Mission-Oriented Seismic Research Program

http://mosrp.uh.edu/news/m-osrp-update-for-your-possible-interest-resending-a-2013-note-to-the-m-osrp-advisory-board

M-OSRP | M-OSRP update: For your possible interest, resending a 2013 note to the M-OSRP Advisory Board University of Houston College of Natural Sciences and Mathematics

https://www.linkedin.com/posts/arthurweglein_youtube-activity-7136085406168616960-hndN/?utm_source=share&utm_medium=member_desktop

Arthur Weglein on LinkedIn: YouTube Dear Friends, I hope that this note finds you well. There are two presentations in the links below. The longer video presentation is titled” A Perspective on…

https://m.facebook.com/story.php?story_fbid=2350596735209811&id=100007785225589

M-OSRP has pioneered, developed, and delivered two new and game changing imaging and inversion methods. The first is a method that requires a velocity model that we call Stolt Claerbout III migration for heterogeneous and discontinuous media. It has significant added value compared to all current migration methods including all forms of RTM. Those advantages include structural and amplitude fidelity and resolution of the target and reservoir, and the automatic accommodation of planar, curved and pinch-out reflectors, for structure and amplitude analysis. The new SCIII migration method is the only migration method that makes no high frequency approximation- thereby processing the recorded data as a recorded wave field and preserving its ubiquitous wave nature in the subsurface. Since waves are ubiquitous they do not suffer illumination issues.
In contrast, e.g., RTM and Kirchhoff migration both make high frequency ray theory like appropriations, that squeeze the recorded wave field back into the subsurface along rays. Hence, processing methods ( like RTM) that make high frequency approximations, are a major contributor to illumination issues. Again, the intrinsic high frequency assumptions, within both the imaging principle and propagation model within all RTM methods, discount the potential within the data, and that choice of processing method, contributes to illumination issues. The papers in the links below examine, analyze and document these statements and conclusions. http://www.mosrp.uh.edu/people/faculty/arthur-weglein
The second imaging method pioneered and delivered by M-OSRP is the ISS direct depth imaging, without a velocity model known , estimated or determined with viability demonstrated on complex synthetic Marmousi data and on field data-below I am sharing a video and a few comments on this topic.
https://youtu.be/Ukquiv5KM7A
ISS depth imaging, the evolution and field data example, demonstrates viability of the direct ISS imaging method without knowing, estimating or determining the velocity model-it’s the only method with that set of properties and a truly game changing idea and method. It derives from precisely the same logic and mathematical physics that derived the inverse scattering series (ISS) methods for free surface multiple elimination and for internal multiple attenuation and elimination. The ISS depth imaging algorithm inputs your data and communicates ( at every step) where and when the depth image is correct, (it knows when the depth and structure are correct-without drilling a well ) or when additional higher order terms in the ISS depth imaging series will be required- a remarkable awesome property and intelligence-the intelligence is because the method is direct, ( like every term in the solution to the quadratic equation, minus b plus or minus…every part of that solution has a role to play , a specific and defined role and objective… and that property is due to the direct nature of the quadratic solution… if you were doing a model matching search for the roots of a quadratic equation you would be in the world of FWI) and analogous to the solution of the quadratic equation, every term in every isolated task sub-series of the ISS is part of a direct solution for that task, and every term and part of a term has a well-defined purpose and objective, towards the goal of the specific task of the isolated task sub-series) there is nothing AI or ML here- and no searching , hoping or model matching-it’s a new direct math physics, and seismic physics and for the task of migration, locating structure , accurately, a direct imaging algorithm, where every term and piece of a term has a distinct and well defined purpose, its own intelligence - and the algorithm informs you on your data where and when its purpose is called for- the algorithm itself communicates on your particular synthetic or field data where its purpose is required, and when and where within your current image, all is well and no further help is needed and where ( at every given stage ) more terms within the ISS depth imaging subseries will be necessary- and then once again with the additional terms included in the imaging series algorithm, where it is now adequate, and no additional terms are called for- and the current location and depth has been achieved and its job is done. When higher order ISS imaging terms are called for, on some part of your image, at some step, those additional terms will not impact or perturb the places where the structure is already correct. The algorithm informs you when and where its purpose is achieved.
And the velocity model is not known, estimated or ever determined.
We didn’t create that intelligence, that intelligence resided inside these task specific sub-series of the inverse scattering series before we understood that they must be there and searched and located them. Another mind-bending thought is how the Q compensating sub-series of the ISS can compensate for the effects of absorption and dispersion without knowing, estimating or ever determining the absorptive and dispersive mechanism. I understand exactly how it works and does that, but it still is amazing and awesome and mind boggling, to me.
In summary: We clearly need more capable imaging ideas and algorithms, and that won’t come from some not new RTM and FWI challenged thinking, but a really new and fundamentally more capable and effective imaging concept and method - we respectfully suggest you might consider these two concepts, methods, and algorithms, mentioned here. When considering methods arrived at without any thought or intelligence, for example, model matching methods , where you can model match anything- if one might be asked why don’t we just model match internal multiples, for example in FWI -or why model match primaries and free surface multiples leaving out internal multiples, the best and honest answer is ‘ why not?’ There is no theory or derivation behind model matching anything- and what data to model match-no theory whatsoever- that’s why it’s so popular and accessible, because it’s easy to understand- and it’s easy to understand because there is nothing to understand. In fact, you can explain FWI in an elevator ride. Take an actual trace and a trace from a model, and move the model parameters around trying to have the two traces match. And the elevator door opens. And then you mention’ by the way , there is a direct method to determine those properties, without moving properties around. You start with an operator identity , a relationship between the change in medium properties and the corresponding change in a wave field, called the Lippmann Schwinger equation. That LS equation produces the forward scattering series and then the inverse scattering series. The forward series is quite unremarkable and requires all details of the model to generate the wave field. On the other hand the inverse scattering series is absolutely remarkable, and communicates that all processing objectives are achievable directly in terms of the recorded data, and with absolutely no subsurface information whatsoever. You offer:’ Are you interested in seeing how that LS equation and the ISS derive and work? ‘ . Your colleague hesitates and then responds’ Wait a minute, what was that thing you were talking about in the elevator? ‘
To have an effective seismic processing strategy, we need direct methods for the part of our data that can be explained by our assumed physics and earth model types, and indirect methods for a part of our data that is beyond our assumed physics and earth model types. Having these two parts working in cooperation is a worthwhile goal. The problem is the current out of balance, with indirect model matching and ML and AI , totally ignoring the physics and real intelligence behind direct methods- with all next steps calling for buying more computers. Not only do the model matching methods require no thought but the next steps require no thinking as well. The total absence of intelligence and thinking- in terms of what you are doing, ( and overselling) but in removing any concerns about what needs to be done next. That’s the exact opposite of the ‘final and ultimate’ seismic processing method. Methods are not the problem; methods do not have egos and ambition, and they don’t overstate their capabilities and avoid their assumptions and shortcomings. That language ‘ having an ultimate and final solution’ has no place in any scientific endeavor and certainly not in seismic physics. The earth is more complicated and complex than our assumed physics, now and for the foreseeable future. While we can support and encourage (and celebrate) seismic processing progress, there is no final and ultimate seismic processing method, and there never will be.

Petrobras Workshop on Game Changing Seismic Technology - Aug. 2016 M-OSRP invited presentation at Petrobras Workshop on Game Changing Seismic Technology - Aug. 2016

http://www.mosrp.uh.edu/people/faculty/arthur-weglein

M-OSRP | Dr. Arthur Weglein Dr. Arthur Weglein-Director, Mission-Oriented Seismic Research Program. Dr. Weglein is Hugh Roy and Lillie Cranz Cullen Distinguished University Professor in Physics.Dr. Arthur B. Weglein is Professor in Department of Physics & Department of Earth and Atmospheric Sciences.

http://www.mosrp.uh.edu/people/faculty/arthur- weglein

M-OSRP | Error University of Houston College of Natural Sciences and Mathematics