introduction to the receiver function method · 2018. 2. 6. · 1introduction. 2crustal vp/vs ratio...
TRANSCRIPT
Introduction to the receiver function method
Frederik Tilmann (+ material from Rainer Kind)
4D-MB SPP Short Course, 1 February 2018
Tilmann (GFZ,FUB) Receiver Functions 1 / 0
Outline
1 Introduction
2 Crustal Vp/Vs ratio and Moho depth from stacking of multiples
3 Common Conversion Point Stacks (CCCP)
paired with lecture on seismic anisotropy (Georg Rumpker)
Tilmann (GFZ,FUB) Receiver Functions 2 / 0
Outline
1 Introduction
2 Crustal Vp/Vs ratio and Moho depth from stacking of multiples
3 Common Conversion Point Stacks (CCCP)
Tilmann (GFZ,FUB) Receiver Functions 3 / 0
Target: Imaging subsurface structure, discontinuities
TRANSALP profile Vibroseis+Explosion+passive 1998,2001 [TRANSALP Working Group, 2002]
Controlled source reflectionseismology:
Use reflections of P energy
Method of choice for shallowstructure but energy does mostlynot penetrate deeply
No information on S velocity
Very expensive to carry out
⇒ Use earthquakes as energy source
⇒ The energy is coming from below
⇒ We need to use conversionsinstead of reflections
Tilmann (GFZ,FUB) Receiver Functions 4 / 0
Target: Imaging subsurface structure, discontinuities
TRANSALP profile Vibroseis+Explosion+passive 1998,2001 [TRANSALP Working Group, 2002]
Controlled source reflectionseismology:
Use reflections of P energy
Method of choice for shallowstructure but energy does mostlynot penetrate deeply
No information on S velocity
Very expensive to carry out
⇒ Use earthquakes as energy source
⇒ The energy is coming from below
⇒ We need to use conversionsinstead of reflections
Tilmann (GFZ,FUB) Receiver Functions 4 / 0
Moho depth and Vp/Vs ratios
Lombardi et al. [2008]
VP/VS varies typically 1.6–2.0, stronglyindicative of lithology,felsic ⇒ mafic implies small ⇒ large VP/VS
values.
Tilmann (GFZ,FUB) Receiver Functions 5 / 0
Basics: P and S waves
VP =
√K + 4
3µ
ρ
K Bulk modulus
µ Shear modulus
ρ Density
VS =
õ
ρ
At the same period S wavelength is shorter due to shorter propagation velocity
Tilmann (GFZ,FUB) Receiver Functions 6 / 0
Basics: P and S waves
VP =
√K + 4
3µ
ρ
K Bulk modulus
µ Shear modulus
ρ Density
VS =
õ
ρ
At the same period S wavelength is shorter due to shorter propagation velocity
Tilmann (GFZ,FUB) Receiver Functions 6 / 0
P and S wave: interaction with interface
Incident P
Incident P ⇒ converted SV wave(except when exactly verticallyincident)
No SH wave (T component)converted by flat interface
Also works the other way round:incident SV ⇒ converted P
Note that in reality the different types of motion overlap and affect the same volume.
Tilmann (GFZ,FUB) Receiver Functions 7 / 0
P and S wave: interaction with interface
Incident P
R
Z
Incident P ⇒ converted SV wave(except when exactly verticallyincident)
No SH wave (T component)converted by flat interface
Also works the other way round:incident SV ⇒ converted P
Note that in reality the different types of motion overlap and affect the same volume.
Tilmann (GFZ,FUB) Receiver Functions 7 / 0
P and S wave: interaction with interface
Incident P
R
Z
Refracted P
Converted S
V
Incident P ⇒ converted SV wave(except when exactly verticallyincident)
No SH wave (T component)converted by flat interface
Also works the other way round:incident SV ⇒ converted P
Note that in reality the different types of motion overlap and affect the same volume.
Tilmann (GFZ,FUB) Receiver Functions 7 / 0
P and S wave: interaction with interface
Incident P
R
Z
Refracted P
Converted S
V
Ref
lect
ed P
Ref
lect
/con
v S
V
Incident P ⇒ converted SV wave(except when exactly verticallyincident)
No SH wave (T component)converted by flat interface
Also works the other way round:incident SV ⇒ converted P
Note that in reality the different types of motion overlap and affect the same volume.
Tilmann (GFZ,FUB) Receiver Functions 7 / 0
P and S wave: interaction with interface
Incident P
R
Z
Refracted P
Converted S
V
Ref
lect
ed P
Ref
lect
/con
v S
V
Incident P ⇒ converted SV wave(except when exactly verticallyincident)
No SH wave (T component)converted by flat interface
Also works the other way round:incident SV ⇒ converted P
Note that in reality the different types of motion overlap and affect the same volume.
Tilmann (GFZ,FUB) Receiver Functions 7 / 0
P and S wave: interaction with interfaceTechnical details
Wavefronts along interface need to stay in phase: ⇒ Snell’s law: sin ι1P
VP= sin ι2S
VS
⇒ Converted S waves propagate more steeply than their originating P waves
Tilmann (GFZ,FUB) Receiver Functions 8 / 0
Using RF for imaging discontinuity depths
P wave from earthquake
Direct P
P-converted-SV at D
Seismic discontinuities
M
D
P-converted-SV at M
Use time-separation of waves converted fromP-to-S at discontinuities underneath thereceiver.
Take home
Interface with velocity increase with depthshows up as positive wiggle, velocity decrease asnegative wiggle.
Fast-to-slow (e.g. Moho)
Slow-to-fast (e.g. magma chamber, LAB)
Using RF for imaging discontinuity depths
P wave from earthquake
Direct P
P-converted-SV at D
Seismic discontinuities
M
D
P-converted-SV at M
Use time-separation of waves converted fromP-to-S at discontinuities underneath thereceiver.
Take home
Interface with velocity increase with depthshows up as positive wiggle, velocity decrease asnegative wiggle.
Fast-to-slow (e.g. Moho)
Slow-to-fast (e.g. magma chamber, LAB)
Diffractions
Ryberg&Weber (2000) Rondenay [2009]
Anything that is not a smooth horizontal interface will give rise to diffractions, in particular also
topography of any interface.
Tilmann (GFZ,FUB) Receiver Functions 10 / 0
Incident P wave in a simple crust
R component(SV+P)
T component(SH)
Z component(P+SV)
P (norm.)
SV (norm.)
Incident P wave in a simple crust
R component(SV+P)
T component(SH)
Z component(P+SV)
P (norm.)
SV (norm.)
What are multiples?
Fig: A. Frassetto
NB: such a clean signal is rarelyachieved
P- and S-receiver functions
Methods – P and S Receiver Functions
P S Surface Waves
Tilmann (GFZ,FUB) Receiver Functions 12 / 0
The making of a receiver functionMultiple effects build the seismograms
R component traces at station in Australia
[Kennett, 2002]
But: the conversions are hard to see in raw data becausein addition to the sequence of interfaces, the seismogramsare influenced by
Instrument response
Source complexity
Near-source structure (e.g. surface reflections)
Each effect is combined with the others by a processcalled convolution.Here, We are interested in receiver structure⇒ we do not need to know these effects in detail.
Tilmann (GFZ,FUB) Receiver Functions 13 / 0
Refresher: convolution and deconvolutionNoise-free synthetic data (T0 = 10s)
0 20 40 60 80
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time (s)
Acc
eler
atio
n (m
/s)
∗Conv.
0 20 40 60 800
0.2
0.4
0.6
0.8
1
Time (s)
V
0 20 40 60 80
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Time (s)
V
NB Deconvolution is inherently unstable and needs to be regularised (different methods
available)! Aster et al. [2005]
Refresher: convolution and deconvolutionNoise-free synthetic data (T0 = 10s)
0 20 40 60 80
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time (s)
Acc
eler
atio
n (m
/s)
∗Conv.
0 20 40 60 800
0.2
0.4
0.6
0.8
1
Time (s)
V
⇒
0 20 40 60 80
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Time (s)
V
−−−→Decon
NB Deconvolution is inherently unstable and needs to be regularised (different methods
available)! Aster et al. [2005]
Refresher: convolution and deconvolutionNoise-free synthetic data (T0 = 10s)
0 20 40 60 80
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time (s)
Acc
eler
atio
n (m
/s)
∗Conv.
0 20 40 60 800
0.2
0.4
0.6
0.8
1
Time (s)
V
⇒
0 20 40 60 80
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Time (s)
V
−−−→Decon
−20 0 20 40 60 80 100−0.2
0
0.2
0.4
0.6
0.8
1
1.2
Time (s)
Acc
eler
atio
n (m
/s)
NB Deconvolution is inherently unstable and needs to be regularised (different methods
available)! Aster et al. [2005]
Refresher: convolution and deconvolutionNoisy synthetic data (T0 = 10s)
0 20 40 60 80
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time (s)
Acc
eler
atio
n (m
/s)
∗Conv.
0 20 40 60 800
0.2
0.4
0.6
0.8
1
Time (s)
V
⇒
0 20 40 60 800
1
2
3
4
5
Time (s)
V
−−−→Decon
NB Deconvolution is inherently unstable and needs to be regularised (different methods
available)! Aster et al. [2005]
Refresher: convolution and deconvolutionNoisy synthetic data (T0 = 10s)
0 20 40 60 80
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time (s)
Acc
eler
atio
n (m
/s)
∗Conv.
0 20 40 60 800
0.2
0.4
0.6
0.8
1
Time (s)
V
⇒
0 20 40 60 800
1
2
3
4
5
Time (s)
V
−−−→Decon
0 20 40 60 80
−4
−3
−2
−1
0
1
2
3
4
5
Time (s)
Acc
eler
atio
n (m
/s)
NB Deconvolution is inherently unstable and needs to be regularised (different methods
available)! Aster et al. [2005]
Refresher: convolution and deconvolutionNoisy synthetic data (T0 = 10s)
0 20 40 60 80
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time (s)
Acc
eler
atio
n (m
/s)
∗Conv.
0 20 40 60 800
0.2
0.4
0.6
0.8
1
Time (s)
V
⇒
0 20 40 60 800
1
2
3
4
5
Time (s)
V
−−−→Decon
0 20 40 60 80
−4
−3
−2
−1
0
1
2
3
4
5
Time (s)
Acc
eler
atio
n (m
/s)
NB Deconvolution is inherently unstable and needs to be regularised (different methods
available)! Aster et al. [2005]
The making of a receiver functionIsolating the receiver effect by deconvolution
The ‘receiver function’ idea: Use the P wave as proxy for the incident wavesignature (source∗ near source∗mantle propagation∗instrument) [Vinnik, 1977,
Langston, 1979].
Deconvolve vertical component from radial component to obtain RF
Deconvolution
Deconvolution - division in freq domain (fourier transformed time series)
Rf (ω) = c(ω)R(ω)
Z(ω)
The deconvolution removes source and instrument effects (cancel in division)
⇒ RF is only dependent on the structure below receiver and incidence angle
The making of a receiver functionIsolating the receiver effect by deconvolution
The ‘receiver function’ idea: Use the P wave as proxy for the incident wavesignature (source∗ near source∗mantle propagation∗instrument) [Vinnik, 1977,
Langston, 1979].
Deconvolve vertical component from radial component to obtain RF
Deconvolution
Deconvolution - division in freq domain (fourier transformed time series)
Rf (ω) = c(ω)R(ω)
Z(ω)
The deconvolution removes source and instrument effects (cancel in division)
⇒ RF is only dependent on the structure below receiver and incidence angle
The making of a receiver functionIsolating the receiver effect by deconvolution
The ‘receiver function’ idea: Use the P wave as proxy for the incident wavesignature (source∗ near source∗mantle propagation∗instrument) [Vinnik, 1977,
Langston, 1979].
Deconvolve vertical component from radial component to obtain RF
Deconvolution
Deconvolution - division in freq domain (fourier transformed time series)
Rf (ω) = c(ω)R(ω)
Z(ω)
The deconvolution removes source and instrument effects (cancel in division)
⇒ RF is only dependent on the structure below receiver and incidence angle
The making of a receiver functionVariants: component decomposition
— —[Rondenay, 2009]
Deconvolve Z from R, or L from Q.
Take-home
ZRT RF will have large peak at 0 s/0 km. Sedimentary basins can lead to a slight shift of
LQT RF will have nearly zero amplitude at 0 s/0 km. In presence of thick sediments, strongamplitudes near 0 km / 0 s.
For deeper structures no strong difference between both methods.
Tilmann (GFZ,FUB) Receiver Functions 16 / 0
The making of a receiver functionVariants: component decomposition
— —[Rondenay, 2009]
Deconvolve Z from R, or L from Q.
Take-home
ZRT RF will have large peak at 0 s/0 km. Sedimentary basins can lead to a slight shift of
LQT RF will have nearly zero amplitude at 0 s/0 km. In presence of thick sediments, strongamplitudes near 0 km / 0 s.
For deeper structures no strong difference between both methods.
Tilmann (GFZ,FUB) Receiver Functions 16 / 0
P- and S receiver functions
Crustal multiples in P-RF obscure mantle discontinuities
As conversion appear before main phase, this does not affect S-RF (buthigher noise level, lower frequency, potential contamination by other phases)
Outline
1 Introduction
2 Crustal Vp/Vs ratio and Moho depth from stacking of multiples
3 Common Conversion Point Stacks (CCCP)
Tilmann (GFZ,FUB) Receiver Functions 18 / 0
Moveout of direct phases and multiples
[Kind and Yuan, 2011]
Fig: A. Frassetto
Tilmann (GFZ,FUB) Receiver Functions 19 / 0
H − κ stacking methodIntroduced by Zhu [2000] – κ = VP/VS
Observables:
∆tPs = H(ηS − ηP ) (1)
∆tPpPms = h(ηS + ηP ) (2)
∆tPsPms = 2hηS (3)
(Definitions:
ηP =√
V −2P − p2, ηS =
√V −2
S − p2)
p is given by distance of earthquakeUnknowns:
H Depth to Moho
VP Average P velocity crust
VS Average S velocity crust
Redundant equations ⇒ only 2 param.can bedetermined independently.Usually one fixes VP and searches forcombinations of H and κ = VP/VS that fit theobserved RFs.Based on trial values, calculate ∆tPS,PpmS,PsPms
and calculate
A(H, κ) = 0.7Rf (∆tPs )+0.2Rf (∆tPpPs )−0.1Rf (∆tPpSs )Tilmann (GFZ,FUB) Receiver Functions 20 / 0
Good data: permanent stations in CaliforniaZhu and Kanamori [2000]
Tilmann (GFZ,FUB) Receiver Functions 21 / 0
Case study: SELASOMA profile (Southern Madagascar)Rindraharisaona et al. [2017]
44˚ 46˚ 48˚
-24˚
-22˚
-20˚
A
B
C
MS
01
MS
02
MS
03
MS
04
MS
05
MS
06
MS
07
MS
08
MS09
MS
10
MS11
MS
12
MS
13
MS
14
MS
15
MS
16 M
S17 M
S18 M
S19
MS
20
MS
21
MS
22
MS
23
MS24
VO
I
LO
NA
AM
01
AM04
AM
05
AM06
AM07
AM08
AM
09 AM10
AM12
AM13
AM
14
AM
15
AM16
AM
17
AM18
AM19
AM
20
AM21
AM22
AM23
RUM1
RUM2
RUM3
RUM4
RUM5
FOMA
MAHA
Archean
Proterozoic
Morondava basin
Br
Bg-Rn
Bg-Rn
Zz
If
If
Am
Ej
Cn
Cr Cb-Jr
Vb
Ad
An
Ik
ItAt
Ag
Vl
SELASOMA
SELASOMA
RHUM-RUM
MACOMO
Permanent
44˚ 46˚ 48˚ 50˚-26˚
-24˚
-22˚
-20˚
-18˚
-16˚
-14˚
-12˚
30
60
90
120
150
180
Typical temporary array data (1-2 years deployment)
Tilmann (GFZ,FUB) Receiver Functions 22 / 0
Case study: SELASOMA profile (Southern Madagascar)ZR Receiver function stacks
0 5 10 15 20 25
time(s)
MS01 (05)
MS02 (09)
MS03 (12)
MS04 (12)
LONA (12)
MS05 (10)
Cenozoic-Cretaceous
0 5 10 15 20 25
time(s)
MS06 (10)
Carboniferous-Jurassic
MS07 (12)
MS08 (12)
AM21 (08)
MS09 (13)
0 5 10 15 20 25
time(s)
MS24 (21)
Cretaceous volcanic
AM02 (02)
AM01 (07)
RUM2 (9)
RUM1 (14)
MAHA (14)
RUM3 (10)
0 5 10 15 20 25
time(s)
MS10 (20)
Androyen domain
MS11 (11)
AM20 (02)
AM22 (13)
MS12 (16)
0 5 10 15 20 25
time(s)
MS13 (16)
Anosyen domain
AM19 (09)
AM23 (09)
AM17 (12)
MS14 (05)
FOMA (36)
RUM4 (22)
RUM5 (21)
0 5 10 15 20 25
time(s)
MS15 (24)
Ikalamavony domain
MS16 (16)
AM16 (10)
AM18 (08)
0 5 10 15 20 25
time(s)
VOI (51)
Antananarivo domain
AM14 (10)
AM13 (08)
MS17 (16)
AM15 (07)
MS18 (08)
AM08 (06)
MS19 (20)
AM06 (11)
AM07 (14)
MS20 (16)
MS21 (11)
AM09 (10)
AM11 (02)
MS22 (16)
AM04 (07)
MS23 (24)
AM12 (10)
AM10 (06)
Tilmann (GFZ,FUB) Receiver Functions 23 / 0
Case study: SELASOMA profile (Southern Madagascar)ZR Receiver function stacks
Archaean:
0 5 10 15 20 25
time(s)
MS01 (05)
MS02 (09)
MS03 (12)
MS04 (12)
LONA (12)
MS05 (10)
Cenozoic-Cretaceous
0 5 10 15 20 25
time(s)
MS06 (10)
Carboniferous-Jurassic
MS07 (12)
MS08 (12)
AM21 (08)
MS09 (13)
0 5 10 15 20 25
time(s)
MS24 (21)
Cretaceous volcanic
AM02 (02)
AM01 (07)
RUM2 (9)
RUM1 (14)
MAHA (14)
RUM3 (10)
0 5 10 15 20 25
time(s)
MS10 (20)
Androyen domain
MS11 (11)
AM20 (02)
AM22 (13)
MS12 (16)
0 5 10 15 20 25
time(s)
MS13 (16)
Anosyen domain
AM19 (09)
AM23 (09)
AM17 (12)
MS14 (05)
FOMA (36)
RUM4 (22)
RUM5 (21)
0 5 10 15 20 25
time(s)
MS15 (24)
Ikalamavony domain
MS16 (16)
AM16 (10)
AM18 (08)
0 5 10 15 20 25
time(s)
VOI (51)
Antananarivo domain
AM14 (10)
AM13 (08)
MS17 (16)
AM15 (07)
MS18 (08)
AM08 (06)
MS19 (20)
AM06 (11)
AM07 (14)
MS20 (16)
MS21 (11)
AM09 (10)
AM11 (02)
MS22 (16)
AM04 (07)
MS23 (24)
AM12 (10)
AM10 (06)
0 5 10 15 20 25
time(s)
MS01 (05)
MS02 (09)
MS03 (12)
MS04 (12)
LONA (12)
MS05 (10)
Cenozoic-Cretaceous
0 5 10 15 20 25
time(s)
MS06 (10)
Carboniferous-Jurassic
MS07 (12)
MS08 (12)
AM21 (08)
MS09 (13)
0 5 10 15 20 25
time(s)
MS24 (21)
Cretaceous volcanic
AM02 (02)
AM01 (07)
RUM2 (9)
RUM1 (14)
MAHA (14)
RUM3 (10)
0 5 10 15 20 25
time(s)
MS10 (20)
Androyen domain
MS11 (11)
AM20 (02)
AM22 (13)
MS12 (16)
0 5 10 15 20 25
time(s)
MS13 (16)
Anosyen domain
AM19 (09)
AM23 (09)
AM17 (12)
MS14 (05)
FOMA (36)
RUM4 (22)
RUM5 (21)
0 5 10 15 20 25
time(s)
MS15 (24)
Ikalamavony domain
MS16 (16)
AM16 (10)
AM18 (08)
0 5 10 15 20 25
time(s)
VOI (51)
Antananarivo domain
AM14 (10)
AM13 (08)
MS17 (16)
AM15 (07)
MS18 (08)
AM08 (06)
MS19 (20)
AM06 (11)
AM07 (14)
MS20 (16)
MS21 (11)
AM09 (10)
AM11 (02)
MS22 (16)
AM04 (07)
MS23 (24)
AM12 (10)
AM10 (06)
Direct wave at 0 s
Clear mid-crustaldiscontinuity (uppercrust/lower crust)
Clear but relatively weakMoho conversion andmultiples
Proterozoic:0 5 10 15 20 25
time(s)
MS01 (05)
MS02 (09)
MS03 (12)
MS04 (12)
LONA (12)
MS05 (10)
Cenozoic-Cretaceous
0 5 10 15 20 25
time(s)
MS06 (10)
Carboniferous-Jurassic
MS07 (12)
MS08 (12)
AM21 (08)
MS09 (13)
0 5 10 15 20 25
time(s)
MS24 (21)
Cretaceous volcanic
AM02 (02)
AM01 (07)
RUM2 (9)
RUM1 (14)
MAHA (14)
RUM3 (10)
0 5 10 15 20 25
time(s)
MS10 (20)
Androyen domain
MS11 (11)
AM20 (02)
AM22 (13)
MS12 (16)
0 5 10 15 20 25
time(s)
MS13 (16)
Anosyen domain
AM19 (09)
AM23 (09)
AM17 (12)
MS14 (05)
FOMA (36)
RUM4 (22)
RUM5 (21)
0 5 10 15 20 25
time(s)
MS15 (24)
Ikalamavony domain
MS16 (16)
AM16 (10)
AM18 (08)
0 5 10 15 20 25
time(s)
VOI (51)
Antananarivo domain
AM14 (10)
AM13 (08)
MS17 (16)
AM15 (07)
MS18 (08)
AM08 (06)
MS19 (20)
AM06 (11)
AM07 (14)
MS20 (16)
MS21 (11)
AM09 (10)
AM11 (02)
MS22 (16)
AM04 (07)
MS23 (24)
AM12 (10)
AM10 (06)
Direct wave at 0 s
Only weak conversionfrom within the crust
Distinct Moho conversionand (at some stations)multiples
Sedimentary basin:
0 5 10 15 20 25
time(s)
MS01 (05)
MS02 (09)
MS03 (12)
MS04 (12)
LONA (12)
MS05 (10)
Cenozoic-Cretaceous
0 5 10 15 20 25
time(s)
MS06 (10)
Carboniferous-Jurassic
MS07 (12)
MS08 (12)
AM21 (08)
MS09 (13)
0 5 10 15 20 25
time(s)
MS24 (21)
Cretaceous volcanic
AM02 (02)
AM01 (07)
RUM2 (9)
RUM1 (14)
MAHA (14)
RUM3 (10)
0 5 10 15 20 25
time(s)
MS10 (20)
Androyen domain
MS11 (11)
AM20 (02)
AM22 (13)
MS12 (16)
0 5 10 15 20 25
time(s)
MS13 (16)
Anosyen domain
AM19 (09)
AM23 (09)
AM17 (12)
MS14 (05)
FOMA (36)
RUM4 (22)
RUM5 (21)
0 5 10 15 20 25
time(s)
MS15 (24)
Ikalamavony domain
MS16 (16)
AM16 (10)
AM18 (08)
0 5 10 15 20 25
time(s)
VOI (51)
Antananarivo domain
AM14 (10)
AM13 (08)
MS17 (16)
AM15 (07)
MS18 (08)
AM08 (06)
MS19 (20)
AM06 (11)
AM07 (14)
MS20 (16)
MS21 (11)
AM09 (10)
AM11 (02)
MS22 (16)
AM04 (07)
MS23 (24)
AM12 (10)
AM10 (06)
Strongest peak fromsediment basementconversion, not directwave (at most stations)
‘Messier’ RF
Moho difficult to separatefrom sediment multiples
Working with less-than-perfect dataWhat can go wrong?
Multiples too weak, or scatteredin time
Cause Lateral heterogeneity(dipping boundary,small-scale heterogeneity),gradual transition
Consequence Result iscontrolled solely by Psconversion, perfect trade-off,need to assume VP/VS ratioto get H or vice versa.
Ambiguous phase identification
Cause Most prominentdiscontinuity might not bethe target one (e.g. Moho);
Consequence Misinterpretation.Completely wrongmeasurements notrepresentative of eitherdiscontinuity can result frominterference of multiples.
Precambrian crust - Madagascar [Rindraharisaona et al., 2017]
Well constrained Strong trade-off
Tilmann (GFZ,FUB) Receiver Functions 25 / 0
Working with less-than-perfect dataWhat can go wrong?
Multiples too weak, or scatteredin time
Cause Lateral heterogeneity(dipping boundary,small-scale heterogeneity),gradual transition
Consequence Result iscontrolled solely by Psconversion, perfect trade-off,need to assume VP/VS ratioto get H or vice versa.
Ambiguous phase identification
Cause Most prominentdiscontinuity might not bethe target one (e.g. Moho);
Consequence Misinterpretation.Completely wrongmeasurements notrepresentative of eitherdiscontinuity can result frominterference of multiples.
Synthetic example [Wolbern and Rumpker, 2017]Sharp Moho Transitional Moho
Tilmann (GFZ,FUB) Receiver Functions 25 / 0
Working with less-than-perfect dataWhat can go wrong?
Multiples too weak, or scatteredin time
Cause Lateral heterogeneity(dipping boundary,small-scale heterogeneity),gradual transition
Consequence Result iscontrolled solely by Psconversion, perfect trade-off,need to assume VP/VS ratioto get H or vice versa.
Ambiguous phase identification
Cause Most prominentdiscontinuity might not bethe target one (e.g. Moho);
Consequence Misinterpretation.Completely wrongmeasurements notrepresentative of eitherdiscontinuity can result frominterference of multiples.
Cretaceous volcanics - Madagascar [Rindraharisaona et al., 2017]
Upper crust - lower crust Moho
Tilmann (GFZ,FUB) Receiver Functions 25 / 0
Working with less-than-perfect dataWhat can go wrong?
Multiples too weak, or scatteredin time
Cause Lateral heterogeneity(dipping boundary,small-scale heterogeneity),gradual transition
Consequence Result iscontrolled solely by Psconversion, perfect trade-off,need to assume VP/VS ratioto get H or vice versa.
Ambiguous phase identification
Cause Most prominentdiscontinuity might not bethe target one (e.g. Moho);
Consequence Misinterpretation.Completely wrongmeasurements notrepresentative of eitherdiscontinuity can result frominterference of multiples.
Cretaceous volcanics - Madagascar [Rindraharisaona et al., 2017]
Upper crust - lower crust Moho
Tilmann (GFZ,FUB) Receiver Functions 25 / 0
Working with less-than-perfect dataWhat can go wrong?
Multiples too weak, or scatteredin time
Cause Lateral heterogeneity(dipping boundary,small-scale heterogeneity),gradual transition
Consequence Result iscontrolled solely by Psconversion, perfect trade-off,need to assume VP/VS ratioto get H or vice versa.
Ambiguous phase identification
Cause Most prominentdiscontinuity might not bethe target one (e.g. Moho);
Consequence Misinterpretation.Completely wrongmeasurements notrepresentative of eitherdiscontinuity can result frominterference of multiples.
Cretaceous volcanics - Madagascar [Rindraharisaona et al., 2017]
Upper crust - lower crust Moho
Tilmann (GFZ,FUB) Receiver Functions 25 / 0
Case study: SELASOMA profile (Southern Madagascar)
Moho depth VP/VS ratio
44˚ 46˚ 48˚
-24˚
-22˚
-20˚
A
B
C
Archean
Proterozoic
Morondava basin
Volcanic
Br
Bg-Rn
Bg-Rn
Zz
If
If
Am
Ej
Cn
Cr Cb-Jr
Vb
Ad
An
Ik
ItAt
Ag
24
23
25
25
24 29
30
29
31 33
35
36 36 36
36
35
40 39
39 40
39
38
38
33
40
40
24
26
29
29
41 4040
40
41
38
38
39
36
38
38
38
37
38
39
36
34
31 35
35
30
29
31
38
37
36
33
35
20 25 30 35 40Moho Depth(km)
44˚ 46˚ 48˚
-24˚
-22˚
-20˚
A
B
C
Archean
Proterozoic
Morondava basin
Volcanic
Br
Bg-Rn
Bg-Rn
Zz
If
If
Am
Ej
Cn
Cr Cb-Jr
Vb
Ad
An
Ik
ItAt
Ag
1.85
1.82 1.80 1.82
1.81 1.80
1.78
1.73
1.76
1.81
1.78
1.72 1.75
1.76
1.78
1.82
1.69
1.75
1.75
1.76
1.76
1.73
1.73
1.98
1.70
1.81
1.97
1.82
1.81
1.67
1.781.74
1.76
1.71
1.761.79
1.751.76
1.69
1.71
1.77
1.81
1.72
1.69
1.71
1.76
1.78
1.77
1.76
1.94
1.80
1.96
1.69
1.69
1.72
1.82
1.68 1.72 1.76 1.80 1.84 1.88Vp/Vs ratio
Case study: SELASOMA profile (Southern Madagascar)
44 45 46 47 48
−500
0
500
1000
1500
Ele
vati
on (
m)
−100
−50
0
50
100
Bouguer
anom
aly
(m
Gal)
Morondava basin Proterozoic Archean Volcanic
Elevation
Bouguer Anomaly
computed Bouguer Anomaly
44 45 46 47 48Longitude
15
20
25
30
35
40
Moho D
epth
(km
)
Cenozoic-CretaceousCarboniferious-Jurassic
Adro
yen
Anosyen BR
SZ
Ikala
mavony
AntananarivoCretaceousVolcanic
1.6
1.7
1.8
1.9
2.0
2.1
Vp/V
s
Moho
Vp/Vs
44 45 46 47 48Longitude
0
10
20
30
40
50
60
Depth
(km
)
Sediment Upper crust
Middle crust
Lower crust
Mantle
2.0
2.4
2.8
3.2
3.6
4.0
4.4
4.8
vs(
km/s
)
Contributions of crustallayers to bulk VP/VS
Interpretation
The recovered VP/VS represents an whole-crust average - it generally cannot be used to directlyidentify lithologies but needs to take into account the varying crustal layers but can inform onthe dominant bulk composition
NB The VP/VS ratio measurement is generally less reliable than Moho depth measurement
Case study: SELASOMA profile (Southern Madagascar)
44 45 46 47 48
−500
0
500
1000
1500
Ele
vati
on (
m)
−100
−50
0
50
100
Bouguer
anom
aly
(m
Gal)
Morondava basin Proterozoic Archean Volcanic
Elevation
Bouguer Anomaly
computed Bouguer Anomaly
44 45 46 47 48Longitude
15
20
25
30
35
40
Moho D
epth
(km
)
Cenozoic-CretaceousCarboniferious-Jurassic
Adro
yen
Anosyen BR
SZ
Ikala
mavony
AntananarivoCretaceousVolcanic
1.6
1.7
1.8
1.9
2.0
2.1
Vp/V
s
Moho
Vp/Vs
44 45 46 47 48Longitude
0
10
20
30
40
50
60
Depth
(km
)
Sediment Upper crust
Middle crust
Lower crust
Mantle
2.0
2.4
2.8
3.2
3.6
4.0
4.4
4.8
vs(
km/s
)
Contributions of crustallayers to bulk VP/VS
Interpretation
The recovered VP/VS represents an whole-crust average - it generally cannot be used to directlyidentify lithologies but needs to take into account the varying crustal layers but can inform onthe dominant bulk composition
NB The VP/VS ratio measurement is generally less reliable than Moho depth measurement
Case study: SELASOMA profile (Southern Madagascar)
44 45 46 47 48
−500
0
500
1000
1500
Ele
vati
on (
m)
−100
−50
0
50
100
Bouguer
anom
aly
(m
Gal)
Morondava basin Proterozoic Archean Volcanic
Elevation
Bouguer Anomaly
computed Bouguer Anomaly
44 45 46 47 48Longitude
15
20
25
30
35
40
Moho D
epth
(km
)
Cenozoic-CretaceousCarboniferious-Jurassic
Adro
yen
Anosyen BR
SZ
Ikala
mavony
AntananarivoCretaceousVolcanic
1.6
1.7
1.8
1.9
2.0
2.1
Vp/V
s
Moho
Vp/Vs
44 45 46 47 48Longitude
0
10
20
30
40
50
60
Depth
(km
)
Sediment Upper crust
Middle crust
Lower crust
Mantle
2.0
2.4
2.8
3.2
3.6
4.0
4.4
4.8
vs(
km/s
)
Contributions of crustallayers to bulk VP/VS
Interpretation
The recovered VP/VS represents an whole-crust average - it generally cannot be used to directlyidentify lithologies but needs to take into account the varying crustal layers but can inform onthe dominant bulk composition
NB The VP/VS ratio measurement is generally less reliable than Moho depth measurement
Outline
1 Introduction
2 Crustal Vp/Vs ratio and Moho depth from stacking of multiples
3 Common Conversion Point Stacks (CCCP)
Tilmann (GFZ,FUB) Receiver Functions 28 / 0
CCCP - Basic principle
Geometry for P-RF S-RF
Assume conversions occur at subhorizontal interfaces
Write seismogram amplitudes into bins along expected ray path of converted phase
⇒ Ignores diffraction (equiv. to assuming specular reflection in reflection problems)
Tilmann (GFZ,FUB) Receiver Functions 29 / 0
Synthetic example – One-layer crust (Movie)Single earthquake
⇒
kpcrst
8
−1.0−0.8−0.6−0.4−0.20.0 0.2 0.4 0.6 0.8 1.0
Amplitude
0
100
200
Dep
th (k
m)
0 100 200
Distance (km)
MultiplesDiffraction
Tilmann (GFZ,FUB) Receiver Functions 30 / 0
Case study: SELASOMA profile Southern MadagascarP-RF - Rindraharisaona et al. [2017]
Colors represent discontinuities, not absolute velocities (here: red: sudden increase withdepth)
In PS stack, multiples cause artifacts at larger depths, in PpPs and PpSs stack, artifactsfrom direct conversion at shallower depths
CCCP stacks of multiples can give additional confidence
Tilmann (GFZ,FUB) Receiver Functions 31 / 0
P-RF vs S-RFExample: Himalaya-Tibet continental collision
P-RF (high frequency):
Nabelek et al. [2009]
Detailed images of the crustal interfaces and Moho. Deeperstructure obscured by multiples and not shown.
S-RF (low-frequency):
Zhao et al. [2010]
Only low-resolution Moho, nointernal crustal structure. Imagesmantle discontinuties, e.g. LAB(lithosphere-asthenosphereboundary)
Note that the cross-sections are not exactly from the same profile.
Tilmann (GFZ,FUB) Receiver Functions 32 / 0
P-RF vs S-RFExample: Himalaya-Tibet continental collision
P-RF (high frequency):
Nabelek et al. [2009]
Detailed images of the crustal interfaces and Moho. Deeperstructure obscured by multiples and not shown.
S-RF (low-frequency):
Zhao et al. [2010]
Only low-resolution Moho, nointernal crustal structure. Imagesmantle discontinuties, e.g. LAB(lithosphere-asthenosphereboundary)
Note that the cross-sections are not exactly from the same profile.
Tilmann (GFZ,FUB) Receiver Functions 32 / 0
P-RF vs S-RF direct comparisonExample: North Chile subduction zone
Time section P-RF vs S-RF
Sodoudi et al. [2011]
For reference: P-RF CCCP
Tilmann (GFZ,FUB) Receiver Functions 33 / 0
What can go wrong?Dipping interfaces [Schneider et al., 2013]
PRF - Pamir continental subduction
Because the assumption of conversion athorizontal interfaces, dipping structure areshown at too shallow dip (significant effect fordips>∼ 30◦)
CCCP stacking under assumption of dippinginterface images the steeply dipping structurecorrectly but will show horizontal interfaces(here the flat Moho) more shallow than inreality.
Tilmann (GFZ,FUB) Receiver Functions 34 / 0
Dipping interfacesSynthetic tests [Schneider et al., 2013]
Tilmann (GFZ,FUB) Receiver Functions 35 / 0
What can go wrong?Dipping interfaces [Schneider et al., 2013]
PRF - Pamir continental subduction
Because the assumption of conversion athorizontal interfaces, dipping structure areshown at too shallow dip (significant effect fordips>∼ 30◦)
CCCP stacking under assumption of dippinginterface images the steeply dipping structurecorrectly but will show horizontal interfaces(here the flat Moho) more shallow than inreality.
Tilmann (GFZ,FUB) Receiver Functions 36 / 0
A more general approach: migration
Make no assumptions on smoothness or geometry of discontinuites/anomalies.Trace seismic energy to all possible locations where conversion might havehappened. ⇒ Constructive interference at real conversion points, destructiveinterference at all others.
[Rondenay, 2009]
Tilmann (GFZ,FUB) Receiver Functions 37 / 0
A more general approach: migrationComparing (horizontal) CCCP to migration
6 in-plane incident waves at different slownesses[Rondenay, 2009]Tilmann (GFZ,FUB) Receiver Functions 38 / 0
A more general approach: migration
However:
Requires very good information on velocitymodel:Correct velocity
Ryberg&Weber (2000)
Get frowns & smiles
Require a high density of receivers:
Sodoudi et al. [2011]
Otherwise: more smiles!
Tilmann (GFZ,FUB) Receiver Functions 39 / 0
A more general approach: migration
However:
Requires very good information on velocitymodel:Too slow:
Ryberg&Weber (2000)
Get frowns & smiles
Require a high density of receivers:
Sodoudi et al. [2011]
Otherwise: more smiles!
Tilmann (GFZ,FUB) Receiver Functions 39 / 0
A more general approach: migration
However:
Requires very good information on velocitymodel:Too fast:
Ryberg&Weber (2000)
Get frowns & smiles
Require a high density of receivers:
Sodoudi et al. [2011]
Otherwise: more smiles!
Tilmann (GFZ,FUB) Receiver Functions 39 / 0
A more general approach: migration
However:
Requires very good information on velocitymodel:Too fast:
Ryberg&Weber (2000)
Get frowns & smiles
Require a high density of receivers:
Sodoudi et al. [2011]
Otherwise: more smiles!
Tilmann (GFZ,FUB) Receiver Functions 39 / 0
A more general approach: migration
However:
Requires very good information on velocitymodel:Too fast:
Ryberg&Weber (2000)
Get frowns & smiles
Require a high density of receivers:
Sodoudi et al. [2011]
Otherwise: more smiles!
Tilmann (GFZ,FUB) Receiver Functions 39 / 0
A more general approach: migration
However:
Requires very good information on velocitymodel:Too fast:
Ryberg&Weber (2000)
Get frowns & smiles
Require a high density of receivers:
Sodoudi et al. [2011]Otherwise: more smiles!
Tilmann (GFZ,FUB) Receiver Functions 39 / 0
Effect of gradual transitions and frequency dependence
Kind et al. [2012]
At long periods Moho peak shifted because of interference with intra-crustal (Conrad)discontinuity
Transitional discontinuity (here LAB) disappears
In addition to 1D effects shown, 2D - 3D effects can influence frequency dependence
Tilmann (GFZ,FUB) Receiver Functions 40 / 0
Not covered because of lack of time:
Waveform inversion (1D, 2D, 3D)
P - RF CCCP - discontinuities shown
Japan [Kawakatsu et al., 2009]
Inversion - actual velocity diffwrt reference shown
Cascadia [Rondenay, 2009]
Deconvolution methods and sidelobes
Tilmann (GFZ,FUB) Receiver Functions 41 / 0
Summary (of sorts)
P receiver functions - high-to-medium frequency Crustal and slab imaging ⇒ CCCPimaging/migration with direct conversion, maybe supplementary use of multiplesAverage crustal velocities ⇒ H − κ-stackingDetailed crustal velocity model ⇒ 1D waveform inversion (in combination with surfacewave dispersion from ambient noise)‘Slab’ velocity model: ⇒ 2D, 3D waveform inversion
P receiver functions - medium-to-low frequency Target: Mantle transition zone (410, 660discontinuites) ⇒ Station or CCCP stacking with direct conversions)
S receiver functions - low frequency Mantle lithosphere imaging(lithosphere-asthenosphere-boundary (LAB) - mid-lithopheric discontinuities (MLD)
Pay attention to:
Artifacts from in- and out-of plane diffractions, processing
Separation of primaries from multiples
RF-methods sensitive to discontinuities not absolute velocities - subject to velocity-depthtradeoff!
References I
R. C. Aster, B. Borchers, and C. H. Thurber. Parameter estimation and inverse problems. Elsevier, 1st edition,2005.
H. Kawakatsu, P. Kumar, Y. Takei, M. Shinohara, T. Kanazawa, E. Araki, and K. Suyehiro. Seismic evidencefor sharp lithosphere-asthenosphere boundaries of oceanic plates. Science, 324:499–502, 2009. doi:10.1126/science.1169499.
B. Kennett. The seismic wavefield, volume II: Interpretation of seismograms on regional and global scales.Cambridge University Press, 2002.
R. Kind and H. Yuan. Encyclopedia of Solid Earth Geophysics, chapter Seismic Receiver Function Technique,pages 1258–1269. Springer, 2011.
R. Kind, X. H. Yuan, and P. Kumar. Seismic receiver functions and the lithosphere-asthenosphere boundary.Tectonophysics, 536:25–43, April 2012. doi: 10.1016/j.tecto.2012.03.005.
C.A. Langston. Structure Under Mount Rainier, Washington, Inferred From Teleseismic Body Waves. J.Geophys.Res., 84:4749–4762, 1979.
D. Lombardi, Braunmiller J., E. Kissling, and D. Giardini. Moho depth and Poisson’s ratio in theWestern?Central Alps from receiver functions. Geophys. J. Int., 2008. doi:10.1111/j.1365-246X.2007.03706.x.
J. Nabelek, Hetenyi G., J. Vergne, S. Sapkota, B. Kafle, M. Jiang, H. Su, J. Chen, B. S. Huang, andHi-Climb-Team. Underplating in the Himalayan-Tibet collision zone revealed by the Hi-CLIMB experiment.Science, 325:1371–1374, 2009. doi: 10.1126/science.1167719.
E. J. Rindraharisaona, F. Tilmann, X. Yuan, G. Rumpker, J. Giese, G. Rambolamanana, and G. Barruol.Crustal structure of southern Madagascar from receiver functions and ambient noise correlation:Implications for crustal evolution. J. Geophys. Res., 122(2):1179–1197, 2017. doi:10.1002/2016jb013565.
Tilmann (GFZ,FUB) Receiver Functions 43 / 0
References II
S. Rondenay. Upper mantle imaging with array recordings of converted and scattered teleseismic waves. Surv.Geophys., 30:377–405, 2009. doi: 10.1007/s10712-009-9071-5.
F. M. Schneider, X. Yuan, B. Schurr, J. Mechie, C. Sippl, C. Haberland, V. Minaev, I. Oimahmadov,M. Gadoev, N. Radjabov, U. Abdybachaev, S. Orunbaev, and S. Negmatullaev. Seismic imaging ofsubducting continental lower crust beneath the pamir. Earth Planet. Sci. Let., 375:101–112, 2013. doi:10.1016/j.epsl.2013.05.015.
F. Sodoudi, X. Yuan, G. Asch, and R. Kind. High?resolution image of the geometry and thickness of thesubducting nazca lithosphere beneath northern Chile. J. Geophys. Res., 116:B04302, 2011. doi:10.1029/2010JB007829.
TRANSALP Working Group. First deep seismic reflection images of the eastern alps reveal giant crustalwedges and transcrustal ramps. Geophys. Res. Let., 29, 2002. doi: 10.1029/2002GL014911.
L. P. Vinnik. Detection of waves converted from P to SV in the mantle. Phys. Earth Planet. Int., 15:39–45,1977.
I. Wolbern and G. Rumpker. Limitations of h − κ stacking: ambiguous results caused by crustal layering. J.Seismol., 21:221–235, 2017. doi: 10.1007/s10950-016-9599-z.
Junmeng Zhao, Xiaohui Yuan, Hongbing Liu, Prakash Kumar, Shunping Pei, Rainer Kind, Zhongjie Zhang,Jiwen Teng, Lin Ding, Xing Gao, Qiang Xu, and Wei Wang. The boundary between the Indian and Asiantectonic plates below Tibet. Proc. Nat. Acad. Sci., 107(25):11229–11233, 2010. doi:10.1073/pnas.1001921107.
L. Zhu. Crustal structure across the San Andreas Fault, Southern California from teleseismic converted waves.Earth Planet. Sci. Let., 179:183–190, 2000.
L. Zhu and H. Kanamori. Moho depth variation in southern California from teleseismic receiver functions. J.Geophys. Res., 105(B2):2969–2980, 2000.
Tilmann (GFZ,FUB) Receiver Functions 44 / 0