IDEAS
- I
spent the last few years perfecting what I think is a great
inversion tool, only to find that most land and shallow water input is so distorted
by coherent noise that
inversion results are meaningless.
WHILE I'M PREJUDICED
in favor of my own
non-linear techniques, this series is really a way to communicate
hard learned knowledge of seismic problems. In a sense I feel I have
to explain what I know in order to justify my approach. Hopefully you will
learn some new concepts if you stay with me. If you find you already know
it all, congratulations!
POWERPOINT SHOWS - Several
basic topics are in HTML form. However, because I consider comparing
before and after examples crucial to process selection, much of the deeper
thoughts are covered in several PowerPoint shows (where toggling is easy
to do, especially in full screen). Most serious seismic people already use the tool, but if you don't
have it you can get it free by Googling PowerPoint.
SEISMIC TUNING IS AN
EXAMPLE of such a
complex subject that benefits from toggling. The basic understanding of
what makes up a typical seismic section should be the first requirement of
interpretation. If you don't have it you may be doomed to error. You will
find dozens of illustrative examples inside the various shows.
THE SHOWS ARE BIG
and slow to load (especially if you don't have fast broad band). Be
patient. While the first load may take several minutes, once you have the file on your disc
it is much faster. Also, make sure you go to full screen on each
one.
On the left panel
is a series of topics. Passing your mouse over one will bring it up. But be careful
with the mouse. If you accidentally hit another topic, you will unexpectedly
go there, so get it to the right immediately.
A summary of what
you'll find when you go on - Noise distortion
will garble structure & stratigraphy.
Where present
it makes such things as AVO analysis a virtual joke. Wavelet work is hit
very hard {and it is the essence of inversion). Since the offending energy is
often coherent, 3D stacking can mistakenly focus on it. Migration is even more dangerous, since it seeks patterns
spatially, assuming all to be true reflections. These themes are developed inside,
as well as in the newer set of PowerPoint shows.
Even if you are a
seasoned interpreter
your
insight into seismic tuning will improve when you really look at what
inversion does. The PowerPoint shows make it easy to
instantly compare the "before" with the "after"
showing you how real events are covered up by long wavelet tails, and
how removing these tails changes the interpretation picture.
But be sure to go to full
screen on all Power Points!!
If you are
interested in my services (and/or want to further this research),
let me know at dpaige1@sbcglobal.net
OVER-AUTOMATION -
Years
ago ,when I was project director for exploration out of Phillips'
computing department, we bought a mapping program from a third party. This
was early on, and the program required a balanced grid of input for its
curve fitting. I found that a lot of time had to be spent keeping the
geologists from believing the enormous structures out on the data fringes.
I see much the same thing happening
today, with the exotic multidimensional graphics logic that people love to
use. Few ultimate users of these "surface" oriented programs
have any idea of the quality of the input data. In fact they get little
chance to see it. Once again my theme picture applies.
While current interpretation software is
powerful, I feel the subtle
stratigraphic detail you will see in two of my PowerPoint shows will
suffer.
These pictures are several
years old,
and some improvements
have been made in the automatic picking. However, at the time they were pushed as wonderful
answers, and I do believe they are fair examples of today's data quality.
The blue surface blows my mind. If one looks at the sections on
the side one sees lousy data. Without even knowing I can tell you this is
either land or shallow water. Remember
both pictures when I get to the comments on coherent noise.
In any case, this type of presentation is probably all the managers saw
(again prompting the Ostrich with head in sand).
Another example of questionable
results the vendor seemed proud of.
The
automatic picking wanders enough for significant mapping error, and the
red fault pick is far off (see the reflection from the fault face to the
left). However the main problem is data quality. What we see here I
believe to be a complex
mix of lots of coherent noise and possibly a
little real reflection data. Please contrast these two pictures with the
example below.
This is what data
should look like.
Not surprisingly, this is deep water stuff. I say this because
all deep water shooting I have seen is much better than land and shallow
water.
There has to be a reason for
this predictable difference in quality.
And this reason must be associated with
the way the energy passes into the subsurface which, in turn affects the
generation of coherent noise. While I don't pretend to
know all the answers, I believe I do understand some things that the
industry has ignored. I will put out my ideas
later in the noise discussion,
Relax
- There is nothing
I can do to speed up the loading of these long Power Point shows.
Hopefully you will find them worth the wait. Once in your system they go
fast, so coming back the 2nd time makes sense.
So start your selection then
come back here to read the comments at the right.
Upon
entering the show manager,
immediately right click and choose full screen.
Then, click on specific
shows.
When you are finished with
a show keep clicking
until you get back to the flow manager.
When you want to reenter
this flow, right
click, choose "end show" then kill the
PowerPoint at the upper right.
