EPOS use case: High-resolution seismic tomography of Italy Giuseppe Fiameni with the collaboration of Alberto Michelini, Federica Magnoni, Emanuele Casarotti.

EPOS use case:High-resolution seismic tomography

of Italy

Giuseppe Fiameni with the collaboration of Alberto Michelini, Federica Magnoni, Emanuele

Casarotti

Outline• Objective• Motivation• EPOS’s use case• Requirements• Actual status• Proposal

Objective• High resolution tomographic velocity

model of Italy

(i.e., invert the recorded seismic waveform data of earthquakes in Italy and adjacent regions for the 3D, laterally heterogeneous velocity structure of Italy.)

3

Motivation1)Big scientific challenge: waveform

tomography of te 3D seismic velocity structure of the Italian region up to frequencies of 0.5 Hz (i.e., maximum spatial resolution ~2 Km) has never been feasible before

2)Very relevant implications for modeling at high resolution strong ground shaking (large earthquakes) scenarios with evident seismic engineering applications

3)The use case workflow conjugates data storage retrieval, staging, HPC modeling and stage-out of large volumes of simulated data, detailed 3D velocity model and adjoint kernels

4

EPOS WorkflowEPOS WorkflowObserved

Data

ForwardModeling

InverseModeling

Visualization

InputParameters

Forward ModelingForward Modeling

velocity modelsource modeltopography/

geology

Synthetic 3D wavefield

simulationsRUN

2009 L’Aquila earthquake simulation for 3D velocity model and finite fault

Inverse ModelingInverse Modeling

Iteratively improve the

velocity model

velocity modelsource model

topography/geologyTOMOGRAPHY

Misfit between synthetics and dataobserved data

Inverse ModelingInverse ModelingForward

Modeling

Inverse

ModelingVisuali

zation

Updated velocity model db(binary & vtk ASCII)

Iterative improvement of the tomographic model

Forward SimulationsForward Simulations

Forward

Modeling

FORWARDFORWARDSIMULATIONSSIMULATIONS

ADJOINTADJOINTSIMULATIONSSIMULATIONS

WINDOW WINDOW SELECTIONSELECTION

MEASUREMENTSMEASUREMENTSADJOINT SOURCESADJOINT SOURCES

KERNEL SUMMATION and KERNEL SUMMATION and REGULARIZATIONREGULARIZATION

STEP LENGTH TESTSTEP LENGTH TEST

42 h+

+

+

+

0.5/31 h

84 h

3 h

169/231 h

C. Italy256 cores63 events

TOTAL TIME 1 ITERATIONTOTAL TIME 1 ITERATION~ 13/16

days

4 days+

+

+

+

0.5/75 h

8 days

3 h

16/22 days

~ 28/37days

Italy2300 cores150 events

Wall Time on 256 CPUs Cluster

Inverse ModelingInverse ModelingForward

Modeling

Inverse

ModelingVisuali

zation

Updated velocity model db(binary & vtk ASCII)

Iterative improvement of the tomographic model

Forward SimulationsForward Simulations

Forward

Modeling

FORWARDFORWARDSIMULATIONSSIMULATIONS

ADJOINTADJOINTSIMULATIONSSIMULATIONS

WINDOW WINDOW SELECTIONSELECTION

MEASUREMENTSMEASUREMENTSADJOINT SOURCESADJOINT SOURCES

KERNEL SUMMATION and KERNEL SUMMATION and REGULARIZATIONREGULARIZATION

STEP LENGTH TESTSTEP LENGTH TEST

60 GB+

+

+

+

15 GB

70 GB

7 GB

240 GB

TOT STORAGE 1 ITERATIONTOT STORAGE 1 ITERATION 392 GB

+

+

+

+

4 TB

2400 GB

144 GB

825 GB

30 GB

600 GB

StorageC. Italy256 cores63 events

Italy2300 cores150 events

VisualizatiVisualizationon

ObservedDataForwa

rdModeli

ng

Inverse

ModelingVisuali

zation

InputParameters

Seismograms (SAC, GMT, ObsPy, Google Earth)

Presently using ‘manual’ scripts Parallel environment

Mesh - Velocity Models (Paraview)

Movies (Paraview) Kernels (Paraview)