Transcript There is no mixing of primaries in the subsurface.

Think stratigraphy,
This picture, all by itself, not only proves strike slip
theory, but also verifies the logic used to bring out the fault detail. This is North Sea data
and to my knowledge, the operator still has no idea this structural relationship exists.
Stratigraphy to the left of the main green
fault fits together, as does that to the
right. However there’s a drastic difference
between the two.
The obvious fact is we are seeing two
locations (originally far enough apart to
make such a big difference geologically
reasonable).
Take your time - but remember: if true here it will be true all over.
The justification for doing “before stack” inversion & integration is
the drastic down-wave shape differences between stations.
But before going on. let’s examine what the data
shows us.:
On the globe, we see the South American coastline
duplicated on the African side, with the Atlantic ocean
in between.
We might explain what we see here two ways – either
saying the right hand block was pulled to the left, or
that the left hand block was pulled to the right. While
plate theorists can play with this, it is immaterial to
our thesis. For us it is enough to say that lateral shift
is here, and it can be huge. If you study the exquisite
detail here I would hope you will agree that this slide
proves the strike slip thesis.
Back to justification for “before stack” inversion and integration.
Primary reflections come from individual interfaces (like bed tops and bottoms). Because of the
two way depth point geometry there are significant differences in the traveled distance between
individual gather traces, Earth filtering (creation of successive side lobes) is tied to distance
traveled. Events do not mix until they reach the receiver, where all overlapping primaries are
composited. In other words, the phase damage is done before anything is recorded, Thus,
contrary to original thinking, the normal stack is adding a fairly wide variety of event shapes, and
this requires inversion resolution before the distorting addition occurs.
Basic reflection theory that we should all keep in mind.
1. The seismic energy continuum consists of thousands of independent primary reflections, each coming
from it’s own reflecting interface.
2. There is no mixing of primaries in the subsurface.
3. Geophones can only record the energy that exists at instants of time, and this forces an accidental form of
compositing at the recording point.
4. The earth filter generates trailing lobes with travel, gradually emphasizing lower frequencies. The total
travel difference from inner to outer station is great and by the time the energy reaches the recording
points there will be significant differences in primary wave shapes. This (mostly ignored) problem is
exacerbated by excessively long spreads. It heavily affects gather trace character, probably dwarfing any
possible AVO effect.
5. The separation between primaries changes with offset, modifying the way they combine at the recording
point, again affecting gather trace character. As with the earth filter, it happens before any processing can
be done. In summary, the stack will produce an almost accidental waveform mixture.
The above 5 points are demonstratable facts, not just my opinion.
It appears to me
that they have largely been ignored by current frequency domain developers. Their effects contribute to
the difficulty of determining a usable average waveform (the weakest link in their logic). Of course my own
system faces the same set of problems, but using advanced pattern recognition to establish reflection
coefficient spikes allows it to get the best answers possible. So my after stack inversion provides a safer
and arguably better route.
However it was clear that inversion should ideally be done before stack. So
that is what I did, and this show is mostly about comparing the before and after results.
So we start with these comparisons. The arrows have been located at specific times to help you keep the
correlation straight. Because my personal goal was to improve the resolution enough to facilitate picking
the strike slip faults I knew were there (in this North Sea data), I have added a final section for that subject.
Normal inversion, no noise removal – arrows show fxd times.
To understand the processing, it is
Almost necessary to toggle between
The indicated slides.The purpose of.
the arrows is to tie locations for the
Sake of comparison.
Toggle
Noise removed, before stack inversion – (note positions).
The seismic resolution has been greatly
Improved by focusing the inversion and
integration on the offset. Notice how the
bed thicknesses have been established.
Toggle
Another line – normal inversion with noise removed
Before stack inversion, noise removed. Note thickness.
Toggle
Before stack inversion, noise removed – Pick section.
The area if interest
Expanded Excerpt - before stack inversion with noise removal.
Use this slide to toggle with the faulted version, to check
evidence. But first remember that strike slip faults may not
exhibit vertical throw, so abrupt changes in character may be
all we have to go on.
And here we are back at the
final composite of the “before
stack” results.
Because it is so important (and generally ignored) I repeat the down-wave dialogue.
Primary reflections come from individual interfaces (like bed tops and bottoms). Because of the two way depth point
geometry there are significant differences in the traveled distance between individual gather traces, Earth filtering
(creation of successive side lobes) is tied to distance traveled. Events do not mix until they reach the receiver, where
all overlapping primaries are composited. In other words, the phase damage is done before anything is recorded,
Thus, contrary to original thinking, the normal stack is adding a fairly wide variety of event shapes, and this requires
inversion resolution before the distorting addition occurs.
Proof of the importance of wavelet shape focus –
Improving the attribute resolution by taking spread geometry into consideration has so increased seismic
resolution that we find inversions of individual channel data almost beating previous full “after stack” results.
On the left, the non-linear inversion logic was applied to the matrix of all stacked traces. On the right it was applied to
the set of gather traces that were collected on single channel E.
Because noise removal is central to my later work, I go on with a discussion on
how intertwined coherent noise creates a random effect that confuses all frequency
domain calculations and generally screws up the works – Pay close attention here,
since this is a vital point.
To the left we have a raw gather and to the right the same data with noise lifted.
The fact is such noise is present on every prospect, to some degree. The problem
is seeing it! To do so requires intense pattern searching on the gathers, and this is not normally done. When we
add the factors I outline next, such things as AVO claims and frequency domain waveform generation become
questionable, and the need for non-linear approaches becomes apparent.
For the reasons listed below the final stack is adding different waveform shapes. Thus
the need to invert before this distortion happens.
1. The seismic energy continuum consists of thousands of independent primary reflections, each coming
from a single reflecting interface.
2. There is no mixing of primaries in the subsurface.
3. Geophones can only record the energy that exists at instants of time, and this forces an accidental form of
compositing at the recording point. In essence this is a preliminary stack, and we have no control over it.
4. The earth filter generates trailing lobes with travel, gradually emphasizing lower frequencies. The total
travel difference from inner to outer station is great and by the time the energy reaches the recording
points there will be significant differences in primary wave shapes. This (mostly ignored) problem is
exacerbated by excessively long spreads. It heavily affects gather trace character, probably dwarfing any
possible AVO effect.
5. The separation between primaries changes with offset, modifying the way they combine at the recording
point, again affecting gather trace character. As with the earth filter, it happens before any processing can
be done. In summary, the final stack is an almost accidental waveform mixture.
Some thoughts on frequency oriented inversions - The frequency domains were essentially
invented mathematically to make solutions by equations possible. The new tool was the transform, which
models what happens in the time domain into this new form. This conversion enabled the designer to invoke
equations to generate filters that change the spectrum of the data to equal a desired one. This was as far as
their early deconvolutions went. Of course these efforts were attempts at inversion. They just had to be limited
to keep the processes stable.
The problem with going farther is that any particular spectrum can represent a variety of wave shapes. Later
phase work has concentrated on determining that shape. This is the weak link in their process. When noise is
present it gets much harder. My non-linear approach by-passes this problem by determining spike location via
pattern recognition. The great well log matches I have shown pretty well prove the validity of this approach.
I continue with a graphic that tries to put the basic seismic problem into perspective.
1 The geology
3. The down wave
2. The reflection coefficients
(spikes in non-linear lingo).
shale
Lime
Sand
shale
The argument for
non-linear methods.
Computing reflection coefficient
spikes via statistical optimization
eliminates frequency and phase
from the picture.
4. Its direction
The crazy down-wave at the left is just
there to show the nodular character that
evolves with depth. The “shape” of the
primary reflections at each offset, as well
as the offsets between them, will depend
on the total distance traveled. They are
stacked at the receiver location, and the
resulting trace character will vary greatly
between offsets.
Rigid, mathematical solutions of complex
problems like this are extremely tough.
Optimization is typically the answer when
coming up with the best answer possible
is what we want.
Because linear inversions are not able to compute reflection coefficients without knowing the wave shape,
the industry has become obsessed with that problem, with opinions coming from everywhere.
I use the oval to emphasize that this is what inversion should be doing. Once we have effectively computed
the reflection coefficient spikes we have raised seismic to the well log level. This is where my non-linear
inversion takes us, avoiding the exact wave shape hurdle by calculating spike position via pattern recognition.
We’re looking at completely different approaches, and the way statistical optimization can handle error is the
key to being able to get answers under difficult circumstances. Again, the proof lies in the well log match.
Processes should be judged by results - One could spend hours studying what there is to learn
from this display of a “simulated sonic log” section. First, notice how bed thicknesses make the “stratigraphic
differences” across the strike slip faults very evident. To me, as an interpreter, this is vital.
When I re-look at this picture I shake my head and wonder why almost no interest has been shown. Nobody had
noticed them before I made this run and I guess they will continue to be ignored.
Strike slip faults are a result of shallower beds being torn
apart as a result of deep plate movement (continental
drift). They are a given fact of geological life. The reason
they have been missed by the industry is that they often
are very hard to see. Where the stratigraphy is regionally
regular, no vertical throw may result, making patterns
very hard to establish. When one can’t see a pattern
between the bursts of energy, it is normal to assume we
are looking at noise.
Each fault block in this pattern exhibits stratigraphic
consistency within itself. Before the pattern was set it
looked like an unconnected hodgepodge. Much of my
later efforts to improve resolution were prompted by the
need to see the fault patterns I was sure existed. My
time domain inversion and subsequent sonic log
simulation are vital parts of this improvement, and that Is
why I place this subject first on the list.
Next you might observe that the changes in amplitude
now seem to make structural trapping sense. This ties
into my claim that direct reservoir spotting is now a good
possibility!
First time In skim to see all subjects covered. For more
on strike slip faults, come back & click on oval -
Raw seismic sections are “coincidental” mixtures of primary reflections.
The amplitudes and polarities of stacked events depends on how the primary reflections were aligned when
they reached the recorder.
These alignments depend on effective velocities (a function of distance from source to receiver). Earth filtering
continually produces trailing side lobes with distance traveled. Because of large differences in recording travel
time, these character changes can be significant. Of course they occur before any processing is possible.
In this mix, the strong will prevail. The amplitude of each primary reflection is a function of both the velocity of
the previous layer, and that of the current one. Thus the lower interface between a sand and a limestone will
generate a strong event,making the sand look like a lime, where that between a sand and a shale will be
weaker. Of course the same is true on the upper interface. The point here is that trusting the amplitude of any
stacked event before inverted data is integrated can be specious, at best. In my own development work I was
continually surprised (and pleased) at how the integrated results matched with available well logs. Not perfect,
but certainly a big improvement.
The simulated sonic log section at
the left is a good example of where
integration after inversion made
good sense out of amplitudes. The
well match on this strong red event
matched beautifully, to the point we
could have predicted success if the
run had been made before drilling.
Unfortunately, due to me being
isolated from the Ikon client, no one
ever saw these visual results.
For more on simulation, click oval -
Stack (input)
Inverted & integrated
This simulated sonic log looks too
good to be true (but it is, and all
the other ones we have matched
at least come close to this quality).
Once again, we inverted down to
the reflection coefficient (spike),
then integrated the results.
While it looks great to me, many
seem to have trouble adjusting to
the squarish nature, not realizing
that was the goal all along. Those
thicknesses are crucial to long
range seismic correlations, and to
detailed stratigraphy.
The curves are not sinusoids, and
frequency analysis does not apply
in any sense. What they really are
is truth and beauty, thank you.
If you need to see more, click on
the oval -
Now let’s talk about noise removal –
To begin, it is vitally important that we lift it off gently,
so as not to disturb the underlying signal we are trying to bring out. Obviously frequency sensitive
filtering is a no no. Predicting the individual noise events and computing correlation coefficients is
the key. The example comparison below proves that can be done. For more, click on oval -
I refer to the fact that all sorts of academic assumptions have been made that ignore the effect of noise
inter-twining with signal.
The next series of pictures introduces several shows I believe are important. After moving through the
possibilities, come back and click on the image that interests you, giving the PowerPoint plenty of time
to load. Since you might have trouble getting back here, it would not hurt to bookmark this file.
This show is my latest attempt to explain my work. It uses a set
of data that was giving the geophysicist problems. Click on the
oval here for the PowerPoint -
Click on oval for direct reservoir detection -
This south Louisiana work is perhaps the one I am most proud of. It really was the beginning of my serious noise
removal efforts. The flank events butting up against the salt dome did not even show up before the noise was
removed. My advanced scanning for noise events was developed here, I was able to track strike slip faults on the
both ends of the dome, virtually
proving that they contributed to
the actual formation.
I started here with data in the
shot point format, with no NMO
applied. This allowed a more
precise and logical scanning to
detect non reflection NMOs.
The system removed so many
refraction events (stemming
from the central noise cone)
that I wondered if there could
be any energy left. The results
showed that there was.
Click on oval to enter this one.
There are more things to
talk about but I will leave it
here for now. Click on oval
below to go to router.