shear) are the two possible sources of this heavy energy cone. In general Industry has taken the easy way out by assuming horizontal ground roll. If this were true the problem could be easily solved by muting. My work has proven to me that vertically traveling shear waves comprise the bulk of the noise we see here. We can still get rid of the really strong stuff by muting, but the problem is that this energy spawns deeper refractions that move into the center of the spread. This was very evident in my Gulf Coast de-noising exercise. It is an omni-present problem on land work. I think of it as coherent noise type 1. Inter-bed multiples are the second noise type. Their removal can be accomplished by first establishing expected velocity curves, then iteratively scanning for events that have reflection curvature, but whose apparent velocities exceed an acceptable range. Point source refractions are the third and most insidious type of coherent noise. They are described in the following series of slides (showing removal examples). All three types should be expected.

Deep Refractions from vertical travel cutoffs A ABC To the left you see the gathers I got to work with in my North Sea effort. A is the North Sea Chalk. B is the critical angle position, where the reflection begins cloning to a refraction. Here vertical travel begins to be cut off. C.is where no more energy gets through The yellow line represents the deep mute I had to use to get any results. Point source reflection theory has always been a mystery to me, but it is one thing I have to buy into. In essence it says that when the reflection process hits an abrupt end to either a reflecting interface or s break in the down wave, a horizontal refraction takes place, thus proving nature abhors a vacuum. This means every deeper event will be accompanied by fairly strong coherent noise that penetrates into the spread. This pre-stack migration was not able to handle refraction patterns. The exotic explosion of energy we see at the upper right is the result. When I was on my way out I was able to obtain some raw data that had no migration (or NMO). The set at the lower left is from the same zone. The A,B and C markers apply in the same way, and you can see there is no abrupt change in energy, proving the use of pre- stack migration to be at fault. Since that process mixes heavily in its attempt to draw in energy that has wandered due to strong dip, it has to really murder the fine detail on the fault breaks. As a result of this phenomenon AVO on this data is virtually a joke. The energy explosion would have triggered the searcher, but this would not have had anything to do with the theory. Yet the major discussion theme when I came on the scene was whether to use inside, middle or outside trace combos. On the next slide I repeat what you see at the left so you can see how infinitely detailed the noise patterns are before being massacred by the migration. These refractions sometimes stack quite nicely, not looking like noise. Once we know to expect them, we can search them out by their offset patterns. Once we can detect them, lifting them off becomes simple. Improved results supply logical proof they were there, A B C Are prevalent on both onshore and offshore work – and few processors take this serious noise into account.

Because the energy is so much stronger at the top of the section, that is where our system starts out. Remember you are looking at a refraction that has been identified. Note the “linear slope” annotation pointed to by the red arrow. In this case it is 2208 samples (or 4416 milliseconds over the total spread length. There is no way events can be seen in the over-lapped jumble until the system has lined them up for you. The fact that you can see the refractions linearly over such a long interval is the proof one needs that we are not mistakenly removing reflections. When it is satisfied that it is safe, it predicts and gently removes the offending energy, trace by trace.

And here is a deeper refraction that has been lined up at a slope of 3256 samples per spread total. It starts up in the shear wave high energy zone, but then works its way well into the spread. Again, all that was done here was to shift each gather traces up by the calculated linear offset.

In the original show I present this set of cutouts and try to explain that each shows a refraction event that was lifted off. To see the events, each gather trace had to be shifted upwards by the amount computed from the linear refraction slope illustrated on the third slide. These slopes are very extreme (ranging upwards to a couple thousand samples per spread). As I said, without correction you wouldn’t see the events at all, yet here they show their intractable truth. I am sure this explains why nobody has done this type of removal before. In any case I cut them out to save space. I just chose these few (out of the hundreds the system lifted off) to make the following crucial points: 1. No shaping after the fact was used, so, as I say above, the mere presence of these obvious events, way into the spread, is firm logical proof that they are coherent noise and that they will affect the stack. The fact that my logic does not see them on it’s next iteration is internal verification that they have all been effectively removed. This passes the programmer honesty test. 2.The last 5 start from within the band of noise the industry commonly attributes to ground roll. As the system removes event after event, this strong inside band gets eventually (almost) eliminated. Again this proves this overpowering energy is traveling vertically through the section (shear wave) rather than along the surface (ground roll). So this is what my logic removes from the gathers. I mentioned the nature of the task was such that all the shot points affecting a gather had to be treated before final answers could be prepared (and that is a bunch). Watching the system removing tons of energy from the data made me wonder if anything would be left. To find out we now jump into those final results and see the answer is a strong yes! Just keep in mind that the detected noise was generally a lot stronger than the signal, even way into the spread.

Allow them time to load! A PowerPoint compendium of seismic topics Basic reflection theory the experts seemed to have missed. Inversion and integration – the all important sonic log simulation. Coherent noise background – an explanation of types. Removal of coherent noise on Gulf Coast – a breakthrough. Vibroseis de-noising – another (but similar) breakthrough. Strike slip faulting – salt dome association – new thinking. North Sea strike slip interpretation – the importance of resolution. About Paige - MS in geology,spent 7 years in Venezuela for Mobil,& then Phillips Maracaibo interpretation found Phillips’ major field there. Back to states, joined Phillips computing, became project manager for exploration. Hired by Western Geo. To start digital operations in Shreveport. Wrote first predictive deconvolution program that put Western on the map in digital processing (and formed the non-linear basis for later ADAPS software), After brief sojourn in commercial processing (where he wrote a table driven programming system), joined Dresser Olympic as both manager of processing and of research. Went on his own to start non-linear development. Consulting package consists of Paige’s personal time, his open-ended software and use of his processing hardware. Unless full segy detail is requested (segy output), the product is a series of PowerPoint studies. He can be reached at