Transcript bozdag_ireland_last - Seismic wave Propagation and Imaging in

Seismic wave Propagation and Imaging in
Complex media: a European network
EBRU BOZDAĞ
Early Stage Researcher
Host Institution: Utrecht University
Place of Origin: Istanbul, Turkey
Appointment Time: February 2005
Project: Testing and Improving Tomographic Models Using
Numerical 3D Wave Propagation
Task Groups: TG Planetary Scale
Cooperation: Oxford University
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
SPICE
Crustal corrections predicted by
ray theory and finite frequency theory
compared to measured time shifts
from SEM seismograms using Crust2.0
Ebru Bozdağ
Jeannot Trampert
Utrecht University, Utrecht, the Netherlands
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
Crustal corrections are important in
surface wave tomography
 Crustal structures have a strong effect on
surface waves.
 Inverting for crust and mantle is difficult.
Therefore crustal corrections are preferred.
 Phase delays from the crust are removed in
surface wave tomography to identify mantle
structure.
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
The objective
Investigate how far
 great circle approximation
 exact ray theory and
 finite frequency theory
predict crustal corrections using the SEM
(Komatitsch and Tromp, 2002) seismograms
computed in 1D (PREM) and 3D
(PREM+Crust2.0) earth models.
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
Data generation
Earthquakes & Stations
VIC
SIR
NCN
SEP
11 earthquakes
253 stations
IJR
CMA
CRT
SOA
XZNG
IJR
WIA
Synthetic seismograms using SEM
 Synthetic seismograms from 1D earth model PREM
(Dziewonski and Anderson, 1981)
 3D crustal model Crust2.0 (Bassin et al., 2000) is
superimposed on top of PREM model and synthetic
seismograms are computed for PREM+Crust2.0 model
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
Measuring time shifts as a function of frequency
PREM
PREM+Crust2.0
Time-variable filter
to extract the
fundamental mode
Rayleigh wave
Phase correction to
PREM seismograms
Cross-correlation
of the seismograms
d
dt 
ω
PREM+Crust2.0 - PREM
PREM+Crust2.0 – PREM+Corr.
Measure the phase d 
of the cross-correlation
Unwrap the phase
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
Methods used for crustal corrections
Great Circle Approximation (GCA)

d  

 dc
c0
c
0
d
0
Exact Ray Theory (ERT)


dc
d   
ds
c0 ray c0
2

1
Finite Frequency Theory (FFT)

dc
d    K ( , ) ( , )dd
c0
0 0
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
Calculation of local phase
velocity perturbations
 At each grid point of Crust2.0, we
superimpose the crustal model (plus
topography) onto PREM and solve for the
exact eigenfrequencies in that 1D earth
model.
 We thus generate exact local phase
velocities at each grid point which are used
to calculate crustal corrections along rays.
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
A spherical harmonic expansion of the
local phase velocities is used to simulate the
smoothing of Crust2.0 in SEM
Rayleigh, 40s
Smoothing with
spherical harmonic expansion
Without smoothing
dc/c0
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
Ray paths showing the time shifts
computed for 150 s using GCA for one earthquake
Time shifts from
PREM+Crust2.0 – PREM
seismograms
time shift (s)
Time shifts
after correction
(PREM+Crust2.0 - PREM+GCA)
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
Time shifts as a
function of distance
calculated for 150 s
using GCA (l=40)
Black lines: average
uncertainties of the
measurements of
Trampert & Woodhouse
Blue bars: before correction
(PREM+Crust2.0 - PREM)
Red bars: after correction
(PREM+Crust2.0 – PREM+GCA)
90%
dt=±5.9 s
76%
dt=±7.2 s
81%
dt=±10.6 s
89%
dt=±19.4 s
92%
dt=±22.7 s
96%
dt=±27.8 s
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
Time shifts as a
function of distance
calculated for 40 s
using GCA (l=40)
Black lines: average
uncertainties of the
measurements of
Trampert & Woodhouse
Blue bars: before correction
(PREM+Crust2.0 - PREM)
Red bars: after correction
(PREM+Crust2.0 – PREM+GCA)
59%
dt=±4.8 s
55%
dt=±7.2 s
69%
dt=±11.2 s
50%
dt=±17.8 s
62%
dt=±20.8 s
78%
dt=±24.8 s
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
Oceans-continents
150 s, GCA, l=40
Blue bars: before correction
(PREM+Crust2.0 - PREM)
Red bars: after correction
(PREM+Crust2.0 - PREM+GCA)
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
Oceans-continents
40 s, GCA, l=40
Blue bars: before correction
(PREM+Crust2.0 - PREM)
Red bars: after correction
(PREM+Crust2.0 - PREM+GCA)
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
Comparison of methods (150 s, l=40)
GCA vs. FFT
GCA vs. FFT (major arc)
GCA vs. ERT
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
FFT vs. ERT
www.spice-rtn.org
Comparison of methods (40 s, l=40)
GCA vs. FFT
GCA vs. FFT (major arc)
GCA vs. ERT
FFT vs. ERT
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org
Conclusions
No pronounced difference between
GCA, ERT or FFT corrections
Corrections at 150 sec are better
than at 40 sec
Corrections get worse as distance
increases
We will now check if the imperfect
corrections will lead to a detectable
mantle signal
SPICE Research and Training Workshop III, July 22-28, Kinsale, Ireland
www.spice-rtn.org