Etude numérique et empirique de l'atténuation haute fréquence (Kappa)

Présentation de Céline Gélis et al. (IRSN) au 2nd Workshop Résif "Aléa sismique & Shakemaps" | 29 - 31 janvier 2018, Montpellier

Transcript

1. Etude numérique et empirique de l’atténuation haute fréquence (Kappa) Projet « Kappa » (appel d’offre RESIF-RAP 2017) C. Gélis, L. Provost, B. Froment, F. Tchawe Nziaha (IRSN) F. Bonilla, Ph. Guéguen (IFSTTAR) M. Calvet, L. Margerin (OMP) F. Gatti, F. Lopez-Caballero, S. Touhami, M. Colvez (Centrale Supélec) F. Courboulex, T. Monfret (GEOAZUR) E. Bertrand, Ph. Langlaude (CEREMA) J. Mayor (EDF)
2. Etude numérique et empirique de l’atténuation haute fréquence (Kappa) – Journées RESIF – 29-31 janvier 2018, Montpellier 2/9 Atténuation du mouvement sismique ( ) ) /( 0 , Vs Q fr ef e A f r A π − ⋅ = Qef = facteur de qualité effectif, dépend de f Attenuation: sc in ef Q Q Q 1 1 1 + = Campbell (2009) ef Q 1 iews on the selection of the pre-event, P- S-wave portions of the record and on the of fE and fX , which can lead to some vari- j between analysts. We found that s in picking of the pre-event, P-wave and ortions did not significantly affect the js -automatic procedure to choose the inter- approach we took to quantify the subjectivity and precision of the obtained js. In the absence of the high-frequency decay quantified here by j Fourier amplitude spectra should be flat above the corner frequency, fc , of the source. When fitting the best-fit lines to determine j it is necessary that fE (the frequency chosen as the start of the best-fit line) is greater than fc otherwise the j 45 55 65 75 85 Time (s) -1.0 -0.5 0.0 0.5 1.0 Acceleration (cm/s2) (a) 1996.197.00.12.45.4060.RA.OGSI.00.ENE.D.SAC 0 10 20 30 40 50 Frequency (Hz) 10-6 10-5 10-4 10-3 10-2 10-1 100 Acceleration Spectra (cm/s) 3 5 3 5 3 5 3 5 3 5 3 5 Noise S-wave (b) 0 10 20 30 40 50 Frequency (Hz) 10-4 10-3 10-2 10-1 100 3 5 3 5 3 5 3 5 = 0.027 s (c) f E f X Figure 2 direct shear-wave and noise spectra computed from a record that shows a clear high-frequency linear trend. Also shown are the d to estimate the pre-event noise and the direct shear-wave spectra (black parts of acceleration time-history) and the frequencies fE and fX chosen by one of the analysts (the other analysts chose similar fE and fX for records such as this) J. Douglas et al. Pure Appl. Geophys. κo m Anderson & Hough (1984), Hough et al. (1988) e r m⋅ + = 0 κ κ Douglas et al. (2010) ( ) X E f f f f e A f A < < ⋅ = − , 0 κ π different views on the selection of the pre-event, P- wave and S-wave portions of the record and on the selection of fE and fX , which can lead to some vari- ations in j between analysts. We found that differences in picking of the pre-event, P-wave and S-wave portions did not significantly affect the js obtained. A semi-automatic procedure to choose the inter- vals used to compute the direct shear-wave spectra and noise spectra was also applied. Since both P- and approach we took to quantify the subjectivity and precision of the obtained js. In the absence of the high-frequency decay quantified here by j Fourier amplitude spectra should be flat above the corner frequency, fc , of the source When fitting the best-fit lines to determine j it is necessary that fE (the frequency chosen as the start o the best-fit line) is greater than fc otherwise the j estimates can be biased. When using strong-motion data from moderate and large earthquakes (Mw C 5.5 45 55 65 75 85 Time (s) -1.0 -0.5 0.0 Acceleratio 0 10 20 30 40 50 Frequency (Hz) 10-6 10-5 10-4 10-3 10-2 10-1 100 Acceleration Spectra (cm/s) 3 5 3 5 3 5 3 5 3 5 3 5 Noise S-wave (b) 0 10 20 30 40 50 Frequency (Hz) 10-4 10-3 10-2 10-1 100 3 5 3 5 3 5 3 5 = 0.027 s (c) f E f X Figure 2 Example of direct shear-wave and noise spectra computed from a record that shows a clear high-frequency linear trend. Also shown are the intervals used to estimate the pre-event noise and the direct shear-wave spectra (black parts of acceleration time-history) and the frequencies fE and fX chosen by one of the analysts (the other analysts chose similar fE and fX for records such as this) Anderson & Hough (1984) Douglas et al. (2010) À r :
3. Etude numérique et empirique de l’atténuation haute fréquence (Kappa) – Journées RESIF – 29-31 janvier 2018, Montpellier 3/9 Utilisation de Kappa GMPEs Zones de sismicité faible à modérée Zones actives Kappa ì  Aléa sismique (« host-to-target ») Campbell, 2003 ; Cotton et al., 2006 ì  Prédiction du mouvement sismique (effet de site) Boore, 2003 zontal propagation of S waves through the showed that the spectral decay of the loga- urier acceleration spectrum with frequency, square source spectrum, is flat (i.e., κ ˆ 0) and Qef ∝ f and is negative (i.e., κ > 0) and Qef ∝ fη…η < 1†. However, only when stant) is the spectral decay described exactly Fitting equation (5) to a model with a frac- dependence of Qef will yield a smaller value el in which Qef is assumed to be constant, es the importance of the standard assumption have adopted it as their high-frequency filter parame revision of the specific barrier model of the ea source (Papageorgiou and Aki, 1983) because of fit to strong-motion data. However, these latter auth tinue to suggest that it could be a source paramet than a site parameter. In the quarter-wavelength method, the site amp of the Fourier amplitude spectrum of acceleration is c from the equation (Boore, 2003) Amp…f† ˆ …ρ S β S = ρ  β†1=2 exp… πκ 0 f†; Campbell (2009) Extraits USGS, M>5
4. Contenu du projet RESIF-RAP « Kappa » Approche empirique ▌ Données (françaises RESIF) ▌ Calcul de Kappa et Q Approche numérique ▌ Milieu connu (Q) ▌ Tests de sensibilité 4/9 Etude numérique et empirique de l’atténuation haute fréquence (Kappa) – Journées RESIF – 29-31 janvier 2018, Montpellier Application au contexte français (Nice) ▌ Vers l’estimation du mouvement pour un site spécifique Participants : C. Gélis, L. Provost, B. Froment, F. Tchawe Nziaha (IRSN) F. Bonilla, Ph. Guéguen (IFSTTAR) M. Calvet, L. Margerin (OMP) F. Gatti, F. Lopez-Caballero, S. Touhami, M. Colvez (Centrale Supélec) F. Courboulex, T. Monfret (GEOAZUR) E. Bertrand, Ph. Langlaude (CEREMA) J. Mayor (EDF) Objectifs: Publications communes (1 résumé soumis à SSA 2018)
5. Etude numérique et empirique de l’atténuation haute fréquence (Kappa) – Journées RESIF – 29-31 janvier 2018, Montpellier 5/9 Contenu du projet – approche empirique ì  Estimation de Kappa sur quelques stations sismologiques françaises à l’aide des données les plus récentes (et autres stations) A non-automatic procedure for estimating j was dopted because we noted that the frequency, fE , at which the acceleration spectral amplitudes show a estimates for j. Similarly, due to varying signal-to- noise ratios (visually inspected), fX shows large variations and therefore it was not possible to use a -5˚ 0˚ 5˚ 10˚ 45˚ 50˚ 1 2 4˚ 5˚ 6˚ 7˚ 8˚ 9˚ 41˚ 42˚ 43˚ 44˚ 45˚ 46˚ 47˚ 48˚ 49˚ 0 50 100 km Besancon Colmar Epinal Nancy Strasbourg Lons-le-Saunier Macon Bourg-en-Bresse Lyon St.-Etienne Valence Privas Gap Grenoble Avignon Nimes Le-Puy-en-Velay Montpellier Marseille Toulon Digne-les-Bains Nice (1) -2˚ -1˚ 0˚ 1˚ 2˚ 3˚ 42˚ 43˚ 0 50 100 km Biarritz Pau Lourdes St-Gaudens Pamplona Huesca Gerona Narbonne Beziers Carcassonne Perpignan (2) Figure 1 arthquake (circles) and station (triangles) locations and travel paths (lines) of the records used for this study. 1 Alps and Co ˆte d’Azur (southern part of map) and 2 Pyrenees Vol. 167, (2010) A j Model for Mainland France 1307 Douglas, Bonilla, Gehl et Gélis (2010) 0 50 100 150 200 250 300 350 400 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 Soil (s) Standard: 0 =0.0350 s, m =0.000156 s/km Weighted: 0 =0.0347 s, m =0.000161 s/km Alps 0 50 100 150 200 250 300 350 400 Distance (km) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 Rock (s) Standard: 0 =0.0268 s, m =0.000156 s/km Weighted: 0 =0.0254 s, m =0.000161 s/km 0 50 100 150 200 250 300 350 400 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 Standard: 0 =0.029 s, m =0.000204 s/km Weighted: 0 =0.029 s, m =0.000205 s/km Côte d’Azur 0 50 100 150 200 250 300 350 400 Distance (km) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 Standard: 0 =0.025 s, m =0.000204 s/km Weighted: 0 =0.024 s, m =0.000205 s/km 0 50 100 150 200 250 300 350 400 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 Standard: 0 =0.024 s, m =0.000153 s/km Weighted: 0 =0.025 s, m =0.000152 s/km Pyrenees 0 50 100 150 200 250 300 350 400 Distance (km) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 Standard: 0 =0.017 s, m =0.000153 s/km Weighted: 0 =0.018 s, m =0.000152 s/km Figure 6 Distance dependence of j values for three regions in mainland France. The top plots present the results for stations located on soil. The bottom plots show the results for stations located on rock 0.06 0.08 0.10 0.12 0.14 0.16 0.18 (s) Unc. standard: 0 =0.023 s, m =0.000125 s/km Unc. weighted: 0 =0.022 s, m =0.000126 s/km Con. standard: 0 =0.0174 s, m =0.000156 s/km Con. weighted: 0 =0.0164 s, m =0.000161 s/km OGAN (Rock) 0.06 0.08 0.10 0.12 0.14 0.16 0.18 Unc. standard: 0 =0.022 s, m =0.000251 s/km Unc. weighted: 0 =0.023 s, m =0.000246 s/km Con. standard: 0 =0.0362 s, m =0.000156 s/km Con. weighted: 0 =0.0354 s, m =0.000161 s/km OGMO (Rock) 1312 J. Douglas et al. Pure Appl. Geophys.
6. Etude numérique et empirique de l’atténuation haute fréquence (Kappa) – Journées RESIF – 29-31 janvier 2018, Montpellier 6/9 Contenu du projet – approche empirique ì  Vers une comparaison avec l’estimation de l’atténuation (Qi et Qsc) aux mêmes stations Mayor, Traversa, Calvet, Margerin (2017) Exemple à Taiwan : Margerin, Gillet, Planès, Calvet, Hung (2017) ▌ Comparaison avec carte existante (Qi) ▌ Station par station : Qi et Qsc
7. Etude numérique et empirique de l’atténuation haute fréquence (Kappa) – Journées RESIF – 29-31 janvier 2018, Montpellier Vs=2000 m/s ; ν = 0.25 ; ρ = 2000 kg/m3 Qs = Vs/20 (=100) ; Qp = 2*Qs (=200) 7/9 Contenu du projet – approche numérique ì  Exploration de la physique liée à Kappa 20 km 5 km Gélis et Bonilla, en cours ▌ Simulations 2D (préliminaires) ▌ Vers des simulations 3D (Gatti, Lopez-Caballero)
8. Etude numérique et empirique de l’atténuation haute fréquence (Kappa) – Journées RESIF – 29-31 janvier 2018, Montpellier 8/9 Contenu du projet - application ì  Application « host-to-target » à Nice Provost, Gélis, Bonilla, en cours GMPE au rocher κGMPE Rocher du site κrocher site Site dans bassin Host-to-target Al Atik et al. (2014) Fonction d’amplification ▌ Résultats préliminaires
9. Etude numérique et empirique de l’atténuation haute fréquence (Kappa) – Journées RESIF – 29-31 janvier 2018, Montpellier 9/9 Etat des lieux ì  Projet d’un an (début 1er juillet 2017), soumis en réponse à un appel d’offre dans le cadre du GIS-RAP, possibilité de prolongation (sans budget supplémentaire) ì  6 000 € budget (+ 4 650 € autres ressources) ì Cofinancement M2 ì Ressources informatiques ì Réunions ì  2 réunions en visio depuis début projet (autres planifiées) ì  Différents objectifs et approches autour d’une même thématique dans un groupe avec des compétences et connaissances très complémentaires ì  Utilisation des données du réseau sismologique français ì  Comment pérenniser ce groupe ?
10. Etude numérique et empirique de l’atténuation haute fréquence (Kappa) – Journées RESIF – 29-31 janvier 2018, Montpellier 10/20
11. Etude numérique et empirique de l’atténuation haute fréquence (Kappa) – Journées RESIF – 29-31 janvier 2018, Montpellier 11/20