Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow,...

15
Synthetic Seismograms Jeroen Tromp Synthetic Seismograms – p.1

Transcript of Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow,...

Page 1: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

Synthetic SeismogramsJeroen Tromp

Synthetic Seismograms – p.1

Page 2: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

Semi-Analytical Seismology

Global Seismology: Spherical models

� Normal-mode summation

� Reflectivity method (Earth flattening)

� WKBJ theory (Asymptotic)

Regional Seismology: layercake models

� Discrete wavenumber/Reflectivity method

Synthetic Seismograms – p.2

Page 3: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

Perturbation Methods

Tomographic inversions:

� Ray methods

� Path-average approximation

� Non-linear asymptotic coupling theory

� mode splitting/coupling

Source inversions:

� 1D models with path corrections

Synthetic Seismograms – p.3

Page 4: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

The Challenge of the Globe

� A slow, thin, highly variable crust

� Sharp radial discontinuities

� Fluid-solid boundaries

� Anisotropy

� Attenuation

� Ellipticity, topography & bathymetry

� Rotation

� Self-gravitation

Synthetic Seismograms – p.4

Page 5: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

Computational Seismology

Calculation of 3D synthetic seismograms:

� Modern numerical methods

� Hardware advances

Synthetic Seismograms – p.5

Page 6: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

The Seismo Lab Beowulf

� 312 processors

� 156 GB of RAM

� 733 MHz Pentium III� 133 MHz Tyan boards

� CISCO switch

� Linux OS

� MPICH communications

Synthetic Seismograms – p.6

Page 7: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

Topography & Crustal Model

Elevation from sea level (km)

-8 +8 10 30 39 50 65 75

Crustal thickness (km)

Bassin et al. (2000)Synthetic Seismograms – p.7

Page 8: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

Mantle Model

Shear velocity variation (%)

-6% +6%

Ritsema, Van Heijst & Woodhouse (1999)Synthetic Seismograms – p.8

Page 9: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

Hjörleifsdóttir, Kanamori & Tromp (2002)

Synthetic Seismograms – p.9

Page 10: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

The Earth Simulator

� 640 nodes

� 16 GB of memory pernode, 10 TB ofdistributed memory

� 8 processors pernode, 5120 total

Collaboration with Seiji Tsuboi (IFREE)

Synthetic Seismograms – p.10

Page 11: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

Earth Simulator

243 nodes (1944 processors), 3.4 TB memory, 14.8 billion degrees of freedom,5 teraflops

Soon: 512 nodes (4096 processors)

Synthetic Seismograms – p.11

Page 12: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

November 3, 2002, Denali

Tsuboi, Komatitsch, Chen & Tromp, 2003

Synthetic Seismograms – p.12

Page 13: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

November 3, 2002, Denali

PREM 3D

Tsuboi, Komatitsch, Chen & Tromp, 2003Synthetic Seismograms – p.13

Page 14: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

The World’s Fastest Computer

10

100

1000

10000

100000

1000000Ju

n− 93

Nov

− 94

Jun− 9

6Nov

− 97

Jun− 9

9Nov

− 00

Jun− 0

2Nov

− 03

Jun− 0

5Nov

− 06

Jun− 0

8Nov

− 09

Per

form

ance

(G

igaf

lops)

1 Teraflop/s

1 Petaflop/s

Courtesy T. Sterling (CACR, Caltech) Synthetic Seismograms – p.14

Page 15: Synthetic Seismograms - CIDER · Synthetic Seismograms Œ p.3. The Challenge of the Globe A slow, thin, highly variable crust Sharp radial discontinuities Fluid-solid boundaries Anisotropy

Conclusions

� 3D global synthetics accurate up to4 seconds (1D mode synthetics are typicallyaccurate up to 8 seconds)

� Tool to probe/assess 3D structures

� 1D & 3D reference synthetics (need REMand CMT)

� 3D (finite) source inversions

Synthetic Seismograms – p.15