Sismovalp final meeting, Martigny, 02/10/2006 SISMOVALP WP6 Alpine ground motion Numerical benchmark of ground motion simulation in the Grenoble valley.

Sismovalp final meeting, Martigny, 02/10/2006

SISMOVALP WP6

Alpine ground motion

Numerical benchmark of ground motion simulation in the Grenoble

valleyDescription and first results


ESG2006 Numerical benchmark


Numerical Benchmark of 3D ground motion simulation in the Grenoble

Valley : Description and First Results

Emmanuel Chaljub, Seiji Tsuno, François Thouvenot, Michel Dietrich, Pierre-Yves

Bard

+ many colleagues & predictors


Overview• Introduction

• Background Data– Tectonic context / Seismicity (F. Thouvenot)– Geophysical and geotechnical data (M. Dietrich)

• General description and first results (E. Chaljub)– Description of the benchmark : model and events– Participation / Methods– Outline of the Grenoble basin seismic response

• Detailed comparisons (S. Tsuno)– Addressed topics and comparison methods– Example results– Preliminary Conclusions

• Recap / Lessons / Benchmark Future


Proposed predictions"Imposed"

• 2 weak motion events– W1, W2

• 2 strong motion events– S1, S2– Extension of W1, W2 to

stronger events– Source : imposed geometry

and kinematics

• + "Free-style"– M6 (S1) : Estimate +

uncertainties• (NL, source variabilities)


Material provided to participants : topographies

Surface : 50 m gridstepBedrock / sediment

interface: 250 m gridstep


Material provided to participants : velocity model

• Basin

– No shallow structure– Qs = 50

• Bedrock

– Vsurf= 3200 m/s– Qs = ∞

Stiff sediments but very hard bedrock : strong impedance contrast


Material provided to participants: source parameters


Material provided to participants : recordings

anchor deterministic calculations

empirical Green's functions

Borehole, at depth

Borehole, surface


Material NOT PROVIDED

• Input motion for 1D and 2D calculations


Results requested from participants : 1D case

• Time series (at least 30 s) of ground velocity at 1 station (borehole surface)


Results requested from participants : 2D methods

• Time series (at least 30 s) of ground velocity along a cross-section (10 stations, through borehole)


Results requested from participants : 3D case

• Time series (at least 30 s) of ground velocity for 40 stations)

• + PGV map


Results requested from participants : EMP methods

• Time series (at least 30 s) of ground velocity for 3 stations (including borehole)


Results received


Results received : details

1818

ParticipationParticipationParticipationParticipationContribution to Numerical benchmark

Frequency band analyzed

0.1 1 10 40

ID18ID09ID08ID17ID16ID15ID14ID10ID07ID13ID12ID11ID06ID05ID04ID03ID02ID01

1D

2D

EM

3D

Method S1 S2 W1 W2 FS NL Fmin (Hz) Fmax (Hz) No. time his.1D ○ ○ ○ 0.05 10 8

1D ○ ○ ○ 0.3 50 120

2D ○ ○ ○ 8 65

2D ○ ○ ○ ○ 8.5 77

2D ○ ○ ○ ○ 0.3 50 1200

2D ○ ○ 10 44

EM ○ ○ 0.8 40 72

EM ○ ○ 10 234

EM ○ 1 40 12

3F ○ ○ ○ ○ 1 480

3F ○ ○ ○ ○ 2 492

3F ○ ○ ○ ○ 0.5 480

3F ○ ○ ○ ○ 0.2 2 480

3F ○ ○ ○ ○ 0.48 480

3F ○ 2.5 120

3T ○ ○ 0.1 2 240

3T ○ 2.5 117

3T ○ ○ ○ ○ 2 492

Total = 5213

Frequency (Hz)

1919

Addressed topicsAddressed topicsAddressed topicsAddressed topics1. Overall variability (S1 case at Borehole site)1. Overall variability (S1 case at Borehole site)

Each method; 1D, 2D, 3D Flat, 3D Topo., EGF-> Waveform, Input motion-> Fourier spectra, Spectral ratio-> Non-linear (1D, 2D) & Topography effects (3T)

2. Comparison method2. Comparison method-> Waveform, Maximum values (PGV)-> Fourier spectra, Spectral ratio, Resonance peak

3. Applying misfit criteria to 3D estimations (S1 case)3. Applying misfit criteria to 3D estimations (S1 case)-> Anderson’s method-> Kristekova’s method

4. Additional learnings4. Additional learnings-> from S2-> from W1/W2, match with observation ?

2020

Overall raw variability Overall raw variability –– Time Time domaindomain

Overall raw variability Overall raw variability –– Time Time domaindomain

Velocity Waveform at bore-hole site (OGFB & OGFH)

Input (GL -541.5m) Surface (GL)

Time (sec)

Vel

ocity

(m

/s)

- E

W a

nd

SH

(1&

2D

)

OGFH (GL) 1D(Linear) 1D(NL) 2D(Linear) 2D(NL) EM 3F 3T

0 5 10 15 20 25 30

-2

0

2

Time (sec)

Vel

ocity

(m

/s)

- E

W a

nd

SH

(1&

2D

)

OGFB (GL -541.5m) 1D(Linear) 1D(NL) 2D(Linear) 2D(NL) EM 3F 3T

0 5 10 15 20 25 30

-2

0

2

1D

2D

EM

3F

3T

1D

2D

EM

3F

3T

2121

Comparison Comparison -- Spectra SpectraComparison Comparison -- Spectra Spectra

Fourier spectra at borehole site (OGFB & OGFH)

Frequency (Hz)Fou

rier sp

ectra

(Vel

.) EW and SH (1&2D)

3F 3T

1D 2D EM

OGFH (GL)

0.1 1 1050.50.001

0.01

0.1

1

10

Frequency (Hz)

Fou

rier sp

ectra

(Vel

.) EW and SH (1&2D)

3F 3T

1D 2D EM

OGFB (GL -541.5m)

0.1 1 1050.50.001

0.01

0.1

1

10

Two groups

2222

Comparison Comparison -- Spectral ratio Spectral ratioComparison Comparison -- Spectral ratio Spectral ratio

Spectral ratio at bore-hole site

Frequency (Hz)

Spe

ctra

l rat

io

SH

1D

0.1 1 1050.50.1

1

10

100

Frequency (Hz)

Spe

ctra

l rat

io

EW

EM

0.1 1 1050.50.1

1

10

100

Frequency (Hz)

Spe

ctra

l rat

io

EW

3F 3T

0.1 1 1050.50.1

1

10

100

1D 2D

EM3D

1D, 2D, EM, 3D and 3T

Frequency (Hz)

Spe

ctra

l rat

ioSH

2D

0.1 1 1050.50.1

1

10

100

Frequency (Hz)S

pect

ral r

atio

EW and SH (1&2D)

3F 3T

1D 2D EM

0.1 1 1050.50.1

1

10

100

2525

3D Flat3D Flat Simulation Simulation- case S1- case S13D Flat3D Flat Simulation Simulation- case S1- case S1

Bedrock depth of Grenoble basin

OGFH

OGFB

Time (sec)

Velo

city

(m/s

)

EW

ID15 ID16 ID17

ID07 ID10 ID14

OGFB(GL -541.5m)

0 5 10 15 20 25 30

-0.5

0

0.5

Time (sec)

Velo

city

(m/s

) EW

ID15 ID16 ID17

ID07 ID10 ID14

OGFH (GL)

0 5 10 15 20 25 30

-0.5

0

0.5

N

2626

Spectra - 3D Flat & case S1Spectra - 3D Flat & case S1Spectra - 3D Flat & case S1Spectra - 3D Flat & case S1

Frequency (Hz)

Fourier sp

ect

ra (V

el.) EW

ID15 ID16 ID17

ID07 ID10 ID14

0.1 1 20.50.001

0.01

0.1

1

10

Frequency (Hz)

Fourier sp

ect

ra (V

el.) NS

ID15 ID16 ID17

ID07 ID10 ID14

0.1 1 20.50.001

0.01

0.1

1

10

Frequency (Hz)

Fourier sp

ect

ra (V

el.) UD

ID15 ID16 ID17

ID07 ID10 ID14

0.1 1 20.50.001

0.01

0.1

1

10

Frequency (Hz)

Fourier sp

ect

ra (V

el.) EW

ID15 ID16 ID17

ID07 ID10 ID14

0.1 1 20.50.001

0.01

0.1

1

10

Frequency (Hz)

Fourier sp

ect

ra (V

el.) NS

ID15 ID16 ID17

ID07 ID10 ID14

0.1 1 20.50.001

0.01

0.1

1

10

Frequency (Hz)

Fourier sp

ect

ra (V

el.) UD

ID15 ID16 ID17

ID07 ID10 ID14

0.1 1 20.50.001

0.01

0.1

1

10

Fourier Spectra at OGFB (Gl -541.5m)

Fourier Spectra at OGFH (GL)

2727

Spectral ratio Spectral ratio -- 3D Flat & case S1 3D Flat & case S1Spectral ratio Spectral ratio -- 3D Flat & case S1 3D Flat & case S1

Frequency (Hz)

Spec

tral r

atio

EW

ID15 ID16 ID17

ID07 ID10 ID14

0.1 1 20.50.1

1

10

100

Spectral ratio at borehole site

Frequency (Hz)S

pec

tral r

atio

NS

ID15 ID16 ID17

ID07 ID10 ID14

0.1 1 20.50.1

1

10

100

2929Spectral ratio at borehole site

Spectral ratio Spectral ratio -- 3D Topo. & case S1 3D Topo. & case S1Spectral ratio Spectral ratio -- 3D Topo. & case S1 3D Topo. & case S1

Frequency (Hz)

Spec

tral r

atio

EW

ID08 ID09 ID18

0.1 1 20.50.1

1

10

100

Frequency (Hz)S

pec

tral r

atio

NS

ID08 ID09 ID18

0.1 1 20.50.1

1

10

100


Example results : PGV map


S1 case : 3D, Flat predictions

3232

ConclusionConclusionConclusionConclusion

Predicted waveforms and Fourier spectra exhibit a exhibit a large (too large, huge, unacceptable ?) variability.

Peak ground velocities on surface also exhibit a significant variability due to the differences in investigated frequency ranges.

BUT

Spectral ratios at bore-hole site are relatively stable, specially for the resonance frequency.

Differences in “Input motion” (i.e., source modelling) have the largest influence for the numerical simulations in this benchmark test.


Main learnings 1

• Difficulties of the exercise– Data and model format : to be standardized– HF / Short wavelength structure– Utility / absolute necessity of imposed exercises

• Simpler cases

– Consistency 1D / 2D / 3D : input motion• Simple for incident plane waves• ? Including source ?

– EGF : bad S/N ratio at low frequencies– Timing too tight

• Plan more time for predictions start earlier !– 2 years / 18 months / 9 months


Main learnings 2

• Outcomes– Large variability in input motion, significantly less

in site response– Consistency of 2D – 3D modelling (amplification

levels / frequency)– (Very) Encouraging results for different 3D models– Topography effects– Use of misfit criteria

• Definitely useful for objectively quantifying similarity / dissimilarity, but not yet enough practice

– Anderson's criteria more robust and engineering –oriented– Kristekova's : for already rather similar signals

• Need for looking at time histories, spectra, and spectral ratios


Example comparison after iteration (and bug correction)


Main learning

• 3D modelling is not yet "press-button"– Too fast applications may yield very wrong results

(and large untrust from end-users)

• BUT very similar results are possible event with completely different numerical schemes– (probably indicative of the "exact" solution)

• Conditions for careful use– well-validated techniques & codes– Well trained users– Careful model implementation– External review


Future of the Grenoble basin benchmark• Short term

– Summary report for Proceedings Volume 2• Not all results for all receivers• Focusing on some specific aspects

• Intermediate term– ? Second round for converging ?

• First results VERY ENCOURAGING ? Sismovalp extension ?– Other aspects

• Other receivers• Wavefield (array)• NL effects (2D)

– SPICE (2?)– ? Part of a collection of real sites for benchmarking

• Long term– Shallow structure and HF / BB response : deterministic or

stochastic ?


Acknowledgments


Acknowledgements

F. Anselmetti, C. Beck, C. Bordes, M. Campillo, E. Chapron, J. Converset,

C. Cornou, F. Cotton, M. De Batist, P. Finckh, E. Flavigny, P. Foray,

J.F Gamond, S. Garambois, J.R. Grasso, J.P Gratier, P. Guéguen, R. Guiguet,

S. Hatton, L. Jenatton, J. Jerram, S. Labanieh, B. Lebrun, F. Lemeille,

G. Menard, O. Méric, G. Nicoud, S. Roussel, P. Roux, L. Stehly,

S. Tadenuma, M. Vallon , P. Van Rensbergen,J. Verbeke, C. Voisin


ESG2006 Contributions relative to alpine valleys

• Predictions : Sxx

• Characteristics of Grenoble Valley

• Similar cases : Alpine Valleys


Prediction PostersS01 Numerical Simulation of Wave Propagation in the Grenoble Basin

H. Aochi, J. Rey and J. DouglasS02 1D and 2D linear and nonlinear site response in the Grenoble area

L.F. Bonilla, S. Nielsen and P.C. LiuS03 A ground-motion simulation approach coupling rock groundmotion prediction equations

and the empirical Green's functions methodM. Causse, F. Cotton and C. Cornou

S04 Spectral element modeling of 3D wave propagation in the alpine valley of Grenoble, France

E.Chaljub S05 Numerical Benchmark of 3D Ground Motion Simulation in the Valley of Grenoble, French

AlpsH. J. Chiang, T. M. Chang and K. L. Wen

S06 Site effects in a deep alpine valley for various seismic sourcesN. Delépine and J.-F. Semblat

S08 An Efficient Ader-Dg Method for 3-Dimensional Seismic Wave Propagation in Media With Complex Geometry

M. Käser, M. Dumbser and J. De La PuenteS09 Ground Motion Simulation of two Moderate Size Earthquakes in the Grenoble Area using

Summation of Small EarthquakesC. Kohrs-Sansorny, F. Courboulex and A. Deschamps

S10 Ground motion simulation on a 2D profile across the Grenoble basin using the Aki-Larner discrete wave-number method

C. Lacave and F. HollenderS12 Modeling of strong earthquake motion in the Grenoble Valley, French Alps

P.Moczo et al.S14 Kinematic composite souce model combined with EGF for modeling strong ground

motion Application to the Grenoble BasinJ. Ruiz, D. Baumont, P. Bernard and C. Berge- Thierry

S15 3D Ground Motion Simulation of the Grenoble valley by GeoELSEMarco Stupazzini

S30 Kinematic modeling of strong earthquake motion in the Grenoble Valley, French AlpsP. Franek and F. Gallovic


Grenoble ValleyS16 Seismicity of the Grenoble Area

F. Thouvenot, L. Jenatton and R. GuiguetS17 High-amplitude reflections in proglacial lacustrine basin fills of the NW

Alps : Origin and Paleoenvironment ImplicationsE. Chapron, M. Dietrich, C. Beck, P. Van Rensbergen, G. Menard, P. Finckh, G.

Nicoud, F. Lemeille, F. Anselmetti and M. De BatistS18 Seismic profiling and borehole measurements in the Isère valley near

Grenoble, France: 1 data acquisition and processingM .Dietrich, C. Cornou, G. Ménard, F. Lemeille, F. Guyoton and R. Guiguet

S19 Seismic profiling and borehole measurements in the Isère valley near Grenoble, France 2 InterpretationG. Ménard, M.Dietrich, M. Vallon , S. Tadenuma, C. Bordes, O. Méric and F.

Lemeille 126S23 Measurement and variability study of site effects in the 3D glacial

valley of Grenoble, French AlpsE. Chaljub, C. Cornou, J. Verbeke, J. Converset, C. Voisin, L. Stehly, J.R. Grasso, P.

Guéguen, S. Roussel, P. Roux, S. Hatton and M. CampilloS24 Characterising the non linearities of the lacustrine clays in the

Grenoble basinJ. Jerram, P. Foray, E. Flavigny and S. Labanieh

S25 Geotechnical, geophysical and seismological data used for the estimate of the highest amplified frequency in the basin of GrenobleP. Guéguen, S. Garambois, S. Tadenuma, B. Lebrun, and F. Cotton


Alpine Valleys

S26 Sites effects in the Vallorcine valleyC. Voisin, P. Guéguen, J.R. Grasso, C. Gomes

S27 Modelling of strong ground in the July 2004, Mw 5.2 Bovec earthquakeM. Vanini, M. Villani, E. Faccioli and A. Gosar

S28 Seismic Response Analysis of La Salle Fluvial Fan (Valle D' Aosta Italy)C. Turino, G. Ferretti, C. Eva, C. Gauzzi and R. Paolucci

Sismovalp final meeting, Martigny, 02/10/2006 SISMOVALP WP6 Alpine ground motion Numerical benchmark of ground motion simulation in the Grenoble valley.

Documents