2006 RESEARCH UTAM ACCOMPLISHMENTS AND 2007 GOALS
- 3D Reverse Time Migration
- RESULT: 3D and 2D tests of wavefront
migration (WMM) and bottom-up target-oriented
migration (BUM). Both methods promise to be substantially faster
than standard reverse time migration (RTM).
Codes appear to be accurately imaging data from 2D models and
simple 3D models.
- PROBLEMS: None.
- 2007 GOALS: Rigorous tests on performance of RTM, WMM, and BUM.
- Seismic Interferometry
- RESULT: First draft of
book "Seismic Interferometry" given to sponsors.
Includes MATLAB exercises. Classification matrix
coherently organizes different research directions.
- Tested surface wave interferometry on 3D field data
and results for eliminating surface waves are encouraging.
- Tested interferometric interpolation on 2D and 3D data.
Results are good for 2D data, but only fair for 3D data.
- SSP transform data migrated using VSP shot gathers as Green's functions.
Visibly enhances image beneath the salt.
- Preliminary attempt at imaging reflectors beneath salt by elastic
PS wave imaging, including transmitted PS waves.
- Contourlet transformed used to filter surface waves.
- IMPLICATIONS: Surface wave filtering by interferometry looks promising.
Interferometry book organizes research directions around
9x9 classification matrix.
- PROBLEMS: 3D interpolation results not too accurate, but shows some
promise. It needs the extra step of a matched filter.
- 2007 GOALS: Rigorous comparison between
standard 3D surface wave filtering and interferometric
surface wave filtering.
Incorporate matching filter with interferometric
interpolation. Continue contourlet transform research.
Try the SSP->VSP-SWP transform on field data?
Advance efforts to image subsalt reflectors with PS imaging.
- Multiple Migration
Salt flank imaging using Kirchhoff migration
with a double-bounce imaging condition.
- IMPLICATION: Compared to RTM, inexpensive way to illuminate some salt flanks that
are underlain by a base reflector. Transforms (surface seismic profile) SSP data
into VSP data.
- PROBLEMS: Need accurate knowledge of base reflector geometry.
- 2007 GOALS: Multiple migration of 3D data.
- RESULT: Test of multigrid tomography shows
smoother images than dynamic smoothing tomography.
Quick application of early arrival waveform
tomography shows slight improvement over tomostatics
from traveltime tomography.
- IMPLICATION: Perhaps multigrid tomography is moderately
better than dynamic smoothing tomography.
- PROBLEMS: Two sequential internships for Chaiwoot prevented
much effort on the EWT.
- 2006 GOALS: Full EWT tests on new Saudi data.
Possibly elastic inversion.
- Hybrid Computing Shootout
- RESULT: Tested CELL, FPGA, and Multi Core computers for
wave equation modeling.
Results show that the FPGA machine runs about 50% faster than
an AMD multi-core (i.e., dual duo) computer,
but is an order-of-magnitude more difficult to program. The CELL (i.e. \$600. Playstation 3)
is about 70% the speed of an AMD dual duo, and is moderately more
difficult to program than the multi-core computer.
- IMPLICATION: Stay with the multi-core machines.
FPGAs might even gain a 4x speedup with special programming,
but code development is quite costly.
- PROBLEMS: PS3 sold out in most stores. Bought one at last minute.
- 2007 GOALS: Stay with the multi-core machines.
- Migration Deconvolution
- RESULT: Implemented poststack MD on new data set from PEMEX.
Only option was poststack MD because source-receiver coordinates
not available. Fast version of MD was shown and some parts of image show improvement
compared to a decimated full MD result.
- IMPLICATION: Fast MD method used for optimal MD filter parameters.
- PROBLEMS: Need prestack data and source-receiver coordinates.
- 2007 GOALS: Improve MD performance; apply to SaudiAramco data.
- Goal Summary
- We hope to explore bottom-up and wavefront versions of RTM
as a practical 3D imaging tool. Elastic wave imaging
beneath salt is a goal with Xiang's 2D simulations
and PS migrations. Using natural
data to assist with SSP extrapolation is useful
and will be a major topic of research.
There will be a major effort devoted
to early arrival wavefrom tomography
with multiscale elements: Chaiwoot
will devote full time to this endeavor and Sam
Brown will likely be involved using some suggestions from Veritas.
be interaction with our reverse time migration researchers.
Prism wave migration will be explored and its
extension to interferometry and tomography will be elaborated.
Natural datuming methods will be employed for VSP data
and using it to allow SSP data to see below the overburden
Interpolation and surface wave filtering
methods will be employed using
contourlets and interferometry.
3D Visualization of seismic data will be explored in conjunction with
the Computer Science department.