Post on 19-Jan-2016
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
1
Return to Passive Imaging
Brad Artman
June, 6 2002
Ph.D. Proposal
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
2
State of Affairs
Claerbout’s Conjecture
Terra/Helio-seismology EngineeringExploration
Success
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
3
• The nutshell
• Why it works
• As it stands
• Toward the goal
• How to finish
• When will all this be ready?
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
4
• The nutshellThe nutshell– ProblemProblem– ImportanceImportance– ProposalProposal
• Why it works
• As it stands
• Toward the goal
• How to finish
• When will all this be ready?
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
5
Problem
• Who knows if passive seismic imaging works?– Why (not)?
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
6
Importance
• Lots of people care about the subsurface.
• No source required opens opportunities for:– Monitoring– Reconnaissance– Restricted access sites– Old long data sets– New long data sets
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
7
Proposal
I propose to answer whether or not the passive seismic experiment can be used as a practicable methodology for subsurface imaging and/or monitoring.
+ or - I’ll tell you why.
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
8
• The nutshell
• Why it worksWhy it works– Intuitive explanationIntuitive explanation– Rigorous explanationRigorous explanation
• As it stands
• Toward the goal
• How to finish
• When will all this be ready?
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
9
Intuitive explanation
a
r1 r2 r1*r1 r1*r2
b c
t lag
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
10
Rigorous explanation I
1
U
-RR
Conventional reflection seismic
1
-UU
P
Earthquake seismology
Y –R( ) R(Z) + [1+R( )] [ 1+ R(Z)] = Y U( ) U(Z)Z1
Z1
Z1
1 k
1+R( ) R(Z) = a U( ) U(Z)Z1
Z1
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
11
Rigorous explanation IY –R( ) R(Z) + [1+R( )] [ 1+ R(Z)] = Y U( ) U(Z)Z
1Z1
Z1
1 k
1+R( ) R(Z) = a U( ) U(Z)Z1
Z1
1
U
-RR
Conventional reflection seismic
1
-UU
P
Earthquake seismology
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
12
Intuitive explanation
a
r1 r2 r1*r1 r1*r2
b c
t lag
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
13
Rigorous explanation II
U( ) U(Z) = U(p,ω) U(p,ω) = Q(p,ω)Z1
Q(k,ω) = Q( , ω) = |ω|U(k,ω) U(k,ω)1|ω| ω
k
Q(p,ω) = Q( , ω)ω k
Q(p,ω) = |ω| Q(k,ω)
q(x,t) = (t) u(x’,t) u(x+x’,t+t’)t’x’
rx’,x+x’(t)x’
DFT
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
14
Synthetic Testsm
s
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
15
Cross-correlation Technique
Passive Seismic Imaging applied to synthetic data, Rickett SEP-92
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
16
Virtual multiples
a
r1 r2 r1*r1 r1*r2
b c
t lag
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
17
• The nutshell
• Why it works
• As it standsAs it stands– It works!It works!– On the shoulders of giantsOn the shoulders of giants– The raceThe race
• Toward the goal
• How to finish
• When will all this be ready?
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
18
It works!
Acoustic Daylight Imaging via spectral factorization, Rickett SEP-100
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
19
On the shoulders of giants• SEP reports 60 – 86, 100
– SEP’s Franciscan effort and analysis
• Schuster– Correlation migration– Drill bit source
• Bostock– Earthquake array seismograms
• Louie– Engineering characterization
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
20
The Race
• Undisclosed submission
• Company interest
• Livermore???
• The earth from satellite?
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
21
• The nutshell• Why it works• As it stands• Toward the goalToward the goal
– Is 2D possible?Is 2D possible?– Reformulation as migrationReformulation as migration– Motivation to invertMotivation to invert– Sampling the noise-fieldSampling the noise-field– Shape of the noise-fieldShape of the noise-field– Artman GeoServicesArtman GeoServices
• How to finish• When will all this be ready?
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
22
Is 2D possible?
• Yes
• PGS OBC 3C Co-parallel PSD !
Is 2D possible?, Artman, SEP-111
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
23
Reformulation as migrationI(z, x)=Σ P (ω,x,s) P (ω,x,s)
ω
szg
z
P is the wave field at each depth level,where source and receiver fields are propagated independently via SSR.
z
Realization: If P = P , correlation requirement of the passive seismic conjecture
is fulfilled in the migration.
Let the wave equation handle the unknown source.
s g
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
24
Areal Source Wave-fieldShot-profile migration of multiple reflections,Guitton, SEP-111
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
25
Synthetic Testsm
s
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
26
Two horizontal planesm m
m
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
27
Point Diffractor
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
28
Cost Comparison
• Correlations– Spatial increase: N to N (+5 O.M.) – Temporal decrease: N to N (- 4 O.M.)– Migrate (S.P.): N N N N
• Skipping Correlations– No spatial or frequency change: N , N– Migrate (S.P.): N N N N
xy
lag
xy
xy2
xy
t
lag zxy2
t
hxy
t zhxy
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
29
Motivation to invert
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
30
Sampling the noise-field
Minimize storage and processing costsBow to acquisition computer limitations
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
31
No
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
32
Shape of the noise-field
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
33
Artman GeoServices
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
34
• The nutshell
• Why it works
• As it stands
• Toward the goal
• How to finishHow to finish– Brad, the sun, and your coffee tableBrad, the sun, and your coffee table– Santa Clara Valley Seismic ExperimentSanta Clara Valley Seismic Experiment– Recast as deconvolutionRecast as deconvolution– Wave-field separationWave-field separation
• When will all this be ready?
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
35
Solar imaging
• Flare structure, density layering
• Tools: – Absorbing boundary layer modeling– Up-down extrapolator
• James Rickett
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
36
Crustal Seismic
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
37
Deconvolution• Who needs physics?
U = T SS = H W
A U = WA = W W H T -1 -1 -1
= H T-1-1
Assume T is white(ish) and events are well spaced-1
Second PEF estimation & application returns T-1
Levinson recursion calculates the reflection coefficients
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
38
Wave-field separation
u = E L R
I
Up-going wave-field
Physics
Surface displacements
E has azimuthal dependency- this may not work.
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
39
Data commitments• Linear acquisition:
– S. California, USGS. Sept. 2002
– S.C. Valley, USGS. Sept. 2002
– 3C OBC, GoM, PGS, June 2002
– Unspecified test, PGS, Sept. 2002
– Publicly available seismologic deployments
• Areal acquisition:– South Texas, CGG, Sept. 2002
– Sun, SOI, in-house
– S.C. Valley, USGS, in-house
– Moss Landing +, Artman GeoServices, in-house
– Publicly available seismologic deployments
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
40
Time-line
• Summer 2002– 3D up-down extrapolator, absorbing BC’s– Image solar flare, submit for publishing– Image Moss Landing beach
• Urban applications?
– Continue collection of outside data– SEG passive seismic workshop
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
41
Time-line
• Fall 2002– Image Santa Clara Valley, submit for publishing– Gather/acquire data– Image data as it arrives– Physics 210: Particle mechanics
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
42
Time-line
• Winter 2002-Spring 2003– Continue data collection and processing– Comp. Sci. 238: Parallel methods in num. analysis– Begin comparative study of successes and failures– Prepare work for publishing
• Summer 2003 – outside internship
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
43
Time-line
• Fall 2003 – Spring 2004– Finalize comparative analysis– Define success parameters– Prepare for publishing
• Summer 2004– Stat. 110: Stat. Methods in eng. and phys. science– Wrap up outstanding projects
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
44
Time-line
• Fall – Winter 2004– PE 284 : Optimization– Write thesis, journal articles
• Spring 2004– Finalize deliverables for graduation
brad@sep.stanford.edu
Passive Seismic ImagingSEP/Crustal Research Group
45
Thank You