JAPAN SEISMIC HAZARD INFORMATION STATION, J-SHIS

4 downloads 0 Views 5MB Size Report
Oct 12, 2006 - national seismic hazard maps of Japan on March 2005, which was initiated by the earthquake research .... as data on seismic activity, source models and underground structure. ... of the Philippine Sea plate. Another fact is ...
4th International Conference on Earthquake Engineering Taipei, Taiwan October 12-13, 2006 Paper No. 274

JAPAN SEISMIC HAZARD INFORMATION STATION, J-SHIS Hiroyuki Fujiwara1, Shinichi Kawai1, Shin Aoi1, Toru Ishii1, Toshihiko Okumura2, Yuzuru Hayakawa1, Nobuyuki Morikawa1, Shigeki Senna1, Kyoko Kobayashi1, and Ken Xian-Sheng Hao1 ABSTRACT

The Headquarters for Earthquake Research Promotion (HERP) of Japan published the national seismic hazard maps of Japan on March 2005, which was initiated by the earthquake research committee (ERC) on a basis of a long-term evaluation of seismic activity, and evaluation of strong ground-motion. At the same time, the National Research Institute for Earth Science and Disaster Prevention (NIED) also promoted the National Seismic Hazard Mapping Project of Japan to support the study. Under the guidance of ERC, we have been carrying on this study which consists of two kinds of maps; One is a probabilistic seismic hazard map (PSHM) that shows the relation between seismic intensity value and its probability of exceedance within a certain period. Another one is a scenario earthquake shaking map (SESM). For the PSHM, we used an empirical attenuation formula following the seismic activity modeling by ERC. Both the peak velocity on the engineering bedrock (Vs=400m/s) and on the ground surface are evaluated for sites with at 1km spacing, The potential JMA seismic intensities on ground surface are also evaluated by using an empirical formula. For the SESMs, based on the source modeling for strong-motion evaluation, we adopted a hybrid method to simulate waveforms on the engineering bedrock and peak ground velocity. In this project, we developed an open web system to provide information interactively, and named this system as Japan Seismic Hazard Information Station, J-SHIS (http://www.j-shis.bosai.go.jp/). Our products are aimed to meet multi-purpose needs in engineering fields by providing information of the uncertainty evaluation. The information provided from J-SHIS includes not only results of the hazard maps but also various information required to produce the hazard maps, such as data on seismic activity, source models and underground structure. Keywords: Seismic Hazard Map, Probability, Shaking map, J-SHIS INTRODUCTION Following the lessons learned from the 1995 Hanshin-Awaji earthquake, which killed more than 6,400 people, a special Japanese Act of Earthquake Disaster Management was enacted on July 1995 and the Headquarters for Earthquake Research Promotion (HERP) was established. Under the HERP strategy, the National Research Institute for Earth Science and Disaster Prevention (NIED) has been carrying on the National Seismic Hazard Mapping Project of Japan to support the preparation of seismic hazard maps from April 2001. Therefore, HERP published the national seismic hazard maps 1

National Research Institute for Earth Science & Disaster Prevention, Japan, [email protected], [email protected]

2

Shimizu Corporation, Japan

of Japan on March 2005, which was initiated by the earthquake research committee of Japan (ERC) on a basis of the long-term evaluation of seismic activity and on a basis of strong-motion evaluation. We have been also developing an open web system to provide information interactively, and named this system as Japan Seismic Hazard Information Station, J-SHIS (http://www.j-shis.bosai. go.jp/). The information from J-SHIS includes not only results of the hazard maps but also various information required in the processes of making the hazard maps, such as data on seismic activities, source models and underground structures. Our products are aimed to meet multi-purpose needs in engineering fields by even providing the information of the uncertainty evaluation. THE J-SHIS WEB SYSTEM (http://www.j-shis.bosai.go.jp) An open web system, which we named as Japan Seismic Hazard Information Station, J-SHIS, has been operating from 2005, and the English version from 2006, as shown in Figure 1. The J-SHIS is achieved after 3 years investigations and technique approaching from the engineering point of view, under an engineering application committee of National Seismic Hazard Maps, NIED. Two main parts of the J-SHIS web system consist of probabilistic seismic hazard maps (PSHM) and the scenario earthquake shaking maps (SESM) as the examples shown in Figure 2 and 3, and discussed in the following sections. The color in a scale shown in Figure 1, the top page of J-SHIS, stands for the probability that ground-motions equal to or larger than the JMA intensity 6 Lower in considering of all of the earthquakes in the period of 30 years. The J-SHIS web system can show various SHMs up to the conditions to be chosen, e.g., type of earthquake, intensity, probability, reoccurring period, as well as the area. A zoom-in of SHMs can go up to a mesh unit of one kilometer2 with boundary of a political area as shown in Figure 2. One can even get a location relatively by the local map with routes and railways. A search function can go through the name of village, town, ward, city, as well as railway line and stations. Data and map images download are also available with probability and intensity in a mesh unit.

Figure 1. Japan Seismic Hazard Information Station, J-SHIS (http://www.j-shis.bosai.go.jp) Color in a scale stands for probabilities.

In the page of SESM as an example shown in Figure 3, there is strong-motion simulation result from detail source model, while the specified seismic fault is known, such as Itoigawa-shizuoka fault, Miyagi-oki fault. The information open to public includes velocity waves, seismic intensity on surface, fault shapes and parameters. Our products are aimed to meet multi-purpose needs in engineering fields by providing information of the uncertainty evaluation. The information provided from J-SHIS include not only results of the hazard maps but also various information required in the processes of making the hazard maps, such as data on seismic activity, source models and underground structure. Over 100,000 accesses to J-SHIS can be reached whenever an earthquake occurs. A questionnaire investigation recently shows that not only professional but also ordinary people access the J-SHIS. Professional concerns are includes theoretical ideas and technique calculations, application fields,

such as insurance industry, engineering application, and even political managements. Public concerns are concentrated in the fields of mitigation disaster, aseismic design, and earthquake reinforcement. The J-SHIS web system will be developed as a sharing platform of the seismic hazard information, for an interactive communication by researchers and users in general.

Figure 2. A zoom-in of probabilistic seismic hazard map (PSHM) in a color scale can go up to a mesh unit in one kilometer2 with boundary of a political area in the page of PSHM. .

Figure 3. Strong-motion simulation results of JMA intensity in a color scale show scenario earthquake shaking maps (SESM) by detail source models and parameters in the page of SESM.

PROBABILISTIC SEISMIC HAZARD MAP (PSHM) Procedure of probabilistic seismic hazard analysis (PSHA) The seismic hazard curve is a key concept which gives a probability of exceeding a specific level from all possible earthquakes. Usually a seismic hazard curve P(Y > y;t) is defined as

P(Y > y;t) =1 − ∏ {1 − Pk (Y > y;t)}

(1)

k

where Pk (Y > y;t) is a probability that a ground motion Y exceeds level y for the k-th earthquake in the time period t. The probabilities or annual rates of earthquake occurrence, and strong-motion levels for all possible earthquakes are evaluated on PSHMs. A PSHA procedure used in the Seismic Hazard Mapping Project is listed as below. (1) Following ERC’s classification of earthquakes and fault zones as shown in Figure 4, we setup different mathematical models for characteristic seismic activities in Japan. (2) The occurrence probability is evaluated for each earthquake. (3) Probabilistic evaluation model of strong-motion level is selected for each earthquake. (4) Probability that an intensity measure of strong-motion will exceed certain level during a specified time period is evaluated for each earthquake. (5) Considering contributions from all possible earthquakes, we evaluate the probability that the intensity measure of strong-motion will be exceeded during a specified time period.

In the procedure (3), we use an empirical attenuation relation (Si and Midorikawa, 1999) for an intensity measure of strong-motion. At each PSHM, peak velocities at the engineering bedrocks (Vs=400m/s), are first evaluated based on contributions by the empirical attenuation formula. Then, amplification factors dependent on surface geological sedimentary as shown in Figure 5, are used to obtain peak ground motions for the sites with approximately 1km spacing. The JMA seismic intensities on ground surface are also evaluated by using an empirical formula (Midorikawa et al., 1999). The PSHM examples

Figure 2. Distributions for subduction zones, and 98 major faults where earthquake occurrence probabilities are evaluated by ERCJ.

By the PSHA procedure as mentioned above, followed the seismic activity modeling by ERC, we produced PSHMs over all of Japan. Figure 6 gives two PSHM examples for the whole of Japan (ERC 2005). The maps show probabilities that seismic intensity exceeds (a) the JMA scale 6-, and (b) the JMA scale 5-, in 30 years starting from January, 2005. The maps reveal a fact that there is a relative high seismic hazard zone in the Tokai, Kii peninsula and Shikoku area due to large earthquakes with high probability of occurrence in the subduction zone of the Philippine Sea plate. Another fact is also clearly that the surface geological condition affects on seismic hazard. The area covered with soft soil such as sedimentary plains and basins has relatively high seismic hazard.

Figure 4. Distribution for subduction zones in black thick line, and 98 major faults in red line where earthquake occurrence probabilities are evaluated by ERC.

Figure 7 shows the JMA seismic intensity corresponding to the exceedance probability of (a) 39% and (b) 5% in 50 years starting from January, 2005, respectively. One can obtain much more detail maps from the J-SHIS web site such as listed as following. The probability distributions that the JMA seismic intensity exceeds 6- (or 5-) in 30 (or 50) years. The PSHM considering all of contributions from all earthquakes, and/or from individual earthquake. The PSHMs considering only the contribution from subduction zone earthquakes, or from major 98 faults earthquakes, or from other earthquakes exclusive the mentioned above.

Figure 5. Distribution of amplification factors in a color scale, generated by shallow surface layer based on geological data and geomorphological data from the Digital Nation Land Information.

(b)

(a)

(a) (b)

(b) Figure 6. The probability distributions in a color scale where seismic intensity exceeds (a) the JMA scale 6-, and (b) the JMA scale 5-, in 30 years starting from January, 2005.

(b) Figure 7. JMA seismic intensity in a color scale corresponding to the exceedance probability of (a) 39%, and (b) 5%, in 50 years starting from January, 2005.

STRONG-MOTION EVALUATION FOR SCENARIO EARTHQUAKES

In the National Seismic Hazard Mapping Project, not only PSHMs, but also scenario earthquake shaking maps (SESM) are produced, in which SESMs are for some earthquakes located in specified seismic faults with a high occurrence probability. By the limitation of computational capacity and information on modeling, it is difficult to evaluate strong-motion by using a simulation method in the PSHM. However, it is possible to evaluate waveforms on an engineering bedrock layer, as well as peak acceleration and peak velocity on ground surface in the SESM by adopting the simulation method based on a source modeling. Some examples of the SESM, with a mesh size of 1km, are shown in Figure 8.

Figure 8. SESMs were carried out for the fault zones with detail investigation information. The colors stand for the JMA intensity from 3 in green to 6 Higher in red.

A hybrid method in a broadband frequency range is adopted as the simulation method for strongmotion evaluation. It is a combination of a deterministic approach using numerical simulation methods, such as finite difference method (FDM) or finite element method (FEM) for low frequency range, and a stochastic approach using the empirical or stochastic Green’s function method for high frequency range as shown in Figure 9. A lot of information on source characterization and modeling of underground structure is required for the hybrid method. The standardization of the setting parameters for the hybrid method is based on the ‘Recipe for strong-motion evaluation of earthquakes in active faults and in plate boundaries’, by the ERC, (ERC, 2005).

Low frequency range

Matching filter Low frequency range

High frequency range

Finite Difference Method

High frequency range

Stochastic Green’s function method

Superposition

deterministic

Stochastic

Figure 9. The hybrid method is a combination of deterministic and stochastic approach.

Modeling for characterized source model

Source parameters required to evaluate strongmotions by using the characterized source model are classified into three parts; The first part is a set of outer parameters that show magnitude and fault shape of the earthquake. The second part is a set of parameters that describe the degree of fault heterogeneity. Characterized source models are composed of asperities and a background slip area surrounding the asperities. The third part is a set of parameters that define characteristics of the rupture propagation. The detail of parameters is given as following sections and Figure 10.

Long-term evaluation of earthquake activities

Outer characteristics of source Type of earthquake Seismic moment, rupture area, average slip High frequency level of acceleration source spectrum

Inner characteristics of source Asperity Number, location, area, average slip, effective stress

Background slip area Area, average slip, effective stress

Modeling of underground structures

Once strong-motion simulation concerned, we need to deal with seismic velocity-structures with attenuations. We consider the deep underground structures either down to bottom of the earth crust, or to bottom of the plate boundary, then up to a seismic bedrock with a shear-wave velocity (Vs) of 3km/s, farther up to a structure of an engineering bedrock layer with Vs=400m/s ~ 700m/s, and final up to a structure of surface layers as discussed below and Figure 11. Deep underground structure The deep underground structure indicates the structure down to bottom of the crust and/or plates up to a seismic bedrock layer with a shear velocity of 3km/s. Using velocity and attenuation models obtained by the seismic tomography or geophysical explorations, we have been modeling the deep underground structure over of the whole Japan as shown in Figure 12. It is required that model’s depth is down to the Moho discontinuity for the earthquakes in the activefault zones, and down to the plate boundaries for the subduction earthquakes. Structure of sediments The structure of sediments indicates the structure from the seismic bedrock layer up to an engineering bedrock layer with a shear velocity of 400m/s-700m/s. This structure strongly controls the low-frequency strong-motions; therefore, it is an important factor for the evaluation of low-frequency strong-motions. For modeling of the structure of sediments, we use various profiles of deep boreholes, reflection and refraction surveys, data from microtremor surveys, as well as data from the gravity surveys.

Slip velocity time function, fmax

Other characteristics of source Starting point of rupture Patern of rupture propagation Rupture velocity

Characterized Source Model

Figure 10. Flowchart for setting source parameters

Geophysical exploration

Geographyical or Geological data

Reflection survey

Topographycal map

Refraction survey

Geological map

Gravity survey

Geographyical data

Microtrmor survey

Geological profiles of subsurface structure

Velocity profiling

Extraction of information on velocity structure

Boring profiles Boring for exploration of resources Boring for hot spring

Geological modeling of subsurface structure

Conparison of velocity stracture and geological structure

Subsurface structure

Contour maps for subsurface structure (initial model) Modification using gravity data and geological information Contour maps for subsurface structure (modified model) Weathered layers Contour maps for subsurface structure (final model)

3-D grid model for numerical simulation

Figure 11. Flowchart of structure modeling.

We need to use an optimized modeling technique for available data sets in a target area because quantity and quality of information on underground structure are not uniform in all areas. In the modeling of underground structure for the strong-motion evaluations, seismic velocity structures are the most important parameters. It is expected that the accuracy of the modeling is proportional to the quantity and the quality of data. In an ideal case we can use all data to be required. We have made a three-dimensional structure model with various available data in all of Japan. These are including using deep-borehole profiles for accurate structure at some sites, refraction profiles for boundary shapes in large sedimentary basins, reflection profiles for determination of the boundary shapes of basin edges, data from microtremor surveys, gravity surveys and geological information for spatial interpolation. Furthermore, we verify and modify the structure model if it is necessary by using the above structure model to compare its simulation result of strong-motion to the recorded seismograms. However, it is difficult to obtain sufficient data required in the above ideal procedure for 3-D modeling of velocity structures in many cases. In such a case, only available information is from gravity survey and geological structure information. Using these data, we estimate velocity structures indirectly. Uncertainty in a velocitystructure modeling is increase if we use only gravity data because a gravity data represent a density structure. Therefore we also use the information on geological structures to reduce the uncertainty.

Figure 12. Information of deep sedimentary structure model in Japan (meter).

Structure of surface soils

In the modeling of a structure of surface soils from the engineering bedrock layer up to ground surface, profiles of boreholes and data of surface geology are basic information. The surface soil structures are locally very heterogeneous and large amount of data is required to model accurately the surface soil structure. In the case that strong-motion evaluation for large area is required, we adopt a rough estimation method to obtain amplifications of surface soils. The rough estimation method (Fujimoto and Midorikawa, 2003) is based on geological data and geomorphological data from the Digital Nation Land Information. The mesh size of these data is about 1km. The average shear-wave velocity for surface structure down to 30m is estimated by using the empirical relation between microgeographical data and the averaged shear-wave velocity. Then site amplification factors for PGV are obtained from the empirical relation between the averaged shear-wave velocity and PGV as shown in Figure 5. In the case there is sufficient information on surface soil structures, in stead of the rough method, we adopt a more accurate method in which we model the surface soil velocity-structure for each mesh using as many boring profiles and geological data as possible.

RELATIONSHIP OF THE TWO KINDS OF HAZARD MAPS

The National Seismic Hazard Maps of Japan consist of two kinds of hazard maps. One is PSHM that shows the relation between seismic intensity value and its probability of exceedance within a given period. The other one is SESM with a specified source fault. Superimposition of the SESMs into the PSHM is considered. As the methodologies of the superimposition procedure, the following two kinds of methods are proposed. z M1: Relating the SESMs into the PHSMs by using a concept of the contribution factor proposed by Kameda et al (1997). We call this method “weak superimposition”. z M2: Making PSHMs by using the hybrid method for all possible earthquakes. Then each SESM is regarded as a phenomenon in the PSHMs. In the proposed method of M1, the scenario earthquake in the area can be chosen by the condition that the predominant of strong-motion level corresponds with the probability level we are interested. It is possible that making the SESM’s probabilistic-position clear by comparing the SESMs with the PSHMs, or by comparing the strong-motion level directly with the seismic hazard curve. Under the present circumstances, we cannot adopt the methodology M2 for the PSHA in the mapping project because the computation for the hybrid method is too large to carry out and also because the information we currently have is not sufficient for precise source modeling and underground structure modeling. In the future, however, the methodology proposed in M2 can be used for the PSHA in order to evaluate strong-motions more precisely. Once the methodology M2 will be achievement, the next generation SHM will become true that we can directly locate a SESM as a phenomenon of the PSHM and make clear relationship of the two kinds of hazard maps. DISCUSSION

1.

2.

3.

Reliability of an intensity level with small probability for an earthquake, whose occurrence probability is high, becomes a subject of discussion in preparation of the PSHMs. An intensity level with small probability is sometimes strongly affected by the amount of uncertainty contained in an empirical attenuation relation used in a seismic hazard analysis. We often adopt a quantity of deviation, which is obtained during a statistical regression analysis, to derive the attenuation relation. However, it consists of uncertainties from various sources because the empirical attenuation relation is based on the data observed at various sites from various earthquakes. It has been pointed out that this procedure overestimates a seismic hazard because a spatial uncertainty is to be considered separately as epistemic uncertainty. On the other hand, if we use a stochastic Green’s function method or a hybrid method, instead of empirical attenuation relation in the strong-motion evaluation, the spatial uncertainty is not mingled and is more realistic. Further studies from both data and theory are necessary to properly quantify the uncertainty in ground-motion evaluation. A Brownian passage time distribution as well as a Poisson process is adopted for the long-term evaluation of earthquake activity by ERC. Because of insufficiency of information on previous earthquake activity in fault zones, the average recurrence interval and elapsed time since the latest earthquake event are often expressed as a time range instead of a fixed value, and consequently the occurrence probability is given by an interval. In the PSHMs, based on discussion on the treatment of the probability given by an interval, we adopt the median and the maximum probability value for the typical case, and for reference case, respectively. Although not only the estimated parameters but also degrees of reliability on estimation of parameters are shown in a long-term evaluation of earthquake activity, it is quite difficult to quantify the degrees of reliability, thus are not reflected in the hazard maps. Selection of a specified earthquake is essential to make a SESM. The basic selective policy of a scenario earthquake in the mapping project is to choose the most probable case in the area.

However, available information to determine the source parameters is often insufficient and in the same time decision has to be made under the situation of the uncertain factors remaining. In that case, we assume several cases of the characteristic source models and compare their results to show deviation of strong-motion evaluation due to uncertainties. It is still under consideration that reliability evaluation of the results with the decision under the uncertain factors remaining. We have been continuing our effects for accumulating data from seismic observation networks, such as K-NET, KiK-net, and for constructing structure database from surveys on the basis of a longterm policy. ACKNOWLEDGMENTS

This study was done under the direction of the ERC (Chairman: Kenshiro Tsumura), the subcommittee for long-term evaluations (Chairman: Kunihiko Shimazaki) and the subcommittee for evaluations of strong ground motion (Chairman: Kojiro Irikura). We are deeply grateful for valuable comments from the committee for methodology of PSHA (Chairman: Saburo Midorikawa) and the engineering committee for use of the seismic hazard maps (Chairman: Hiroyuki Kameda). REFERENCES

Aoi S and Fujiwara H. (1999), 3D finite-difference method using discontinuous grids. Bull. Seism. Soc. Am., 89, 918-930. Dan K and Sato T. (1998), Strong-motion prediction by semi-empirical method based on variable-slip rupture model of earthquake fault. (in Japanese) J. Struct. Constr. Eng., AIJ, 509, 49-60. Earthquake research committee of Japan (2005), General seismic hazard map covering whole of Japan, (in Japanese). Fujimoto K and Midorikawa S. (2003), Average shear-wave mapping throughout Japan using the Digital National Land information. (in Japanese) Journal of JAEE, 3, 13-27. Fujiwara H and Fujieda T. (2002), Voxel finite element method for 3-D elastodynamic analysis. (in Japanese) Proceedings of the 11th Japan Earthquake Engineering Symposium, 94. Geller RJ. (1976) Scaling relations for earthquake source parameters and magnitudes. Bull. Seism. Soc.Am., 66, 1501-1523. Irikura K. (1983), Semi-empirical estimation of strong ground motions during large earthquakes. Bulletin of the Disaster Prevention Research Institute, Kyoto University, Vol.33, Part2, No.298, 63-104. Irikura K. (1986), Prediction of strong acceleration motions using empirical Green’s function. Proceedings of the Seventh Japan Earthquake Engineering Symposium, 151-156. Kameda H, Ishikawa Y, Okumura T and Nakajima M. (1997), Probabilistic scenario earthquakes definition and engineering applications-. (in Japanese) Journal of Structural Mechanics and Earthquake Engineering, JSCE, 577, 75-87. Midorikawa S, Fujimoto K, Muramatsu I. (1999), Correlation of new J.M.A. instrumental seismic intensity with former J.M.A. seismic intensity and ground motion parameters. (in Japanese) Journal of ISSS, 1, 51-56. Nakamura H and Miyatake T. (2000), An approximate expression of slip velocity time function for simulation of near-field strong ground motion. (in Japanese) Zisin, 53, 1-9. Pitarka A. (1999), 3D elastic finite-difference modeling of seismic motion using staggered grids with nonuniform spaceing. Bull. Seism. Soc. Am., 89, 54-68. Si H, and Midorikawa S. (1999), New attenuation relationships for peak ground acceleration and velocity considering effects of fault type and site condition. (in Japanese) J. Struct. Constr. Eng., AIJ, 523, 63-70.