1y ago

13 Views

2 Downloads

2.80 MB

12 Pages

Transcription

Bull. Earthq. Res. Inst.Univ. TokyoVol. 2 ,**0 pp. ,, ῌ,-,National Seismic Hazard Maps of JapanHiroyuki Fujiwara*, Shinichi Kawai, Shin Aoi, Nobuyuki Morikawa, Shigeki Senna,Kyoko Kobayashi, Toru Ishii, Toshihiko Okumura and Yuzuru HayakawaNational Research Institute for Earth Science and Disaster Prevention, JapanAbstractThe Headquarters for Earthquake Research Promotion (HERP) of Japan published the nationalseismic hazard maps of Japan in March ,**/, at the initiation of the earthquake research committeeof Japan (ERCJ), on the basis of a long-term evaluation of seismic activity and a strong-motionevaluation. Meanwhile, the National Research Institute for Earth Science and Disaster Prevention(NIED) also promoted the special research project named National Seismic Hazard Mapping Projectof Japan to support preparation of the seismic hazard maps. Under the guidance of ERCJ, we carriedout a study of the hazard maps. There are , types of hazard map : one is a probabilistic seismichazard map (PSHM), which shows the relation between seismic intensity value and its probabilityof exceedance within a certain time ; the other is a scenario earthquake shaking map (SESM). Forthe PSHM, we used an empirical attenuation formula for strong-motion, which followed seismicactivity modeling by ERCJ. Both peak velocity on the engineering bedrock (Vs .** m/s) and onthe ground surface are evaluated for sites with a spacing of approximately km, The potential JMAseismic intensities on the ground surface are also evaluated using an empirical formula. For theSESMs, based on source modeling for strong-motion evaluation, we adopted a hybrid method tosimulate waveforms on the engineering bedrock and peak ground velocity. For this project, wedeveloped an open web system to provide information interactively and retrievally, and named thesystem Japan Seismic Hazard Information Station, J-SHIS (http : //www.j-shis.bosai.go.jp). Weaimed to distribute the process of uncertainty evaluation, and to meet multi-purpose needs in engineering ﬁelds. The information provided by J-SHIS includes not only the hazard map results, butalso various information required in the processes of making the hazard maps, such as data onseismic activity, source models, and underground structure.Key words : Seismic Hazard Map, Probability, Shaking map, J-SHISIntroductionmic Hazard Maps of Japan’ should be promoted as aThe great Hanshin-Awaji earthquake, which oc-major subject of earthquake research. The Nationalcurred on January 1, 33/, killed more than 0,.** per-Research Institute for Earth Science and Disastersons. Following the lessons learned from this disas-Prevention (NIED) then started a special project inter, A Special Act of Earthquake Disaster Manage-April ,** named ‘National Seismic Hazard Mappingment was enacted in July 33/ to promote a compre-Project of Japan’ to support preparation of seismichensive national policy on earthquake disaster pre-hazard maps. After . years of study, in March ,**/vention. The Headquarters for Earthquake ResearchHERP published the ‘National Seismic Hazard MapsPromotion (HERP) was established in accordance withof Japan,’ which was initiated by the earthquake re-this act. In April 333, HERP issued ‘On Promotion ofsearch committee of Japan (ERCJ) on the basis of aEarthquake Research-Comprehensive and Fundamen-long-term evaluation of seismic activity and strong-tal Measures for Promotion of Observation, Measure-motion evaluation. .ment and Research on Earthquakes’. In this article,Under the guidance of ERCJ, we carried out aHERP concluded that preparation of ‘National Seis-hazard map study. There are , types of hazard map.* e-mail : fujiwara@bosai.go.jp (-ῌ , Tennodai, Tsukuba, Ibaraki, -*/ῌ***0, Japan)221

H. Fujiwara, S. Kawai, S. Aoi, N. Morikawa, S. Senna, K. Kobayashi, T. Ishii, T. Okumura and Y. HayakawaOne is a probabilistic seismic hazard map (PSHM),quakes, we evaluate the probability that the inten-which shows the relation between seismic intensitysity measure of strong-motion will be exceeded dur-value and its probability of exceedance within a cer-ing a speciﬁed time.tain time. The other is a scenario earthquake shak-In procedure (-), we use an empirical attenuationing map (SESM). For the PSHM, we used an empiri-relation (Si and Midorikawa, 333) for an intensitycal attenuation formula for strong-motion, which fol-measure of strong-motion.lowed the seismic activity modeling carried out byThe seismic hazard curve is a key concept, whichERCJ. Both peak velocity on the engineering bed-gives the probability of exceeding a speciﬁc level fromrock (Vs .** m/s) and on the ground surface wereall possible earthquakes. Usually, a seismic hazardevaluated for every site spaced approximately kmcurve P (Y y ; t) is deﬁned asapart. The potential JMA seismic intensities on theP Y y ; t ῌ Pk Y y ; t ground surface were also evaluated using an empiri-k where Pk (Y y ; t) is the probability that a groundcal formula. For the SESMs, based on source modeling for a strong-motion evaluation, we adopted a hy-motion Y exceeds level y for the k-th earthquakebrid method to simulate waveforms on the engineer-within time t.ing bedrock layer and peak velocity on the groundIn the following sections we describe detailedsurface.procedures for calculating the seismic hazard curveFor this project, we developed an open web sys-in di#erent source faults and using di#erent prob-tem to provide information interactively and retriev-ability models.ally, and named it Japan Seismic Hazard InformationPSHA for earthquakes with speciﬁed,. . .source faultsStation, J-SHIS (http : //www.j-shis.bosai. go.jp/). Weaimed to distribute the process of uncertainty evalu-The occurrence probability of a potential earth-ation, and to meet multi-purpose needs in engineer-quake with a speciﬁed source fault is evaluated usinging ﬁelds. Information provided by J-SHIS includesa) a renewal process such as a Brownian passage timenot only the results of the hazard maps, but also vari-distribution, or b) a Poissonian process.ous information required in the processes of makingThe following procedure shows how to calculatethe hazard maps, such as data on seismic activity,probability Pk (Y y ; t) using time-dependent and time-source models, and underground structure.independent probability models.a) PSHA using time-dependent probability modelProbabilistic seismic hazard map (PSHM),.Assuming an independency of ground motionlevel for each earthquake, the probability Pk (Y y ; t),. Procedure of probabilistic seismic hazardanalysis (PSHA)that the ground motion exceeds a speciﬁc level y iswritten asProbability or annual rate of earthquake occurrence, and strong-motion levels for all possible earth-Pk Y y ; t quakes are evaluated for PSHMs. The PSHA proce- ῌ P Ek l ; t P Y y Ek l dure used in the Seismic Hazard Mapping Project isl * itemized below.where P (Ek l ; t) is a probability for l occurrences of( ) Following classiﬁcation of earthquakes by ERCJ,the k-th earthquake over time t and P (Y y Ek) is awe set up di#erent mathematical models for charac-probability that a ground motion Y exceeds a level y,teristic seismic activities in Japan.where the occurrence of the k-th earthquake is given.(,) Occurrence probability is evaluated for each earth-P Y y Ek quake.(-) ῌῌ P Y y mi, rj Pk mi Pk rj mi Probabilistic evaluation model of strong-motionij level is selected for each earthquake.where Pk (mi) is a probability density function of the(.) Probability that the intensity measure of a strong-magnitude for the k-th earthquake ; Pk (rj mi) is themotion will exceed a certain level during a speciﬁedprobability density function of the distance from thetime is evaluated for each earthquake.source to the site, and P (Y y mi, rj) is the conditional(/) Considering contributions from all possible earth-probability of exceedance while the occurrence of an222

National Seismic Hazard Maps of Japanearthquake is given with magnitude mi and distanceu Y y u Ek P Y yΐEk rj. k u Ek P Y yΐmi, rj Pk mi Pk rjΐmi An attenuation relation is used to estimate meankground motion “(mY i, rj) as a function of magnitudeand distance. Then P (Y yΐmi, rj) is written asy P Y yΐmi, rj FU “ mY i, rj ῐity that the ground motion exceeds level y given theoccurrence of an earthquake in the k-th area. Pk (mi)is the probability density function of magnitude foris a cumulative function of U.earthquakes in the k-th area ; Pk (rjΐmi) is the proba-In the situation of a small occurrence-probability,bility density function of distance to the site ; and,where an earthquake reoccurs twice or more duringP (Y yΐmi, rj) is the conditional probability of ex-time t, the probability is so small that it can be ig-ceedance while the occurrence of an earthquake isnored ; therefore, equation (,) can be written asgiven for magnitude mi and distance rj.Pk Y y ; t The probability density function of magnitude P Ek ; t P Y yΐEk P (mi) is derived from the Gutenberg-Richter relation. P Ek ; t P Y yΐmi, rj Pk mi Pk rjΐmi We set the maximum and minimum magnitudes, mujand ml, for each area, respectively. Then, the numberwhere P (Ek ; t) is the occurrence probability of thek-th earthquake over time t.of earthquakes classiﬁed by mu and ml, is given byThen, the Brownianpassage time distribution is adopted for evaluatingP (Ek ; t).b) PSHA using time-independent probability modelP (Y y ; t) that a ground motion Y exceeds a speciﬁc N miῑMῑm N Mῒml N Mῒm N Mῒm *a bmlevel y over time t for the k-th earthquake, is given byPk Y y ; t exp uk Y y ῌt N miῑMῑmu N Mῒml N Mῒmu Using the Gutenberg-Richter relation,Assuming a Poissonian model, the probabilitythe distribution function of magnitude is given by FM m P Mῑm where uk (Y y) is an annual rate exceeding a speciﬁcshaking level y for the k-th earthquake, and it can be N Mῒml N Mῒm N Mῒml N Mῒmu ῒ ῌῒ ΐῐῑῒῐ ῒ ῌ exp bln * m ml exp bln * mu ml written asῒ ΐῐῒῐ uk Y y u Ek P Y yΐEk u Ek P Y yΐmi, rj Pk mi Pk rjΐmi ij Then, probability Pk (mi) is written aswhere u (Ek) is the mean annual rate of the k-thearthquake.,. . ,.jsource area, P (Y yΐEk) is the conditional probabil where U is a logarithmic normal distribution and FU (u)iiwhere u (Ek) is the mean annual rate in the k-thP mi P m ῑmiῑm, FM m, FM m PSHA for earthquakes with non-speciﬁed where we putsource faultsIn the case of a seismic hazard evaluation formi Dmῌ, m ῑmi m, miearthquakes with non-speciﬁed source faults, theprobability of earthquake occurrence can be modeled,. ,Dmῌ,Examples of PSHMas a Poissonian process. The probability P (Y y ; t),Using the PSHA procedure mentioned in ,. , fol-of a ground motion Y exceeding a speciﬁc level y overlowed the seismic activity modeling by ERCJ, wetime t, is given byproduced PSHMs throughout Japan.P Y y ; t exp u Y y ῌt At each PSHM, peak velocity on the engineering bedrock (Vs .** m/s) is ﬁrst evaluated based onwhere u (Y y) is the annual rate exceeding shakingcontributions by the empirical attenuation formulalevel y, and it can also be written as(Si and Midorikawa, 333). Then, an ampliﬁcation223

H. Fujiwara, S. Kawai, S. Aoi, N. Morikawa, S. Senna, K. Kobayashi, T. Ishii, T. Okumura and Y. HayakawaFig. . A distribution of ampliﬁcation factor generated by shallow surface layer based on DigitalNation Land Information on geological dataand geomorphological data.Fig. /. The probability distributions in a color scalewhere seismic intensity exceeds the JMA scale 0during the -* years from January ,**/. The PSHM(a) considers all contributions from all earthquakes,and is reprinted from Figure - (a) for comparison :the breakdown cases only consider contributionsfrom (b) subduction zone earthquakes, (c) major 32faults earthquakes, and (d) other earthquakes exclusive the cases (b) and (c).Fig. ,. Distributions for subduction zones, and32 major faults where earthquake occurrenceprobabilities are evaluated by ERCJ.Fig. -. The probability distributions in a color scalewhere seismic intensity exceeds (a) the JMA scale0-, and (b) the JMA scale /-, during the -* yearsstarting from January. ,**/.Fig. 0. SESMs were carried out for the fault zoneswith detailed investigation information. The colorsstand for JMA intensities from - in green to 0 or 1in red.Fig. . JMA seismic intensity in a color scale correspondingto the exceedance probability of (a) -3ῌ, and (b) /ῌ,during the /* years starting from January ,**/.ῌ 224 ῌ

National Seismic Hazard Maps of Japanfactor dependent on surface geological sedimentaryon ground surface. The mesh size of SESM is aboutrocks, as shown in Figure , is used to obtain peak km.ground motions for the sites at spaces of approxi-Figure 0.Some examples of the SESM are shown inmately km. JMA seismic intensities on the groundA hybrid method is adopted as the simulationsurface are also evaluated using an empirical for-method for the strong-motion evaluation. The hy-mula (Midorikawa et al., 333).brid method aims to evaluate strong-motions in aFigure , shows distributions for subduction zones,broadband frequency range. It is a combination of aand 32 major faults where earthquake occurrence pro-deterministic approach using numerical simulationbabilities were evaluated by ERCJ.methods, such as ﬁnite di#erence method (FDM) orFigure - gives , PSHM examples for all of Japanﬁnite element method (FEM) for a low-frequency(ERCJ ,**/). The maps show probabilities that seis-range, and a stochastic approach using the empiricalmic intensity exceeds (a) the JMA scale 0-, and (b) theor stochastic Green’s function method for a high-JMA scale /-, during the -* years starting from Janu-frequency range. A lot of information on source char-ary, ,**/. It reveals the fact that there is a relativeacterization and modeling of underground structurehigh seismic hazard zone in the Tokai, Kii peninsula,is required for the hybrid method. Standardizationand Shikoku area due to large earthquakes with aof setting parameters for the hybrid method is stud-high probability of occurrence in the subduction zoneied in the National Seismic Hazard Mapping Project.of the Philippine Sea plate. Another fact also clearlyIn the following sections, we summarize technicalshown is that surface geological condition a#ects seis-details of the hybrid method based on the ‘Recipe formic hazard. An area covered with soft soil such asstrong-motion evaluation of earthquakes in activesedimentary plains and basins has a relatively highfaults’ and the ‘Recipe for strong-motion evaluationseismic hazard.of earthquakes in plate boundaries,’ which are pub-Figure . shows the JMA seismic intensity corre-lished by the ERCJ, (ERCJ, ,**/).sponding to the exceedance probability of (a) -3 -. Setting parameters for characterized sourcemodeland (b) / during the /* years starting from JanuaryCharacterized source models are composed of as-,**/, respectively.Figure / shows the probability that the JMA seis-perities and a background slip area surrounding themic intensity exceeds 0- over -* years. The PSHM (a)asperities. Source parameters required to evaluateconsidered all contributions from all earthquakes.strong-motions using the characterized source modelThe individual cases only considered contribution (b)are classiﬁed into - parts. The ﬁrst part is a set offrom subduction zone earthquakes, (c) from major 32outer parameters, which shows magnitude and faultfaults earthquakes, and (d) from other earthquakesshape of the earthquake. The second part is a set ofexclusive of cases (b) and (c).parameters, which describes the degree of fault heterogeneity.-.The third part is a set of parameters,Strong-motion evaluation for scenario earth-which deﬁnes the characteristics of rupture propaga-quakestion. Details of parameters are given in the followingIn the National Seismic Hazard Mapping Project,sections and Figure 1.-. . Outer parameters for characterized sourceboth PSHMs and scenario earthquake shaking mapsmodel(SESM), which are for some earthquakes located inspeciﬁed seismic faults with a high occurrence prob-Outer parameters of the characterized sourceability, are prepared. Because of the limitations ofmodel include location of earthquake, size of rupturecomputational capacity and information on model-area, depth, magnitude or seismic moment, and aver-ing, it is di cult to evaluate strong-motion using aage slip on the fault.simulation method in PSHMs. In a SESM, however,In the National Seismic Hazard Maps, the ﬁrst -we can adopt a simulation method based on sourceparameters are given from the ERCJ results for amodeling. Using the simulation method, it is possiblelong-term evaluation of earthquake activities.to evaluate waveforms on an engineering bedrockFor earthquakes occurring in active fault zones,layer, as well as peak acceleration and peak velocityseismic moment M* (dyneῌcm) is given by referringῌ 225 ῌ

H. Fujiwara, S. Kawai, S. Aoi, N. Morikawa, S. Senna, K. Kobayashi, T. Ishii, T. Okumura and Y. HayakawaWe assume that e#ective stress in the asperities sa isto the relation (ERCJ, ,**/), ,.,-ῌ *ῑ /ῌM*,ῌ-ῌ .,.ῌ *ῑ ῌM* ῌ,ῌSῒM* .1ῌ *,/.1ῌ *,/ΐM* .*ῌ *,2equivalent to average stress drop. ῌsbῒwhere S (km,) is a rupture area estimated by theERCJ.For earthquakes occurring in a subduction zoneor at a plate boundary, we determine the seismictics for each earthquake using the seismic data set. DbῌWbῐῌsa DaῌWaῐῐῌwhere Wa and Wb show width of asperity and background slip area, respectively.We determine a cut-o# frequency fmax individu-moment individually by considering its characteris-. . ,The e#ectivestress in the background domain is given byally for each earthquake considering regional charac-Inner parameters for characterized sourceteristics. For a slip velocity time function, we adoptmodela function obtained from a consideration based onInner parameters for the characterized sourcemodel include locations, quantity, and areas of as-the rupture simulation using a dynamic sourcemodel (Nakamura and Miyatake, ,***).perities, average slips and e#ective stresses inside-. . -Other parameters for characterized sourcemodelthe asperities and in the background slip area, fmax,Other parameters for characterized source modeland slip velocity time function.For earthquakes occurring in active fault zones,we determine locations of asperities by consideringare starting point, rupture velocities of the ruptureprocedure, and pattern of rupture propagation.the investigation results for active faults, such asFor earthquakes occurring in active fault zones,trench surveys, and put the asperities just under thewe can determine the starting point of the rupturepositions where the large dislocations are observed.using a branch pattern of the fault if there is informa-There are usually or , asperities for segment oftion. The starting point then can be set up at thean active fault. For earthquakes occurring at plateouter asperities. Otherwise, we put the starting pointboundaries, we determine locations and quantity ofat the bottom of an asperity, if there is no informa-asperities by considering the characteristics of eachtion on the starting point. For earthquakes occur-previous earthquake, using its seismic data set andring at plate boundaries, we determine the startingresults from inversion of rupture process.In thispoint by considering the characteristics of each pre-procedure, we often assume that the locations ofvious earthquake using its seismic data set and re-asperities are invariants.sults of inversion of rupture process. We have as-Using the empirical relation between seismic mo-sumed several cases of characteristic source modelsments and high-frequency level of a source spectrumwhen there was not enough information, because theA (dyneῌcm/s,), the area of asperities can be givenspatial distribution of strong-motion is strongly de-by the following relations,pendent on locations of asperities and starting pointof rupture. When we have no information on ruptureSaῒpr,rῒM*1pῌῌb,.AῌRpattern, we assume that the rupture propagates in ῌa concentric circle at a constant velocity from thestarting point.Aῒ,.0ῌ * 1ῌM* ῌ-The rupture velocity is given asfollows (Geller, 310)where R is the radius of a circular crack whose areaVrῒ*.1, b is the same as the fault area, and b is the shear wavevelocity in the fault region.-. ,ῑῌModeling underground structuresThe ratio of the average slip in asperities Da andFor the simulation, we need a seismic velocity-the average slip in background slip area Db is as-structure with an attenuation model to evaluatesumed to be , : . The average stress drops in thestrong-motions. When modeling underground struc-asperities are given byture, we consider a deep underground structure downM*1Dsaῒ ῌ , 0 r R ῌto the earth’s crust, and/or to the plate boundary,up to the seismic bedrock (Vsῒ- km/s), then to the 226

National Seismic Hazard Maps of JapanFig. 1.Flowchart for setting source parameters.Fig. 2.Flowchart of modeling structure.structure of an engineering bedrock layer (Vs .**modeling the structures of sediments, we use variousm/s 1** m/s), and ﬁnally to the structure of surfaceproﬁles of deep boreholes, reﬂection and refractionlayers as discussed below and in Figure 2.surveys, data from microtremor surveys, as well as-. ,. Deep underground structuredata from gravity surveys. We need to use an opti-The deep underground structure is the structuremized modeling technique for the available data setsfrom the crust and plates up to a seismic bedrockin a target area because quantity and quality of infor-layer with a shear velocity of - km/s. Using velocitymation on underground structure are not uniform inand attenuation models obtained by seismic tomo-all areas. When modeling underground structure forgraphy or geophysical explorations, we have mod-strong-motion evaluations, seismic velocity structureseled the deep underground structure for all of Japan.are most important parameters. It is expected thatA model depth down to the Moho discontinuity isthe accuracy of modeling is proportional to the quan-required for earthquakes in active fault zones, andtity and the quality of data. In an ideal case we candown to the plate boundaries for subduction earth-use all data required.We have made a --dimensional structural modelquakes.-. ,. ,Structure of sedimentswith various available data from throughout Japan.The structure of sediments from the seismic bed-These include deep-borehole proﬁles for accuraterock up to an engineering bedrock layer with a shearstructures at some sites, refraction proﬁles for bound-velocity of .** m/s 1** m/s, strongly a#ects low-ary shapes in large sedimentary basins, reﬂection pro-frequency strong-motions, and is an important factorﬁles for determining the boundary shapes of basinfor evaluating low-frequency strong-motions. Whenedges, data from microtremor surveys, gravity sur-ῌ 227 ῌ

H. Fujiwara, S. Kawai, S. Aoi, N. Morikawa, S. Senna, K. Kobayashi, T. Ishii, T. Okumura and Y. Hayakawaveys and geological information for spatial interpola-can be explained using deterministic simulations basedtion. Furthermore, we verify and modify the struc-on physical models given by the elastodynamic the-tural model if necessary using the above structuralory. On the other hand, it is di cult to evaluate themodel to compare its simulation result of strong-characteristics of high-frequency strong-motions us-motion to recorded seismograms.ing deterministic approaches because the uncertaintyIn fact, however, in many cases it is di cult towhen the setting parameters for the simulations be-obtain su cient data from the above ideal procedurecomes too large due to a lack of information on bothfor --D modeling of velocity structures. In such cases,source modeling and structure modeling. Instead ofthe only information distributed spatially availablea deterministic approach, we need to adopt a stochas-are data from gravity surveys and geological struc-tic approach to evaluate high-frequency strong-mo-ture information.tions. Using broadband strong-motion simulations,Using these data, we estimate velocity structureswe aim to evaluate strong-motions in the frequencyindirectly. Uncertainty in velocity-structure model-range from *. Hz to * Hz. The frequency range in-ing is increased if we use only gravity data becausecludes both low-frequency range and high-frequencygravity data represent a density structure. There-range of strong-motions. To evaluate strong-motionsfore, we also use information on geological struc-in the broadband frequency range, which includes ,tures to reduce uncertainty.frequency ranges whose physical characteristics are-. ,. -Structures of surface soilsdi#erent, it is e cient to use a di#erent approach toWhen modeling the structures of surface soilssimulate strong-motions for each frequency range.from the engineering bedrock layer up to the groundTherefore, a hybrid method is proposed. The hybridsurface, proﬁles of boreholes and surface geology datamethod is a combination of a deterministic approachprovide basic information. The surface soil struc-using numerical simulation methods, such as the ﬁ-tures are locally very heterogeneous, and a largenite di#erence method (FDM) (Pitarka ( 333), Aoi andamount of data is required to accurately model theFujiwara ( 333)) or the ﬁnite element method (FEM)surface soil structure.(Fujiwara and Fujieda, ,**,), to evaluate strong-If a strong-motion evaluation for a large area ismotions based on theoretical models obtained fromrequired, we adopt a rough estimation method to ob-the elastodynamic theory in the low-frequency range,tain ampliﬁcations of surface soils. The rough esti-and a stochastic approach using the empirical or sto-mation method (Fujimoto and Midorikawa, ,**-) ischastic Green’s function method to evaluate strong-based on Digital Nation Land Information on geo-motions in the high-frequency range.logical data and geomorphological data. The meshstrong-motions can be obtained by superposing low-Broadbandsize of these data is about km. The average shearfrequency strong-motions and high-frequency strong-wave velocity for surface structure down to -* m ismotions using matching ﬁlters.estimated using the empirical relation between mi--. -. Deterministic approach for simulatinglow-frequency strong-motionscrogeographical data and the averaged shear wavevelocity. Then, the ampliﬁcation factors for PGV areLow-frequency strong-motions are evaluated byobtained from the empirical relation between aver-solving elastodynamic equations describing the seis-aged shear wave velocity and PGV as shown inmic wave propagation for the physical model, whichFigure .consists of a characteristic source model and an un-If there is su cient information on surface soilderground structure model. We use numerical simu-structures, instead of the rough method, we adopt alation methods, e.g., FDM and FEM, to solve the equa-more accurate method by which we model the sur-tions. Rapid progress of computer technology andface soil velocity-structure for each mesh using asnumerical simulation techniques enable us to solvemany boring proﬁles and as much geological data aspractical problems of strong-motion evaluations. Forpossible.example, in the strong-motion evaluation for earth--. -Broadband strong-motion simulation usingquakes in the Morimoto-Togashi fault zone, we dis-the hybrid methodcretize the underground structure model of domainCharacteristics of low-frequency strong-motions3* km*0* km*.* km into *. km meshes from the sur-ῌ 228 ῌ

National Seismic Hazard Maps of Japanderived from the empirical Green’s function method(Irikura ( 32-), Irikura ( 320)). The empirical Green’sfunction method is an evaluation method for strongmotion waveforms caused by a large earthquake using the ground motion records of earthquakes thatoccur in the source fault as Green’s functions. Theempirical Green’s function method is e#ective forevaluating high-frequency strong-motions that arestrongly a#ected by heterogeneities of propagationpaths and local site conditions. In many cases, however, we have no ground motion record of a properearthquake that occurred in the source fault of thetarget large earthquake. With the stochastic Green’sfunction method, we use functions generated by astochastic method instead of ground motion recordsfor Green’s functions. The stochastic Green’s function method can be applied if we have no groundmotion record for the Green’s function. The Green’sFig. 3. Information from deep sedimentary structuremodel in Japan (meter).functions used in the stochastic Green’s functionmethod are stochastically approximated, and haveno information on phases.The stochastic Green’sfunction method should be used to evaluate the envelope of strong-motion waveforms.Fig. *. The hybrid method is a combination of adeterministic approach and a stochastic approach.face down to a depth of . km, and *.- km meshes fordeeper parts. It takes ./ hours to calculate 0,*** timesteps for this model using our FDM code in origin-2**, 0.CPU. However, if we calculate using a meshof half the size for same domain, required computation time and memory size becomes 0 times and 2times, respectively.-. -. ,Stochastic approach for simulating highfrequency strong-motionsWe adopt the stochastic Green’s function method(Dan and Sato, 332) to evaluate high-frequency strongmotions. The stochastic Green’s function method isῌ 229 ῌFig. . Japan Seismic Hazard Information Station,J-SHIS (http : //www.j-shis.bosai.go.jp).

H. Fujiwara, S. Kawai, S. Aoi, N. Morikawa, S. Senna, K. Kobayashi, T. Ishii, T. Okumura and Y. Hayakawa-. -. -The hybrid methodis regarded as a phenomenon in the PSHMs.Using the hybrid method, we obtain broadbandIn the proposed method (M ), we can select astrong-motion waveforms by superposing low-fre-scenario earthquake that is dominant to the strong-quency strong-motion w

not only the results of the hazard maps, but also vari-ous information required in the processes of making the hazard maps, such as data on seismic activity, source models, and underground structure.,. Probabilistic seismic hazard map (PSHM),. Procedure of probabilistic seismic hazard analysis (PSHA) Probability or annual rate of earthquake .

Related Documents: