*******************************************************************************
Fast 3D Modeling and Prestack Depth Migration
Using Multi-Screen Wave Propagators

Summary of Second General Meeting

December 15, 1995

University of California/Santa Cruz

Hosted by the Earth Science Department and Institute of Tectonics at UCSC


******************
Meeting Attendees:

This was the second general meeting held for the project. Representatives 
from six industry collaborators, (Conoco, Cray Research, nCUBE, Schlumberger-
Cambridge, Shell, and Unocal), two universities, (University of California/
Santa Cruz, and Massachusetts Institute of Technology), the U.S. Department of 
Energy/Office of Basic Energy Sciences, and Los Alamos National Laboratory 
attended. Representatives from ARCO and Sun Microsystems were unable to 
attend the meeting. 


***********************************
Meeting Purpose and Project Status:

Technical work for this project began officially in August, 1995, when the
sub-contract between LANL and UCSC was initiated. Although the equivalent 
contract with MIT is still under negotiation(!), technical work is already in
progress by all university collaborators. This meeting was held to update all
participants on the project status, to introduce new participants, as well as 
those who missed the first meeting, to the project, and to re-assess project 
goals based on initial technical results and industry priorities.


*****************
Meeting Overview:

Meeting attendees convened at 9:00 AM. After a short welcome and introduction, 
three primary topics were addressed during the course of the meeting: 
	1.) technical progress to date. 
	2.) project plans, technical goals, priorities and contributions.
	3.) administrative issues.

A detailed agenda for the meeting was distributed to all project participants
via the project's bulletin board listserver, as well as by hardcopy handouts to
all meeting attendees. The agenda can be obtained either by connecting to the
project's world-wide-web site home page (see below), or by direct request to 
Peter Roberts, phone: 505-667-1199, email: proberts@lanl.gov. The web site 
also contains HTML versions of this meeting summary, a list of attendees, and
other information relevant to the project and the meetings held. Technical
presentations started at 9:30 AM and concluded after the lunch break. Open
discussions on the last two topics above began at 1:15 PM and concluded when 
the meeting adjourned at 5:00 PM.


*******************
Technical progress:

Summaries of the technical presentations will be installed on the web site as 
the various contributors provide the material. Downloadable copies of some of 
the viewgraphs will also be available there. Hardcopies of all relevant 
viewgraphs will be mailed to each organization. The following gives a cursory 
overview of the technical progress reported at the meeting:
	1.) Improvements to the acoustic and elastic Multi-Screen methods for 
	    modeling reflected and backscattered wave fields have been made, 
	    including wide-angle scattering approximations.
	2.) Numerical analysis and comparisons with other wave propagation
	    methods, such as fnite difference, have given some initially
	    promising results for accuracy of Multi-Screen methods.
	3.) A theoretical framework for determining the accuracy of screen
	    propagators has been established using Hamiltonian path integrals.
	4.) The phase-screen and wide-angle approximations have been used in
	    initial tests for prestack depth migration. Numerous 2-D cases have
	    been tested, and work will begin on a 3-D version for MPP machines. 
	5.) Methods have been developed for generating 3-D structural models
	    for testing on various wave propagation codes.
	6.) MPP versions of a 3-D Finite Difference code have been implemented
	    using MPI and tested on the nCUBE and Cray T3D. The code can be
	    used for accuracy comparisons with Multi-Screen methods.
	7.) A PVM version of a PSPI 3-D prestack depth migration code has been
	    implemented on a workstation cluster and the Cray T3D. Although the 
	    code cannot be distributed, it can be used by LANL personnel 
	    for performance comparisons with the phase-screen migration codes.


*****************************
Project goals and priorities:

The project's initial technical goals and industry priorities were defined at 
the first general meeting at LANL, May 8-9, 1995. The detailed summary of that
meeting is available on the project home page.

The following lists the top five priorities identified previously:
	1.) Develop fast scalar 3D imaging codes based on phase-screen method.
		-start with 2D, serial, post-stack and simple models.
		-extend to prestack, 3D, parallel and realistic models.
	2.) Perform accuracy analysis and define limitations of screen methods.
		-compare accuracy with other methods numerically.
                -perform theoretical/asymptotic analysis of accuracy.
		-define application range and limits of the methods.
	3.) Perform benchmarking comparisons against other methods.
		-tests on simple models.
		-comparisons with currently accepted imaging techniques.
		-evaluate computational speed and throughput.
	4.) Extend screen methods to deal with anisotropy and attenuation.
	5.) Improve modeling for wide-angle scattering and multiples.

Of the eight industry collaborators, only ARCO and Sun Microsystems were not
represented at the second meeting. Everyone who did attend the meeting agreed
that the priorities above are still appropriate and should not be changed. The
overall consensus was that technical progress on the majority of these tasks
is promising so far and work should continue as originally planned. The central
focus of the project will continue to be the development and assessment of
multi-screen and hybrid wave propagation methods for modeling and imaging. The
highest priority to industry is to develop new methods for 3D scalar-wave 
imaging in complicated media which are faster and more accurate than existing
in-house capabilities. The project will exploit the inherent computational
efficiency of screen methods, determine the accuracy and applicability, and
implement parallel strategies for processing large 3D seismic surveys.


*************************************
University and LANL Responsibilities:

The following general tasks were confirmed for UCSC, MIT and LANL personnel:
	1.) Development of phase-screen migration codes (UCSC, LANL)
	2.) Development of hybrid modeling and migration methods (MIT, UCSC)
	3.) Accuracy analysis and calibration of algorithms (All)
	4.) Parallel implementations and benchmarking (All)
	5.) Extensions/improvements to screen methods (UCSC)
Los Alamos and MIT participants agreed to provide the project access to useful
results of other ongoing exploration research efforts in their groups if 
allowable under the terms of existing agreements for those projects.


**********************************
Industry Comments and Suggestions:

Conoco:
  -Accuracy and speed of one-way, scalar-wave imaging is highest priority.
  -Define range of applicability of different sets of imaging tools.
  -Work closely with MIT on 3D propagation problems.
	-extensions to finite difference
	-calibration of various methods
  -Exploit natural overlaps with other ACTI projects.
	-library I/O routines
	-parallel strategies
	-test models and data
  -Industry partners should help with defining and building test models.
	-SEG salt model
	-GOCAD support

Cray:
  -Evaluation of code performance on MPP machines is highest priority.
	-optimization issues for high performance computing
	-determine relative speed of different algorithms for imaging
  -Most interested in prestack migration using F-K methods.
  -Prefers MPI parallel implementations over PVM.

nCUBE:
  -Evaluation of code performance on MPP machines is highest priority.
	-optimization issues for high performance computing
	-determine relative speed of different algorithms for imaging
	-direct comparisons with parallel 3D finite difference
  -Prefers MPI parallel implementations over PVM.

Schlumberger:
  -Accuracy issues are a high priority.
  -If methods not good for migration, may still work for velocity analysis.
  -Continue development of 3D elastic modeling with screen methods.
	-wide angle scattering
	-multiples
	-anisotropy
  -Continue asymptotic analysis.

Shell:
  -Accuracy and speed of one-way, scalar-wave imaging is highest priority.
  -Most interested in 3D migration as soon as possible.
  -Keep test models simple at first, then more realistic.
	-velocity contrasts appropriate for salt bodies
	-2D: Marmousi model
	-3D: eventually try SEG salt model
  -Comparisons with existing in-house codes are important.

Unocal:
  -Accuracy and speed of one-way, scalar-wave imaging is highest priority.
  -Framework so far is promising. It's time to get more rigorous.
  -Begin testing more realistic sample models.
	-large lateral velocity variations
	-steeply dipping relectors
  -Continue forward modeling efforts as lower-priority, long-term goal.
	-elastic waves
	-include multiples
	-include anisotropy


***********************
Industry Contributions:

Industry collaborators agreed to provide the following in-kind services:
  -technical guidance and assessment for all aspects of the project.
  -code evaluation and testing against existing in-house methods.
  -access to results of tests performed on project-generated codes.
  -access to test data when appropriate and allowable.
  -help with design and contruction of test models.
  -access to computing facilities as appropriate.
  -support for implementing codes on specific MPP machines.
  -help with data and image visualization.


**********************
Administrative Issues:

1.) Communication:
	-Primary communication of project results, reports, updates and
	 general information will be maintained through a central network web
	 site at Los Alamos. The site will be maintained by Peter Roberts, who
	 should be notified when anyone wishes to add something to the home 
	 pages. Links to other participant's web sites will be added as needed.
	 The LANL web site is already under construction. It is located at:
		http://www.ees4.lanl.gov/ACTI-NEWMETHODS
	 The top-level home page is called "ACTI-NEWMETHODS.HTML", which has
	 links to the three sub-projects. Ours is called "MULTI-SCREEN.HTML".
	 Please tour this site frequently and feel free to make suggestions
	 to Peter (proberts@lanl.gov).
	-The project listserver will remain the primary means of making
	 announcements, posting important issues and carrying on conversations
	 of general interest to all participants. Peter will continue to post
	 project updates, meeting announcements and the like, as before. He 
	 will also use it to inform everyone when something new and interesting 
	 has been added to the web site. Information about the listserver and 
	 how to subscribe, etc., has already been installed in the multi-screen
	 home page. The list will remain open to the world. Please use it.
	-Peter Roberts has been designated as the project's primary point of
	 contact for any and all issues of general or specific importance.
	 There will be no main industry contact. This is a service, NOT a
	 restriction. He will see to it that your concerns are communicated to
	 the project as a whole. If he does not hear from you, he will assume
	 you are happy, whatever the circumstances may be. If, however, there
	 is an urgent issue at hand, requiring feedback from everyone, he will
	 seek you out individually if necessary.
2.) Project Updates:
	-Progress reports and updates will be prepared approximately every
	 month or two, or as warranted. Brief summaries will be posted to the
	 list and more detailed reports will be installed on the web site.
	-It is the responsibility of the project PI's to keep Los Alamos people
	 up-to-date with new results, so they can be disseminated efficiently.
	-List postings and web pages will provide current lists of specific
	 technical tasks that various PI's are working on and what the expected
	 near-term deliverables are.
3.) Participants Agreement:
	-The first version of the agreement was determined to be incomplete in
	 the wording of the right-to-use clause. Stronger wording is being
	 added that specifically grants all particpants a lifetime royalty-free
	 license to use any results and algorithms produced by the project.
	 The next version will be posted and put on the web when it's done.
	-Everyone at the meeting agreed, once again, that no legal agreement
	 regarding the protection of intellectual property or non-disclosure
	 of project results should be imposed on the participants. The project
	 listserver and web site will remain open, with no password protection.
	-Anything contributed to the project is done so at the discretion of
	 the contributor and will be considered open to view by all official
	 project participants.
	-Any public disclosures must be agreed upon by concensus of at least	 	 one representative from each participating organization. However, all
	 participants agreed to impose no restrictions on UCSC, MIT or LANL
	 regarding timely publication of results or distribution of codes
	 developed by them in fulfillment of project funding requirements.
	-First-hand access to project results and codes was considered to be
	 sufficient protection of official collaborators' priveleges.
4.) Project General Meetings:
	-One comprehensive, "mandatory" meeting will be scheduled per year.
	 This will be planned sufficiently in advance to ensure that at least
	 one representative from each organization will attend.
	-Additional meetings will be held as warranted. It was agreed that
	 meetings would be scheduled if enough critical issues arise that need
	 to be discussed in person, rather than by electronic means as above.
	-The next full-blown general meeting has been proposed to be held 
	 in July, 1996 at MIT. Please start thinking about what dates will be
	 convenient and whether or not the location is acceptable.

*******************************************************************************