Posted on March 23, 2018
You are warmly and most cordially invited to attend the 2018 M-OSRP Annual Technical Review and Meeting on Tuesday and Wednesday June 5 -6 , 2018. The meeting will be held in Room 102 of the HSC building at UH.
For your possible interest, a discussion of how we choose the prioritized challenges we address, and a few of the topics to be discussed and presented in the technical review meeting are found in the e mail below.
The meeting will be video recorded with synced slides and available several weeks after the meeting. Further details and a provisional agenda will be sent in mid-April. We look forward to staying in touch and greeting you at our Annual Technical Review June 5-6.
A major activity within M-OSRP has been and remains the development and delivery of fundamentally new and more effective methods for removing free surface and internal multiples, for offshore and on-shore plays, without damaging proximal or interfering events. That is., removing multiples that interfere with target or reservoir identifying primaries, without damaging the primaries. More effective multiple removal remains an active and priority seismic research topic , with open issues to address, and where advances and the next generation of deliverables will have a further significant positive impact on drilling success rates for locating and developing reservoirs.
We recognize that there is considerable attention and communication these days on ‘using multiples’. In the note below and in the executive summary video http://mosrp.uh.edu/news/executive-summary-progress-2017 we present a new perspective on the removal and using multiples discussion. That perspective is behind both the near-term and longer -term strategy and resource allocation decision that M-OSRP is making going forward.
The logic is as follows:
To start : (1) to image recorded primaries, with a smooth velocity model, recorded multiples need first to be removed . If not removed, each multiple will produce a false, misleading and injurious image, and secondly (2) for unrecorded primaries: to use a recorded multiple and a recorded subevent of the multiple to find an approximate image of an unrecorded primary subevent of the recorded multiple, with a smooth velocity model, any unrecorded multiple that is a subevent of the recorded multiple must once again be removed.
Hence, to image recorded primaries recorded multiples must first be removed, and to find an approximate image of an unrecorded primary requires unrecorded multiples to be removed. The very use of multiples speaks to the primacy of primaries. Multiples are only useful if it contains as a subevent an unrecorded primary. A multiple that has all its subevents recorded has absolutely no use or value. All primaries are useful- and there is no substitute for a complete set of recorded primaries. Multiples can at times be useful but are not in any sense the ’new primary’.
The recorded multiple event that can be used ( at times) to find an approximate image of an unrecorded primary, must as an event be removed in order to image recorded primaries. It’s remove and use multiples , it’s primaries, both recorded and unrecorded primaries that we seek and require, and removing and using multiples are not adversarial , they serve the same single purpose and objective.
For your possible interest, in the Executive Summary Presentation in the link above there is a new perspective on removing and using multiples.
Basically : (1) to image recorded primaries, with a smooth velocity model, recorded multiples must be removed and (2) for unrecorded primaries, to use a recorded multiple and a recorded primary subevent of the multiple to find an approximate image of an unrecorded primary subevent of the recorded multiple, any unrecorded multiple that is a subevent of the recorded multiple must be removed.
Hence, to image recorded primaries recorded multiples must be removed, and to find an approximate image of an unrecorded primary requires unrecorded multiples to be removed.
The recorded multiple event that can be used ( at times) to find an approximate image of an unrecorded primary must as an event be removed in order to image recorded primaries.
The key point is that it’s primaries, both recorded and unrecorded primaries that we seek and require, and removing and using multiples are not adversarial , they serve the same single purpose and objective: the imaging of primaries.
Multiples (recorded and unrecorded) need to be removed in order to image primaries( recorded and unrecorded, respectively).
What use is a multiple where all primary sub-events of the multiple have been recorded. The answer : absolutely no use or value , none whatsoever- the only interest for us in such a multiple is ( as always) to remove that recorded multiple to not produce false, misleading and injurious images when migrating recorded primaries.
It is only when a primary sub-event of the multiple has not been recorded that the multiple can be useful. If the unrecorded subevent of the multiple is itself a multiple then that subevent must be removed- and the unrecorded multiple subevent is problematic and has no usefulness.
Both recorded and unrecorded multiples must be removed in order to image recorded primaries and unrecorded primaries, respectively.
Hence multiples are NOT now rehabilitated events on equal footing with recorded primaries. They are NOT the new primaries and multiples are NEVER migrated (That idea and thought of ‘migrating multiples’ has no meaning, (Please see the attached Multiples: signal or noise? Paper), but as events themselves must always be removed. For those pursuing the use of multiples , it is suggested to inform us as to how unrecorded multiples will be removed.
The use of multiples is worthwhile to pursue, and to develop and deliver. Their value directly depends on the lack of adequate recorded primaries. There is no substitute for recorded primaries for the extraction of complex structural information and subsequent amplitude analysis.( see e.g., the high water mark of migration capability, Stolt CIII migration for heterogeneous media, requires recorded primaries, and methods that use a recorded multiple to obtain an approximate image of an unrecorded primary subevent, cannot achieve a Stolt CIII migration delivery and effectiveness in general, and in particular for complex structure determination and subsequent amplitude analysis (please see the Stolt CIII project below).
WHAT ABOUT MODEL MATCHING METHODS LIKE FWI ? THEY OFTEN USE PRIMARIES AND CERTAIN MULTIPLES( TYPICALLY RELATIVELY SIMPLE FREE SURFACE MULTIPLES, BUT NOT TOO COMPLICATED MULTIPLES, LIKE INTERNAL MULTIPLES, WHICH ARE OFTEN “REMOVED” BEFORE THE MATCHING BEGINS).
There is a very important difference between the role recorded data plays and how data is treated in the direct methods for locating and identifying reservoirs , migration and migration-inversion, compared to model matching methods like FWI.
In migration and migration-inversion the data is actively involved and interrogated for information, whereas in model matching the data is passive, and the modeling and updating and searching are where the action resides. The only direct inverse methods for parameter estimation - the parameter estimation subseries of the inverse scattering series, pioneered by Haiyan Zhang( Please see the attachments) specify the data required and the algorithms, and the required data are a complete set of shot records with multi-component primaries.
In contrast, with model matching there is no guide, no underlying theory or conceptual platform- one trace, many traces, multi-component traces, and horizontal and vertical derivatives of displacement and pressure, and stress measurements and gravity data…in fact, absolutely any data can be chosen to be model matched, including only one trace, or one trace with only multiples. ….It seems reasonable that adding more data and data types would provide more constraints to search algorithms that might benefit and assist the parameter identification objective…however while including free surface multiples with primaries is often viewed as helpful, with added constraints for the modeling to match, the addition of internal multiples seems in practice to be ‘too full’ model matching with too many complicated constraints. Seems that model matching with only primaries is viewed as not ‘full’ enough, with primaries and free surface multiples that feels just right and perfectly full, and with the addition of internal multiples, apparently a little ‘too full’. We are back to the lack of an underlying theory and framework.
Model matching has a role and place, and as with all seismic methods ( including all migration, Green’s theorem and ISS methods) we welcome and encourage a balanced view of the benefits, shortcomings and open issues.
With that broad perspective and context we provide a few comments below of the topics to be discussed at the 2018 M-OSRP Annual Technical Review
Analysis , testing and comparison of the inverse scattering free surface multiple elimination algorithm, and the industry standard SRME plus energy minimization adaptive subtraction, for removing both isolated free surface multiples and those free surface multiples that interfere with primaries, without subsurface information and without damaging the primaries. Results and analysis for both an elastic and inelastic subsurface. Guide to when each is the appropriate choice within the free surface multiple removal toolbox. Basically it will be shown that SRME can, at times, be effective for isolated free surface multiples not proximal to other events -whereas the ISS free surface (FS) multiple elimination algorithm is the appropriate choice when the FS multiple is proximal or interfering with other events.
Inverse scattering series (ISS) internal multiple removal , going beyond the current industry standard ISS internal multiple attenuation plus adaptive subtraction, needed when internal multiples interfere with primaries, ISS internal multiple elimination: evaluation on elastic and inelastic media: recent progress, new tool box capabilities and options and open issues.
The Stolt CIII migration for heterogeneous media developed by M-OSRP has significant benefits for imaging and inverting complex structure, and for resolving and identifying reservoirs-with added value in comparison with all RTM methods . Resolution differences are examined and quantified in this study.
Towards a new formulation of several ISS task specific subseries, for example, the subseries for performing Q compensation without needing to know, estimate or determine Q, and without the need for low frequency data.
Removing sea surface generated random noise without needing any knowledge of the noise generator: implications for low frequency acquisition and processing.
Hugh Roy and Lillie Cranz Cullen Distinguished University Chair in Physics
Director, Mission-Oriented Seismic Research Program
Professor, Dept. of Physics, and Professor, Dept. of Earth and Atmospheric Sciences
Physics Department , SR1 617
University of Houston
Houston, Texas 77204-5005
Office phone 713-743-3848
E mail email@example.com
The link below contains a recent executive summary overview video of our projects, deliverables, impact and plans
and in the link below, we are sharing some recent positive news