OBSERVATIONS AND MODELING OF EXTREME SUBDUCTION EARTHQUAKES

Transcription

1 October -7, 8, eijing, China OSERVTIONS ND MODELING OF EXTREME SUDUCTION ERTHQUKES M. Chave,, E. Cabrera 3, N. Perea Institute of Engineering, UNM, C.U., 45, Meico DF, Meico Laboratoire de Géologie CNRS-ENS, 4 Rue Lhomond, Paris, France 3 DGSC, UNM, C.U., 45, Meico DF, Meico STRCT The 9/9/985 a Ms 8. thrust mechanism subduction superficial (TSS earthquake occurred in the Meican Pacific ocean coast with an epicentral distance of 38 km from Meico Cit (MC. The estimates of the human and economical losses related to this large event, were of about 3, people and more than 6 billion US dollars, the largest share of these were in MC. The largest, Ms 8., instrumentall observed TSS earthquake in Meico, occurred in the Colima-Jalisco region the 3/6/93, with epicentral distance of about km from Guadalajara (G in northwestern Meico. The estimated frequenc of occurrence, in Meico, of these tpes of events varies from decades to about ears and their upper Ms bound are still under discussion. appling a hbrid method, previousl validated for the 9/9/985 and for the 9//995 Ms 7.6 Colima-Jalisco earthquakes, we generated broadband snthetic accelerograms, epected in MC stiff and compressive soils, and in G sand soils, associated to etreme seismic scenarios for TSS Mw 8.5 earthquakes. ased on an acceptable risk criteria, in order to minimie the probabilit of eceedance associated to an epected economical loss of 6. illion US dollars, for the proposed Mw 8.5 MC earthquake scenario, a compressible soil design spectra Sa(5% ordinate of.4g (g=98 cm/s, seems appropriate (for the 9/9/985 earthquake the observed Sama(5% was equal to about g. From the information of the construction stock of G, the economical loss, associated to the Mw 8.5 earthquake scenario analed amounts to 9 billion US$ dollars. KEYWORDS: Observations, hbrid snthetics, subduction, etreme earthquakes, economics.. INTRODUCTION The 9/9/985 a Ms 8. thrust mechanism subduction superficial (TSS earthquake occurred in the Meican Pacific ocean coast with an epicentral distance of 38 km from Meico Cit (MC. The estimates of the human and economical losses related to this large event, were of about 3, people and about 6 billion US dollars, the largest share of these were in MC. The largest, Ms 8., instrumentall observed TSS earthquake in Meico, occurred in the Colima-Jalisco region the 3/6/93, with epicentral distance of about km from Guadalajara (G in northwestern Meico. The estimated frequenc of occurrence, in Meico, of these tpes of events varies from decades to about ears and their upper Ms bound are still under discussion. The importance of obtaining reliable estimates of the seismic haard and of its economical consequences, associated with the possible occurrence of etreme earthquakes in those regions of Meico are the objectives of this work. appling a hbrid method, previousl validated for the 9/9/985 and for the 9//995 Ms 7.6 Colima- Jalisco earthquakes (Chave et al. 4 we generated broadband snthetic accelerograms, epected in MC stiff and compressive soils, and in G sand soils, associated to etreme seismic scenarios for TSS Mw 8.5 earthquakes. The economic impact of the scenarios is also analed. The paper is divided in 6 parts. In the nd we briefl discuss the main seismotectonic features of the Meican subduction one, a snthesis of the hbrid broadband modeling procedure used to generate snthetic accelerograms is presented in part 3, the comparison of the recordings obtained for the 9/9/985 Ms 8. Michoacan and the 9//995 Ms 7.6 Colima-Jalisco earthquakes are presented in the 4 th, the results of the modeling for the Mw 8.5 scenario earthquakes, as well as of their economic impacts on MC and G are presented in the 5 th part. Finall the main conclusions of the work are presented.

2 October -7, 8, eijing, China. SEISMOTECTONIC FETURES OF THE MEXICN SUDUCTION ZONE The seismotectonic of the Meican subduction region is associated with the dnamics of the Cocos, Rivera, Caribean and Northamerican plates (Fig.. The Cocos and the Rivera plates are subducting the Northamerican one and had generated the most recent, largest surficial thrust earthquakes in the region, such as the Jalisco 3/6/93 Ms 8., the Ms 7.6 (Mw 8 Colima-Jalisco 9//995 and the 9/9/985 Ms 8.(Mw 8. Michoacan earthquakes (Fig.. The geometr and some geophsical properties of the mentioned interacting plates are shown in Figs. and F, respectivel. The rupture area and the kinematic slip distribution of the Michoacan earthquake is shown in Fig. C; and the rupture area and the location of the four subevents of the Colima-Jalisco earthquake are presented in Fig. E. (SS D P (SD (SC MEXICO km 8 km 995 MC 6 km SS Ms 8. GDL Jalisco Colima..5 N N 36 E 6 5 km E 5 km P GDL COL Ciudad Guman (CG Depth (km Guerrero Coast Popocatepetl Vp Vs Den. km/skm/sg/cm MC SD Mainshocks 3.9 CS ftershocks 3. SD Mainshocks Distance from the trench, km C HIPOCENTER L strike km L dip km S slip km F Depth km (M W Trench Coast Colima fthershock of /9/995 event - Local network and teleseismic data 9/8/993 event Distance from the trench ais, Km S4 S3 S S Fig.. Outer rectangle is the surface projection of the 564 km 3 earth crust volume 3D discretiation, inner rectangle is the rupture area of the 9/9/985 Ms 8. earthquake; profile P-P ; C Kinematic slip distribution of the rupture of the 985 earthquake (Mod. Mendoa and Hartell, 989; D Location of the Colima Jalisco region; E surface projection of the 6398 km 3 earth crust volume 3D discretiation, inner rectangle (69 km rupture area of the 995 earthquake, the 4 smaller rectangles represent the four (S to S4 subevents of the 995 earthquake (Escobedo et al, 998; F Profile from the coast to Guadalajara which includes the densities of the laers. (and et al., RODND MODELING PROCEDURE hbrid procedure (Fig., combining long period and high frequenc simulations (Chave et al., 4 was utilied for the computation of broadband snthetics accelerograms for the Ms 8. (Mw 8. Michoacan 985 and the Ms 7.6 (Mw 8 Jalisco-Colima 995 earthquakes, as well as for etreme Mw 8.5 earthquake scenarios for Meico Cit and Guadalajara. The long period (<.3 H wave field was simulated using a recentl optimied 3D seismic wave propagation parallel finite difference code, that uses nd order operators in time and 4th order differences in space (Cabrera and Chave et al, 7. The high frequenc (.3H snthetics were generated with the Empirical Green function (EGF method, Irikura (986. In this method the ground motion of a large event is epressed as the superposition of the records of small events (elementar sources. Finall, the low and high frequenc snthetics are combined using matched filters (Chave et al., 4.

3 October -7, 8, eijing, China SIMULTION Y th 4 th ORDER 3D 3D FINITEDIFFERENCE METHOD FREQUENCY RNGE (f<.3 <.3 H h SIMULTION Y EMPIRICL GREEN FUNCTION METHOD FREQUENCY RNGE f.3 (f.3 H h v t v t = ρ = ρ t v = ρ ( + f ( + f ( + f MTCHED PIR OF FILTERS SYNTHESIS OF RODND STRONG GROUND MOTION Y THE COMINTION OF THE LOW ND HIGH FREQUENCY SYNTHETICS FREQUENCY RNGE (.<f<5 f <5 H h Fig. Hbrid procedure combining long period and high frequenc simulations (Chave et al., 4 4. OSERVTIONS ND MODELING OF THE 9/9/985 Ms 8. MICHOCN ND THE 9//995 Ms 7.6 COLIM-JLISCO ERTHQUKES The broadband hbrid modeling procedure snthesied in Fig. 3, was utilied to generate snthetic accelerograms to compare them with the ground motions recorded at stations SCT (located on a 4 m laer of compressible soils of MC and Colegio (located on a 35 m laer of sand soils overling rock in Guadalajara for the Ms 8. Michoacan and the Ms 7.6 Colima-Jalisco earthquakes, respectivel. For the low frequenc modeling of the 985 Michoacan event, a 5 6 4(depth km 3 volume was used, spatial and temporal discretiations of.5 km and.s, respectivel, and the geophsical parameters shown in Fig. (Cabrera and Chave et al. 7 were utilied. For the high frequencies modeling of the 985 event, recordings of the 4/3/997 Ms 7.3, /9/985 Ms 7.6 and 4/9/995 Ms 7.4 subduction earthquakes were utilied (Chave et al, in preparation. For the low frequenc modeling of the 995 Colima-Jalisco earthquake, a volume of (depth km 3 was utilied, and for the high frequenc, the recordings of their 6//995 Ms 5.8 and //995 Ms 5.9 for and after shocks were used (Chave et al. 4. In Figs. 3 and 4, the snthetic accelerograms and their associated Fourier amplitude spectra for the SCT site N-S direction 985 Michoacan earthquake and the Colegio (Guadalajara site W-E 995 Colima-Jalisco earthquake, respectivel, are compared with their observations. Notice that the agreements between them are satisfactor in both cases, including the local soil effects at.5 and. H for stations SCT and Colegio, in MC and G, respectivel. 5 SCT 9/9/85, NE ma=94. cceleration (cm/sec Time (sec Snthetic ma= Fourier mplitude SCT 9/9/85 NE Snthetic... Frequenc (h Fig. 3 Observed and snthetic accelerograms and their corresponding Fourier amplitude spectra in the N- S direction at station SCT (Meico Cit for the 9/9/985 Michoacan earthquake.

4 October -7, 8, eijing, China cceleration (cm/sec Observed 9//95 ma=8.59 Station COL (surface Snthetic 9//95 ma= Time (sec. 9//95 Observed Snthetic... Frequenc (H Fig. 4 Observed and snthetic accelerograms and their corresponding Fourier amplitude spectra in the W- E direction at station COL (Guadalajara for the 9//95 Colima-Jalisco earthquake. Fourier mplitude Station COL (surface 5. MODELING OF Mw 8.5 SUDUCTION SURFICIL EXTREME ERTHQUKE SCENRIOS FOR MEXICO CITY ND GUDLJR ND THEIR ECONOMICL CONSEQUENCES. The importance of obtaining reliable estimates of the seismic haard and of its economical consequences, associated with the posible occurrence of etreme earthquakes in regions of Meico such as the Colima Jalisco were in 93 a Ms 8. earthquake occurred (Fig. and Guerrero where a Ms 8. + is epected (in the socalled Guerero seismic gap, could be hardl overstated, particularl if the Ms 8., 985 Michoacan earthquake (Fig. destructive effects on MC are taken into account (about 3, deaths and more than 6 billion dollars loss. ased on the satisfactor results obtained for the modeling of the 985 Mw 8. and the 995 Mw 8 events (Figs. 3 and 4, the hbrid method described in 3 was also utilied to generate snthetic accelerograms for the ground motions epected in MC and G for thrust subduction surficial (TSS Mw 8.5 etreme earthquake scenarios. For the computation of MC snthetics, in Fig. 5 the surface projection of the 5 6 4(depth km 3 volume used for the low frequenc modeling of the Mw 8.5 scenario is presented, the spatial and temporal discretiations as well as the geophsical parameters are the ones mentioned in part 4. The snthetic fractal slip distributions considered for the scenario ruptures are shown in Fig. 5. For the high frequenc modeling, the recordings mentioned above for the 985 Ms 8. event were used, as well as the observations of the later. n eample of the tpe of results obtained is shown in Figs. 6. Fig 6 shows MC distribution of the estimated maimum horiontal ground acceleration (ma and spectral acceleration Sa(5% at s for the Ms earthquake and Fig. 6 shows the corresponding distribution for the Ms 8.5 scenario earthquake. Notice that for the SCT site, for the scenario earthquake, ma is equal to 74 cm/s versus 68 cm/s and for Sa(5% 39 cm/s versus 973 cm/s. lso notice, that the Meico Cit ones were Sa(5% 7 cm/s increases considerable with respect to the ones estimated from the 985 earthquake observations. The economical impact of the considered scenarios, analed b an acceptable risk criteria, is shown in Fig 7. Notice in this figure, that the previous 985 earthquake (976 and actual (4 seismic recommendations of the MC Construction Code, as well as the 985 observation, correspond to a Sa(5% of 4, 45 and 96 cm/s, respectivel, and that the are in the so called unacceptable risk one for the consequence cost of 6. US billion dollars (estimated for the 985 earthquake. The Sa(5% of 4 cm/s (obtained from statistics of the etreme earthquake scenarios considered, Chave et al., in preparation is in the so called transition risk one for the same abscise. For the modeling of the G snthetics, in Fig. 8, the surface projection of the (depth km 3 volume used for the low frequenc simulation of the Mw 8.5 earthquake scenario is presented (with a rupture area which includes large parts of the 93 and the 995 earthquakes, Fig., the discretiations used to model the 995 earthquake were also utilied. The distribution of the subevents considered to represent the Mw 8.5 (each one with a seismic moment = Mo of the scenario earthquake / is also included in this figure. For the high frequenc modeling, the recordings of the 995 Ms 7.6 (Mw 8 event were used. efore showing eamples of the modeling results obtained, let us briefl discuss about the population, geotechnical and construction characteristics of G shown in Fig. 8 (Chave et al 4. a It has about 6 million inhabitants; b 5%, 8%, and 47% of its construction stock (about 3Km were built more than 4, between 5 to 4 and less than 5 ears ago, respectivel; c From that construction stock onl 4% has been built on firm soils or rock sites,

5 October -7, 8, eijing, China therefore, from the observations of the 9//995 Mw 8, about 86% of that stock is on sand soils in which the local site effects are guaranteed, as the observed at COL (Fig. 4; d The actual construction code of Guadalajara does not recognie the seismic site effects. 6 km P 9 km 5 km 5 km P (SS (SD (SC SS Mw 8.5 Scenario Fig. 5 Surface projection of the 3D volume (5 6 4 km utilied in the low frequenc modeling of the Mw 8.5 TSS earthquake scenarios. Fractal slip simulations (based on Mai and eroa, used in the hbrid modeling of the Mw=8.5 Subduction Surficial earthquake Meico Cit scenarios. The rectangles in model are an eample of the finite sources used for the high frequenc Empirical Green Functions simulations. rea Number of km constructions , , , , , ,965 SCT ma=68. cm/s Sa =973 cm/s ma cm/s = gal SCT ma=73.95 cm/s Sa ma =39 cm/s cm/s= gal Fig. 6 Meico Cit distribution of the estimated horiontal spectral acceleration Sa (damping 5% at s for the Ms earthquake and the corresponding distribution for the Ms 8.5 scenario earthquake. The respective ma and Sa(5% values for the SCT site are also included. MC built area (km and the number of constructions of one to three floors are also listed.

6 October -7, 8, eijing, China PROILITY OF EXCEEDNCE OF Sa(g CCEPTLE RISK SIMOLOGY Construction Code 976 Construction Code 4 Earthq. Obsv. 985 m+σ m+3σ CCEPTLE RISK ZONE TRNSITION ZONE Sa (g MRGINLLY CCEPTED RISK UNCCEPTLE RISK ZONE (9/9/985, cost 6.. m. m m m m b b MONETRY VLUE OF CONSEQUENCE COST $(U.S. Fig. 7 Probabilit of eceedance for a time period of 5 ears of several Sa(g values, g=9.8 m/sec, for the Mw 8.5 scenario vs Consequence cost in $ US, of the 9/9/985 Ms 8. Michoacan earthquake. Rivera Plate 88 km S S9 S S5 S S8 S6 S4 S7 S3 Guadalajara COL North merica Plate C. Tpe Good qualit Inter. qualit C ad qualit E < 5 ears 5 E 4 E 3 > 4 C. ge E3 E E E3 E E E3 E E GDL REC. SITES.- COL.- RC 3.- ROT 4.- OL 5.- PL 6.- OR 7.- GR 8.- JR 9.- MIR.- RF.- TON S 3 km 3 Postulated earthqueke S Cocos Plate Subone IV H5 m Subone III 5H m Subone II H5 m km Subone I H5 m Fig. 8 Surface projection of the km 3 volume used in the low frequenc modeling, including the rupture area for the Mw=8.5 postulated earthquake. S-S depict the location of its assumed subevents. Distribution of Guadalajara s construction stock and its four geothechnical subones (H depth of rock under Guadalajara s sand soils. The encircled dots ( to identif the recording sites of the 9//995 Ms 7.6 main shock and its 6/9/995 and /9/995 for and after shocks.

7 October -7, 8, eijing, China cceleration (cm/sec 5-5 Postulated Ms=8.5 ma= COL (surface 9//95 Ms=7.4 5 ma=8.5 Mean M + S -5 M + 3 S rcos 9//95 WE Time (sec Period (sec Spectral cceleration (cm/sec Fig. 9 Observed and snthetic accelerograms for Colegio site in Guadalajara for the 9//995 Mw 8 (Ms 7.6 earthquake and the scenario Mw 8.5 etreme earthquake; cceleration response spectra statistics for subone III (Fig. 8 of Guadalajara for the etreme scenario earthquake. Table. Guadalajara s built surface stock, ij; Mean damage ratio, Dij; Epected built surface damage Sij ij (km (uilt surface stock i=,, C; j=e, E, E 3 SUZONE C. Tpe I H 5m II 5 H m III H 5m IV H > 5 Σ E E E 3 E E E 3 E E E 3 E E E C Σ H=depth to rock of soil; E 5 ears, 5 E 4 ears, E 3 > 4 ears; =Good qualit, =Intermediate qualit, C=ad qualit D ij (Mean damage ratio % SUZONE (IMM C. Tpe I (7. II (7.5 III (8.5 IV (9. E E E 3 E E E 3 E E E 3 E E E C S ij (km (Epected built surface damage = ij D ij SUZONE (IMM [(ma + 3σ cm/s ] C. Tpe I (7. [67] II (7.5 [83] III (8.5 [5] IV (9. [7] Σ E E E 3 E E E 3 E E E 3 E E E C Σ Table. Economical impact of the Mw 8.5 scenario earthquake in Guadalajara built stock C ij / km Cost 6 US$ C. Tpe I (7. [67] II (7.5 [83] III (8.5 [5] IV (9. [7] E E E 3 E E E 3 E E E 3 E E E C C ij = S ij C ij / km Total cost 6 US$ C. Tpe I (7. [67] II (7.5 [83] III (8.5 [5] IV (9. [7] Σ E E E 3 E E E 3 E E E 3 E E E ,6., ,475.6,65. 8, ,78.4, ,75.6 C , ,38.9 Σ ,3.7,476.,87.6 3,3. 4, ,66.7, ,87.6 The scenarios analed, included that the epicenter of the Mw 8.5 earthquake could occur at S, S and S7. n eample of the tpe of snthetic accelerograms obtained for the Colegio site (subone III of Fig. 8 is shown in Fig. 9, which also includes the observation of the 995 Mw 8 event. Notice that the ma = 96 cm/s for the

8 October -7, 8, eijing, China Mw 8.5 scenario, compared to 8.5 cm/s for the Mw event. In Fig. 9 the statistics of the Sa(5% for the subone III are shown. Similar graphs were obtained for the other subones of G. The economical impact of the etreme earthquake scenarios analed is snthesied in Tables and. Table includes the statistics of Guadalajara s built surface stock, ij; the seismic vulnerabilit of its construction stock, epressed b the mean damage ratio, Dij (as a function of the Mercalli Modified Intensit and the corresponding ma obtained from the statistics of the etreme earthquake scenarios, as the one shown in Fig. 9 for each geotechnical subone; the epected built surface damage Sij (equal to the product ij Dij associated to the Mw 8.5 scenario. The economical consequences of the etreme Mw 8.5 scenario considered, estimated as the product of the actual cost, Cij (of the different tpes of Guadalajara s ij b Sij, are presented in Table. Notice that its total amount is of about 9 illion US dollars. 6. CONCLUSIONS appling a hbrid method, previousl validated for the Ms 8. 9/9/985 and for the 9//995 Ms 7.6 Colima-Jalisco earthquakes, we generated broadband snthetic accelerograms, epected in Meico Cit stiff and compressive soils, and in Guadalajara sand soils, associated to etreme seismic scenarios for thrust subduction surficial Mw 8.5 earthquakes. ased on an acceptable risk criteria, in order to minimie the probabilit of eceedance associated to an epected economical loss of 6. billion US dollars (observed in 985, for the proposed Mw 8.5 earthquake scenario, a Sama(5% ordinate of.4g (g=98 cm/s, seems appropriate (for the 9/9/985 Ms 8. earthquake the observed Sama(5% was equal to about g. From the information of the construction stock of Guadalajara, the economical loss, associated to the possible occurrence of the Mw 8.5 earthquake scenario analed amounts to 9 billion US$ dollars. REFERENCES and W, V. Kostoglodov, H. Hurtado-Día, M. Mena, (999. Structure of the southern Jalisco subduction one Meico, as inferred from gravit and seismicit. Geofisica Int., 38, 3, Cabrera E., M. Chave, R. Madariaga, N. Perea, M. Frisenda, (7. 3D parallel elastodnamic modeling of large subduction earthquakes. F. Capello et al. (eds: Euro PVM/MPI 7, LNCS 4757, pp , Springer-Verlag erlin Heidelberg. Chave M., K. Olsen, E. Cabrera, (4. roadband modeling of strong ground motions for prediction purposes for subduction earthquakes occurring in the Colima-Jalisco region of Meico. 3WCEE, Vancouver,.C., Canada, ugust -6, Paper No 653. Escobedo D, J.F. Pacheco, G. Suare, (998. Teleseismic bod-wave analsis of the 9 october, 995 (Mw=8., Colima-Jalisco, Meico earthquake, and its largest foreshock and aftershock. Geophsical Research Letters, 5, 4, Mai M. and G.C. eroa, (. spatial random filed model to characterie compleit in earthquake slip. J. Geophsical Res., Vol 7, No, 38. Mendoa C, S. Hartell, (989. Slip distribution of the 9 september 985 Michoacan, Meico, earthquake: near source and telesismic constrains, ull. Seismol. Soc. mer., 79, pp Irikura K., (986. Prediction of strong acceleration motion using empirical Green s function. Proc. 7 th JEES, CKNOWLEDGMENTS We would like to thank the support of J.L. Gordillo, the Supercomputing staff and M. mbri and Leonardo lcantara, of DGSC, and the Institute of Engineering, UNM, respectivel.we also acknowledge the support of R. Madariaga and M. Mai of the ENS, Paris and ETH, Zurich, respectivel. We acknowledge DGSC, UNM for the support to use Kanalam, as well as the STFC Daresbur Laborator to use HECToR Supercomputers.The authors also acknowledge support from the Scientific Computing dvanced Training (SCT project through Europe id contract II-537-FC-F (