Sei sulla pagina 1di 14

Bulletin of the Seismological Society of America, Vol. 93, No. 6, pp.

25022515, December 2003

Probabilistic Seismic Hazard Assessment Methodology for Distributed Seismicity


by Janez Lapajne, Barbara Sket Motnikar, and Polona Zupancic

Abstract This study deals with probabilistic seismic hazard analysis (PSHA) based on earthquakes that cannot be assigned to specic geologic structures (distributed seismicity). To calculate seismic hazard from distributed seismicity, the PSHA methodology is extended in numerous ways. A simple quantitative seismotectonic model enables statistical incorporation of certain seismotectonic knowledge and the application of attenuation relationships that are based on fault distance. The inuence of a large historic earthquake on seismic hazard is taken into account with an energybased seismic activity rate, obtained by spatial distribution of the released seismic energy. It will be shown that for at least a rough seismic hazard assessment, the type of geometrical seismicity modeling is not a deciding factor. To illustrate this, the classical seismic source zone approach and the improved spatially smoothed seismicity approach are studied in detail to determine their common and distinctive components. Except the geometry of modeling the seismic activity, there is no essential difference between the two approaches. Seismic activity in Slovenia can be treated as distributed seismicity. For this reason, the presented methodology is applied to the calculation of seismic hazard maps of Slovenia as a case study.
Introduction
This study focuses on modeling seismicity that cannot be assigned to specic geologic structures due to inaccurate and insufcient seismologic and geologic data. It is generally termed distributed seismicity (Stirling, 2000). Such earthquakes are sometimes referred to as oating or random earthquakes (dePolo and Slemmons, 1990). Seismotectonic knowledge of a given area often does not allow the exact determination of faults that cause earthquakes of magnitude below 5. However, many historical earthquakes with magnitudes in the range between 5 and 6.5 are also oating, because their records in the earthquake catalog are very inaccurate. As an example, in the probabilistic seismic hazard analysis (PSHA) of California, earthquakes with a magnitude of 46.5 have been modeled separately from the well-known fault sources. In this case, distributed seismicity is named background seismicity (Cao et al., 1996), as opposed to the seismicity of the well-known faults. McGuire (1993) classied the main procedures of PSHA into deductive and historic categories. Historic methods use only information from earthquake catalogues, for example, the nonparametric historic method by Veneziano et al. (1984). On the other hand, deductive methods focus also on the determination and estimation of the cause and origin of earthquakes. The most popular deductive method follows the Cornell (1968) approach and is based on the determination of seismogenic source zones with homogeneous seismic activity rate. The main problem in this approach can be large subjectivity in the delineation of seismic source zones, particularly where the seismotectonic knowledge of the observed area is poor. Some interesting approaches that use the advantages of both methods are the spatially smoothed seismicity procedure developed by Frankel (1995) and Frankel et al. (2000), the parametric-historic procedure proposed by Kijko and Graham (1998, 1999), and the zoning free kernel approach proposed by Woo (1996). All of these methods may be applied to the modeling of distributed seismicity. Although the characteristics of distributed seismicity increase the unreliability of deductive approaches, the classical seismic source zones method is still widely used. On the other hand, the simplicity and natural disposition of the spatially smoothed seismicity approach are convenient for avoiding the subjective delineation of seismic source zones. The particular suitability of this approach for modeling distributed seismicity leads us to study the spatial smoothing in detail and to make improvements in numerous ways (e.g., Lapajne, 2000; Sket Motnikar et al., 2000). An important goal of this research is also to bring both methods

2502

Probabilistic Seismic Hazard Assessment Methodology for Distributed Seismicity

2503

to a common base and to prove that the geometrical (spatial) modeling of seismicity is in fact the only differentiating factor. Numerous seismic hazard assessment studies are based only on the earthquake catalog. However, for detailed and particularly for site-specic estimates, seismotectonic knowledge should also be taken into account. The amount and the quality of the input data inuence the assumptions made about parameters and on the types of modeling of the seismic activity, magnitude distribution, attenuation relationship, and seismotectonics. Later in this article, many alternatives in the input data will be proposed with the aim to generalize the methodology as much as possible.

The corresponding probability density function of the magnitude for the log-linear case is p(m) and p(m) bln10 10 b(m m0) 1 10 b(mu m0) bln10 10
b(m m0)

(2a)

(2b)

Common Characteristics in Modeling Distributed Seismicity


Earthquake Time Series In PSHA, it is common practice to assume that the Poisson process describes the earthquake occurrence rate. In areas where causal faults are unknown, the Poissonian model is particularly suitable. But it is justied even in many situations where the fault memory actually exists (Cornell and Winterstein, 1988). To enable the application of the Poisson process, all preshocks and aftershocks should be removed from the earthquake catalog. Then, at least one complete subcatalog should be determined. In the complete subcatalog, all earthquakes above the corresponding lower bound magnitude m0 are assumed to be recorded. This lowest value of the magnitude in the observed complete subcatalog determines the level or threshold of completeness. Magnitude Distribution The standard log-linear magnitudefrequency relationship was established by Gutenberg and Richter (1944), n(m) n(m0) 10
b(m m0)

(1a)

and later generalized to the doubly truncated exponential recurrence relationship (Cornell and Vanmarcke, 1969), n(m) n(m0) 10
b(m m0)

10
b(mu

b(mu m0)

m0)

10

(1b)

where n(m) is the cumulative annual number of earthquakes with magnitude equal to or greater than m, the seismic activity rate n(m0) is the total annual number of earthquakes with magnitude equal to or greater than the lower bound magnitude m0, mu is the upper bound magnitude, and b is the decay rate. Usually, the seismic activity rate is normalized to an annual rate. Assuming that the magnitude is bounded, the doubly truncated exponential distribution is recommended, as it naturally follows from the maximum entropy principle (Berrill and Davis, 1980).

for the doubly truncated exponential recurrence relationship. Parameters n(m0), m0, mu, and b of this relationship may be determined for the entire observed area (global parameters) or for its smaller parts (local, variable parameters), for example, for a particular source zone or a set of grid cells. However, a too small or a very low hazard area may not be appropriate for statistical evaluation. Of course, the estimates depend on the chosen estimation method and on the time span of the catalog. As observed earlier, the lower bound magnitude m0 generally equals the lowest magnitude in the observed complete subcatalog and is therefore taken as a global parameter for the whole area. For the estimation of decay rate b, it was shown (Page, 1968; Weichert, 1980) that the maximum likelihood estimate (MLE) has the best properties. It may be a local or a global estimate. A common way to determine the seismic activity rate n(m0) is to count the events in the complete earthquake subcatalog. This method usually fails for very large earthquakes due to historical records that are generally too short. Sometimes, the rate of large earthquakes may be determined from the geological record, for example, from fault slip rates (Youngs and Coppersmith, 1985). When such geological information is not given, even large earthquakes can be treated as part of distributed seismicity. Later, we propose an alternative way to emphasize the inuence of larger earthquakes on the resulting seismic hazard, which is based on the released seismic energy (Lapajne et al., 1997). For the well-known fault sources, the upper bound magnitude mu can be based on the maximum observed size of ruptures using empirical relationships between the rupture dimension and the magnitude. For distributed seismicity, the causal faults are unknown, which is why other approaches should be used. Statistical evaluation of the upper bound magnitude mu was most thoroughly treated by Kijko (e.g., Kijko and Sellevoll, 1989; Kijko and Graham, 1998). However, for an approximate estimation of the upper bound magnitude mu, the maximum observed magnitude in the (sub)catalog might be increased by a chosen proportion of the magnitude unit, for example, 0.5. A simple criterion for the determination of the magnitude increment was proposed by Lapajne and Sket Motnikar (1996). In their paper, a welldened complete earthquake subcatalog was chosen to estimate the decay rate b with various values of the upper bound magnitude mu. The value of mu, at which the corre-

2504 sponding standard error of decay rate is minimal, was taken as an estimate of the upper bound magnitude for the period of the chosen subcatalog. The difference between this estimate and the largest observed magnitude in the given complete subcatalog was taken to be a magnitude increment also for other subcatalogs in the given area. Grid of Cells In the PSHA procedures, it is convenient to use a horizontal at surface grid of rectangular cells. All input and output values are given in the centers of grid cells. All epicenters and fault centers inside a particular grid cell are assigned to its center. Due to the inaccuracy of input data and numerical rounding the grid can be rather coarse, for example, rectangles of length sides 0.1 or squares of length side 10 km. Finer grids give smoother maps, but the computation takes more time. A grid of cells is suitable for representing different kinds of seismic sources. The center of a cell may either represent a point source or a point on a fault (usually a center on a surface fault trace), while an areal source zone may be considered as a set of corresponding grid cells. Distribution of Faults: Quantitative Seismotectonic Model One of the main problems in PSHA is how to handle the available seismotectonic knowledge. Also the classication of a given approach into a deductive or historical group strongly depends on this. When modeling distributed seismicity, seismotectonics is usually incorporated by using subjective engineering judgment, or only implicitly through the earthquake catalog. Nevertheless, if in a given area at least a general knowledge about the main types and directions of seismogenic structures exists, some quantitative statistical modeling is possible. We suggest a simple seismotectonic model that statistically connects earthquakes and faults and enables the use of attenuation relationships for epicenters and faults. In applying a grid of cells, each center of a cell has the following attributes: type of dominating structures (i.e., strike slip, normal or reverse fault, thrust), their directions (azimuths), and the corresponding weights. Grid cells with the same attributes may be grouped into regions of homogenous seismotectonic characteristics, such as in Poljak et al. (2000b). Different Geometry of Seismicity Model The geometry of distributed seismicity may be modeled with the delineation of areal source zones (e.g., EPRI, 1988; Bernreuter et al., 1989). Areal source zones are dened with polygons and have a uniform distribution of epicenters. Areal source zones may or may not contain earthquakes from the neighboring area. An alternative to the classical source zone approach is Gaussian smoothing of seismicity. It allows fuzzy modeling of activity rate, where grid cells are the only geometrical elements. Simple one-stage circular Gaussian smoothing

J. Lapajne, B. Sket Motnikar, and P. Zupancic

was introduced and further improved by Frankel (1995), Frankel et al. (2000), Cao et al. (1996), and Lapajne et al. (1997). Two-stage, circular and elliptical smoothing was im plemented by Lapajne (2000) and Sket Motnikar et al. (2000) and applied also by Hamdache et al. (2001). The corresponding procedure and smoothing equations are given in detail later. Computer programs for seismic hazard assessment, using the classical source zone approach (e.g., FRISK88 [Risk Engineering, 1988] and Seisrisk III [Bender and Perkins, 1987]), as well as their later versions, are widely used. However, there has been a lack of software for computing seismic hazard using the spatially smoothed approach. OHAZ (Oriented HAZard) is a program for computing seismic hazard from distributed seismicity. The version OHAZ 2.0 was developed in Java (Zabukovec, 2000) and includes a userfriendly interface, while a more generalized version, OHAZ 4.0, is written in C Builder to allow faster computation (it has not yet been published). OHAZ emphasizes the spatially smoothing approach, but it can also manage the seismicity ascribed to regional or local areal seismogenic zones with or without neighborhood. It can model and maintain maps of spatially smoothed seismicity, taking into account a simple seismotectonic model. Distribution of Past Epicenters The estimation of the hazard is most often derived from the distribution of past earthquake epicenters, given as latitude and longitude coordinates in one or more complete subcatalogs. Usually, this information is shown in the epicenter map where epicenter location errors are not considered. For distributed seismicity and for historical earthquakes, the location uncertainty may be quite considerable. In catalogs, many earthquakes with epicenters from a larger area may even have the same coordinates, which represent large settlements that historical earthquakes have been ascribed to. Both source zoning and smoothing approaches of seismic activity rate modeling take this into account. In the seismic source zone approach, earthquakes may be subjectively classied into a particular source zone and are uniformly smoothed throughout the source zone area. In the one-stage circular smoothing approach, location error is encountered in the subjectively dened radius of smoothing (Frankel, 1995; Lapajne et al., 1997), while the estimated maximum location error directly denes the rst-stage radius in the two-stage smoothing procedure (Lapajne, 2000). Seismicity Models: Models of Seismic Activity Rate It is commonly assumed that future earthquakes will occur in the vicinity of past earthquakes. The input for hazard calculations is the model of seismic activity rate, derived from the distribution of past epicenters and from seismotectonic knowledge. Earthquakes occur in (known or unknown) seismogenic zones; thus some seismotectonic information is already indirectly contained in the catalog. However, we may often benet from additional seismotectonic informa-

Probabilistic Seismic Hazard Assessment Methodology for Distributed Seismicity

2505

tion, a map or a model that is not included in the earthquake catalogs. We would also like to clarify the steps in the procedure and avoid subjective decisions as much as possible, or at least make them easily tracked. In seismic source zone modeling, seismotectonic information is considered subjectively in the determination of source geometry, depending therefore on the authors knowledge. The one-stage circular smoothing procedure is based only on the earthquake catalog. Thus, it is a historic method. Seismic source zone modeling and the one-stage circular smoothing procedure do not explicitly dene the distribution of past epicenters that would include location error. In the two-stage smoothing approach, such a distribution of past epicenters is obtained with circular smoothing of the seismic activity rate. Further elliptical smoothing, based on a seismotectonic model, produces a seismicity model (i.e., model of seismic activity rate), used as the main input for the calculation of seismic hazard. Multimodel Approach Due to the uncertainty in all kinds of data, more than one seismicity model is usually proposed. Different models may reect different earthquake subcatalogs, different authors (teams or single experts), or alternatives in parameters, organized in the logic tree. Usually there is more than one complete subcatalog, which enables different models of distributed seismicity. Rare large earthquake areas often do not coincide with the areas of more frequent smaller earthquakes. Subcatalogs and the corresponding models have a different time span, different lower bound magnitude, and corresponding seismic activity rate, while the upper bound magnitude and the decay rate may or may not remain the same. To be comparable, models should be normalized to the same total annual seismic activity rate (above the same magnitude and in the same area). Due to the most accurate and complete data, the subcatalog of the latest time span is usually the most appropriate base for the reference seismic activity rate above the chosen threshold magnitude in the given inuence area. Even when only one subcatalog is given, we may develop various seismicity models to consider the uncertainty in parameters and in source zone geometry. Subjective determination of alternative parameter values should encompass an uncertainty range in seismological, geological, and tectonic data. Such parameter alternatives, as well as global alternatives of different authors, are usually modeled in a logic tree. Typical parameters, for which uncertainty is considered in this way, are source zone coordinates, seismic activity rate, b-value, upper bound magnitude, and attenuation model. As earthquakes from distributed seismicity cannot be assigned to particular faults, a multimodel approach is even more advisable. Different subcatalogs as well as the logic tree procedure may be applied to the classical source zone approach and to all variants of smoothing. However, multimodeling of source zones is usually based on the logic tree

approach and is performed by different authors, while smoothing approaches are usually based on different subcatalogs. For a given return period, civil engineers usually want to have only one site estimate of a given ground-motion parameter. The nal result of a multimodel approach should be one estimate of seismic hazard, which is either obtained from a compromised seismotectonic model and seismic activity rate or, more often, calculated as a weighted average of different estimates. Different approaches give different nal results: 1. From a compromised model only one seismic hazard estimate is calculated. An example of a seismic hazard map based on the compromised seismotectonic model and the corresponding seismic source zone model is the latest published seismic hazard map of Italy (Slejko et al., 1998). 2. For each model of seismic activity rate or for each selection of parameter values, the corresponding seismic hazard estimate is calculated, and their weighted estimate is the nal result. Many seismicity-smoothing applications are good examples of this (Lapajne et al., 1997; Frankel et al., 2000; Sket Motnikar et al., 2000). A prototype of such weighting in the seismic source zone approach is the study of seismic hazard for 69 sites of nuclear power plants in the central and eastern United States (EPRI, 1988; Bernreuter et al., 1989). Both approaches can be applied to a single site estimate and to a map (grid of sites). Distribution of Released Seismic Energy: EnergyBased Seismic Activity Rate Thus, I conclude that, while proximity to previous small earthquakes is generally an indicator of areas where larger earthquakes are likely to occur, small earthquake seismicity alone is not sufcient to identify areas of concern regarding the occurrence of larger earthquakes (Kafka, 1999). From this quote it follows that larger historical earthquakes, not registered in the complete subcatalogs, need to be considered. The problem is how to handle old strong historical events. Even for well-documented data, the time span of the complete catalog cannot be extended over centuries. Sometimes, the activity rate for larger earthquakes can be determined from geological information, for example, from fault slip rates. Such detailed knowledge is often not available. To take into account the hazard from large reported historical earthquakes, which occurred centuries ago, we introduced an approach for the estimation of the seismic activity rate based on the released seismic energy. The area of large released seismic energy coincides with the epicentral area of a big earthquake. It is assumed that all big earthquakes are registered in the catalog time span. In this sense, the catalog is complete. The exact level of completeness (lower bound magnitude) does not need to be determined,

2506 as the released seismic energy from small and moderate earthquakes can be neglected. The distribution of the released seismic energy will be used to develop additional models of seismic activity rate (distribution of epicenters). For this purpose, the released seismic energy of a given area has been transformed to a corresponding normalized activity rate. The released seismic energy E has a simple empirical relation to the earthquake magnitude (Gutenberg and Richter, 1956a, 1956b; Richter, 1958; Bath, 1973; Willmore, 1979): logE Am B.

J. Lapajne, B. Sket Motnikar, and P. Zupancic

ln u

f(m, d)

e,

(6)

(3)

The regression coefcient A is sometimes estimated to be 1.5 (Richter, 1958; Willmore, 1979; Reiter, 1990). Coefcient B is not important for further evaluation. With the assumption that earthquakes t ideally the log-linear magnitudefrequency law (equation 1a), an energy-based seismic activity rate n(m0) is obtained (Lapajne et al., 1997): n(m0) (A b) 10A(mR b (10(A b)(mu m0)
m0)

where d is the nearest epicentral or fault distance to the site and e is a normally distributed error with standard deviation r. When the territory of interest is divided into grid cells, centers of grid cells may be taken as seismic point sources. As mentioned earlier, epicenters in a cell are assigned to its center. Epicentral distance is dened from the site to the center of the corresponding grid cell. On the other hand, each cell center can be assumed to be the center of a hypothetical fault. Its type, direction, and length are determined from the described quantitative seismotectonic model (distribution of faults). Fault distance denes the distance from the site to the nearest point of the hypothetical fault (see Fig. 1). Equation (6) may therefore be used also with coefcients that are determined for fault sources.

Circular and Elliptical Smoothing of Seismic Activity Rate


The Original Circular Smoothing Procedure The method of spatially smoothed seismicity (Frankel, 1995; Frankel et al., 2000) is very suitable for regions where detailed knowledge of the related seismotectonics is not available (distributed seismicity) and where the delineation of seismic source zones involves too much subjectivity. Geometrical delineation of seismic source zones is not a mathematical procedure. It is based on subjective engineering judgment. On the other hand, the mathematical character of smoothing automatically reduces subjectivity. Circular Gaussian smoothing of seismic activity rate ni(m0) is applied with nj(m0)e
(Dij /c)2

1)

(4a)

or, generalized for the doubly truncated exponential law (equation 1b) case: n(m0) (A b) 10A(mR b (10(A
m0) b)(mu

(1
m0)

10 b(mu 1)

m0)

, (4b)

where mR denotes the magnitude of an earthquake that would have all the released seismic energy in the given time span: mR 1 log A
N

10Amk .
k 1

(5)

Note that equations (4a) and (4b) are valid either for the computation of the seismic activity rate for a small area (a cell of a grid or a source zone) or for the entire region. An arbitrary complete subcatalog or the whole earthquake catalog may be considered. The calculated energy-based seismic activity rate is sensible only for very large historic earthquakes, so the completeness of the whole catalog for smaller earthquakes is not required. Distributions of past epicenters and seismicity models that are based on the described transformation from the released seismic energy can be used together with ordinary seismicity models in a multimodel approach. Attenuation Model, Epicentral, and Fault Distance To estimate seismic hazard for a given site, besides the seismicity model also an attenuation model of a chosen ground-motion parameter u (e.g., peak ground acceleration [PGA], peak ground velocity [PGV], pseudo spectral acceleration [PSA], or pseudo spectral velocity [PSV]) should be known. The general form of an attenuation equation is

ni(m0)

, e
j (Dij /c)2

(7)

where c is a correlation distance and Dij is the distance between the ith and jth cells. In each grid cell i, the seismic activity rate ni(m0) is counted from the earthquake catalog or, as suggested earlier, calculated from the released seismic energy (equation 4a or 4b). The radius of smoothing equals 3c. In the original procedure the correlation distance, and consequently the radius of smoothing, was dened more or less arbitrarily following certain qualitative criteria. The smoothing comprised the error of the epicenter location and the assumption that future earthquakes will take place in the vicinity of past epicenters. A smoothed map of the activity rate represents the seismicity model. Two-Stage Smoothing Procedure To precisely determine various sources of smoothing and to estimate the radius in a qualitative way, the smoothing

Probabilistic Seismic Hazard Assessment Methodology for Distributed Seismicity

2507

grid cell

grid cell
w1 w2 w3

fault ruptures for a given magnitude

epicentral distance o site

weighted fault distances o site


Epicentral and fault distance from the site.

Figure 1.

procedure is separated in the following two stages (Lapajne, 2000). In the rst stage, the seismic activity rate is smoothed circularly as in the original approach (equation 7), but the radius of smoothing is the function of epicenter location error. Simply, the radius may equal the maximum estimated location error in the considered subcatalog. The circularly smoothed map of the activity rate represents the distribution of past epicenters in which the location error is considered. In the second stage it is assumed that earthquakes occur on faults or in fault zones of past earthquakes. Although there is for distributed seismicity no direct relationship between the known faults and earthquakes, the empirical relationship may be applied to obtain the hypothetical fault length. The type and the direction of faults can be determined statistically from the geological data. In the quantitative seismotectonic model (Poljak et al., 2000b), fault types and fault direction are determined in selected regions. Seismic activity rate is further smoothed according to the determined directions of seismogenic faults. For this purpose, fault-rupture-oriented elliptical Gaussian smoothing is proposed: nil(m0) e
j r r 1/2 d ilTVTRV d il r r , 1/2 dijTVTRV dij

ing goes to 3r and to 3s, respectively. The matrix V determines the direction of the rst principal axis of the ellipse and is dened with azimuth : V cos sin sin cos .

(10)

The rst principal axis lies in the direction of the seismogenic fault. It is reasonable to dene its length proportionally to the fault rupture length L (or even simpler: r L). Fault length may be calculated from the estimated upper bound magnitude mu: log L a1 b1mu ,

(11)

where a1 and b1 are regression coefcients, which depend on the fault type and may be determined from empirical relationships (e.g., Wells and Coppersmith, 1994). The axis s represents the width of a seismogenic zone and is assumed to be proportional to L: s kL, k 1. The circularly and elliptically smoothed activity rate map represents the seismicity model. Equivalent One-Stage Smoothing Procedure When the smoothing procedure is based only on the earthquake catalog, an equivalent one-stage circular Gaussian smoothing (Lapajne, 2000) is an alternative to the original circular smoothing procedure (Frankel, 1995). When no statistical seismotectonic model is available, this procedure approximately substitutes the two-stage smoothing approach. It enables one to objectively determine the radius of one-stage circular smoothing. It is assumed that the equivalent-smoothing surface approximately equals the surface of the two-stage smoothing area, which is approximately equal to the surface of the ellipse with principal axis re c r and se c s (see also Fig. 2). Consequently, the equivalent correlation distance ce is determined with c2 e rese

(8)

where nil(m0) is the number of earthquakes in cell i that is during the elliptical smoothing shifted to cell l. Indexes l and j denote all cells in the smoothing area around cell i. T means the transposition and the vector rij denes the distance from d cell i to cell j. The correlation matrix 1 r2 0 0 1 s2

(9)

denes the length of the rst (r) and the second (s) principal half-axis of the ellipse of smoothing. Their meaning is analogous with the correlation distance c; the elliptical smooth-

(12)

2508

J. Lapajne, B. Sket Motnikar, and P. Zupancic

Circular smoothing

Elliptical smoothing

C Ce

Two-stage smoothing

Equivalent circular smoothing


Smoothing areas and radii.

Figure 2.

and therefore depends on the maximum estimated location error (given in the catalog) and on the fault rupture length L. Its evaluation from equation (11) requires the value of the upper bound magnitude mu and the fault type. When the type is not known, either the most common structure or the average of different types may be assumed. The worst case may also be taken into account. Once the correlation distance is determined, the procedure is the same as the original onestage circular smoothing procedure.

P[u

u0 | m, r]

lnu0 1 U* 2

lnu(m, r) r 2

(14)

Seismic Hazard Calculation


The expected annual rate of exceedance of groundmotion level u0 at a site is usually calculated using (e.g., Reiter, 1990) k(u u0)
i mu

where U* represents the complementary error function and ln u is an attenuation model, for example, given by equation (6). For small values, the annual rate of exceedance is practically equal to the annual probability of exceedance. After calculating values k for several levels of ground motion u0, the ground-motion value for a given probability of exceedance can be obtained with the interpolation of k values. For point sources, equation (13) has a simplied form: k(u u0) ni(mmin)
i mu

(15)
u0 | m, ri ] pi (m) dm.

P[u ni(mmin) P[u


mmin r

(13)

mmin

u0 | m, r]pi(m)pi(r) dr dm,

where ni(mmin) is the possibly smoothed annual seismic activity rate above the starting magnitude mmin m0 in a seismic source i, P[u u0 | m, r] is the conditional probability that an earthquake of magnitude m at a distance r from the site produces a ground-motion level u which is greater than level u0, pi(m) is the probability density function of magnitude (2a) or (2b) within source i, and pi(r) is the probability density function of the distance between the point of source i and the site, for which the hazard is being estimated. The conditional probability P[u u0| m, r] is calculated from

Equation (15) may be used for a grid of point sources (source i represents the center of the grid cell), thus it is applicable for an arbitrary geometrical presentation of epicenter distribution. When grid cells are small enough, equation (15) gives practically the same results for areal sources as equation (13), but the computation time with equation (13) is shorter. On the other hand, signicant advantages of equation (15) are its simplicity and its applicability, irrespective of the type of geometry of the seismic activity rate model. Equation (15) may also be used for fault sources, when attenuation model (6) with fault distance is given. In this case, the center of the ith grid cell is the center of a hypothetical fault, and distance ri is determined from the site to the closest point on that fault.

Probabilistic Seismic Hazard Assessment Methodology for Distributed Seismicity

2509

The proposed PSHA methodology for distributed seismicity is described with a ow chart in Figure 3.

ical knowledge, most seismic activity may be considered as distributed seismicity. Input Data

Case Study: Seismic Hazard Map of Slovenia


To illustrate the PSHA methodology for distributed seismicity, a seismic hazard map of Slovenia is calculated using two-stage smoothing of the seismic activity rate. In Slovenia, no large earthquakes of magnitude 7 or above have been observed, and no evidence of characteristic earthquakes has been found. According to current geological and seismolog-

The earthquake catalog covers the period of 5671998 and an area between 12.5 and 17.5 E and 44.5 and 47.5 N of approximately 100,000 km2. The borders of this area are approximately 100 km away from the international borders of Slovenia. The catalog consists of 4687 earthquakes (2426 mainshocks and 2261 foreshocks or aftershocks). Foreshocks and aftershocks are removed from the catalog as

start

earthquake catalogue

dividing region into grid of cells

seismotectonic knowledge

determining subcatalogues

classifying epicenters into grid cells

geometrical seismicity modeling smoothing

zoning

delineation of source zones

statistical seismotectonic model

NO

equivalent circular smoothing

YES epicenter location error radius circular smoothing distributions of past epicenters

statistical seismotectonic model

elliptical smoothing

models of expected seismicity

attenuation model

hazard calculation

seismic hazard estimates / maps

weights

weighted seismic hazard estimate / map

Figure 3.
seismicity.

A ow chart of the proposed methodology for the PSHA of distributed

2510 the Poissonian process is assumed. Most events are dated from the last two centuries, while older recordings are of increasingly inferior quality and limited only to large earthquakes. In the observed area, the doubly truncated exponential recurrence relationship (equation 1b) ts very well the data from the last 120 years above magnitude 3.7 (Lapajne and Sket Motnikar, 1996). Five seismicity models have been proposed (Sket Motnikar et al., 2000). Two of them are based on counting earthquakes from two complete subcatalogs. The catalog may be considered complete from 1880 for m0 3.7 (M1, reference subcatalog) and from 1690 for m0 5.0 (M2). An upper bound magnitude of 6.5 for both subcatalogs is estimated as recommended by Lapajne and Sket Motnikar (1996): the maximum observed magnitude in 16901998 is increased by 0.2. The decay rate b of the magnitudefrequency relationship (equation 1b) is estimated from the reference subcatalog to 0.84 (MLE), and this value is also adopted in the

J. Lapajne, B. Sket Motnikar, and P. Zupancic

M2 model. The maximum location error for M1 is estimated to 22 km and for M2 to 33 km. Three additional models (M1e, M2e, and M3e) of the energy-based seismic activity rate are proposed: two for both complete subcatalogs and one for the whole catalog. Model M1e has the same parameters as model M1 (subcatalog 19801998, magnitude range 3.76.5, b-value 0.84, maximum location error 22 km), and model M2e has the same parameters as model M2 (subcatalog 16901998, magnitude range 5.06.5, b-value 0.84, maximum location error 33 km). Model M3e covers the whole earthquake catalog period of 5671998, with a magnitude range of 3.77.0. The magnitude of the maximum observed historical event (in 1511) was estimated to 6.8. The maximum location error in the catalog is estimated to 56 km, and the MLE of decay rate b is 0.85. The seismic activity rate for all three models is calculated using equation (4b), and it depends only on large earthquakes. For all models, b and mu values are chosen to be constant in the whole observed region.

47.4 47.2 47.0 46.8 46.6 46.4 46.2 46.0 45.8 45.6 45.4 45.2 45.0 44.8 44.6

x 10 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

-2

3.7 <= M < 5.0 5.0 <= M < 6.9


13.0 13.5 14.0 14.5 15.0 15.5 16.0 16.5 17.0 17.5

Figure 4.

Past epicenters and circularly smoothed seismic activity rate for model M1: annual number of past earthquakes (M 4) in a grid cell 5 km 5 km.

Probabilistic Seismic Hazard Assessment Methodology for Distributed Seismicity

2511

47.4 47.2 47.0 46.8 46.6 46.4 46.2 46.0 45.8 45.6 45.4 45.2 45.0 44.8 44.6 13.0 13.5 14.0 14.5 15.0 15.5 16.0 16.5 17.0 17.5 0.56 0.44 0.16 0.52 0.32 0.10 0.33 0.16 0.11 0.30

A1

A2

0.50 0.08 0.42

B1
0.36 0.47 0.17

B3 B2
0.56 0.33 0.11

C1
0.50 0.50 0.24 0.54 0.22

E
0.58 0.29

0.50 0.50

C2

0.13

B4

Figure 5.

Quantitative seismotectonic model of Slovenia. The numbers next to the structural elements represent weights.

The entire observed region is divided into a grid of cells (5 km 5 km). The proposed ve models of seismic activity rate are smoothed in the described two-stage procedure. The estimated maximum epicenter location error is used as the radius of circular Gaussian smoothing. The correlation distance is by denition (Frankel, 1995) one-third of the radius. The circularly smoothed map of activity rate (see example in Fig. 4) represents the distribution of past epicenters with considered location error. A simple seismotectonic model of Slovenia that meets the requirements for the elliptical fault-rupture-oriented smoothing is shown in Figure 5 and was described in Poljak et al. (2000b). The areas of similar seismotectonic characteristics were delineated according to the available seismotectonic maps of Slovenia and neighboring areas. The required attributes of each delineated area have been determined with the following statistical analysis of structures. Structures were divided into 5-km segments and then counted and grouped according to their azimuth of a chosen

span (e.g., 10 ). A few dominant directions were chosen according to their position within the recent maximum principal stress of the region. Their weights were determined by the relative frequency of the selected structure segments. The recent stress regime in the investigated area is compressional with generally northsouthoriented maximum stress axis (Grunthal and Stromeyer 1986; Poljak et al., 2000a). This can produce several structural deformations, that is, the dextral strike-slip movement along northwestsoutheast oriented Dinaric faults, the sinistral strike-slip movement along northeastsouthwestoriented faults, and thrusting along eastwestoriented faults of the southern Alps. Circularly smoothed models of seismic activity rate are further elliptically smoothed and represent seismicity models (see example in Fig. 6). All models are also normalized to the reference model M1. The ve normalized seismicity models are used as input of the seismic hazard calculation (Sket Motnikar et al., 2000). In Figure 7, the corresponding maps for all ve models are presented.

2512

J. Lapajne, B. Sket Motnikar, and P. Zupancic

47.4 47.2 47.0 46.8 46.6 46.4 46.2 46.0 45.8 45.6 45.4 45.2 45.0 44.8 44.6 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

x 10

-2

3.7 <= M < 5.0 5.0 <= M < 6.9


13.0 13.5 14.0 14.5 15.0 15.5 16.0 16.5 17.0 17.5

Figure 6. Past epicenters and two-stage smoothed seismic activity rate for model M1: annual number of expected earthquakes (M 4) in a grid cell 5 km 5 km.

The ground-motion horizontal attenuation model, lnu c1 c2m c3 ln d 2 h2 e,

(16)

with coefcients from Pugliese and Sabetta (1989) and Sabetta and Pugliese (1996), is chosen, where c1, c2, c3, and h are regression parameters. It is developed for all the main ground-motion parameters (PGA, PGV, PSA, and PSV) and for both types of distance, that is, epicentral distance (denoted by PS-e) and nearest fault distance (PS-f). Many accelerograms used in the estimation of coefcients were from the neighboring area (Friuli). The PGA map of Slovenia for rock and rm soil, for a return period of 475 years and for attenuation model PS-f, is calculated using equation (15). The starting magnitude mmin is chosen to be 4.0, as at the given accuracy all lower magnitudes may be neglected. All calculations were obtained using the OHAZ program (Zabukovec, 2000). PGA values in Figure 8 were averaged from ve PGA values ob-

tained from ve seismicity models. For comparative reasons, equivalent one-stage circular smoothing (Lapajne, 2000) and classical seismic source zone modeling (Fajfar et al., 2000) were also performed. The results showed that at least for a rough seismic hazard assessment, the type of geometrical seismicity modeling is not a deciding factor.

Conclusions
The proposed PSHA methodology for distributed seismicity is described with a ow chart in Figure 3. All steps of the improved spatially smoothed seismicity approach are discussed in detail and compared with the classical seismic source zone approach. The following advantages are presented: The two-stage circular and elliptical smoothing procedure and equivalent one-stage smoothing are particularly suitable for distributed seismicity modeling.

Probabilistic Seismic Hazard Assessment Methodology for Distributed Seismicity

2513

Figure 7. (a) Five models of normalized circular smoothed seismic activity rate. (b) Five models of normalized two-stage smoothed seismic activity rate. (c) PGA maps for ve spatial seismic activity models.

2514

J. Lapajne, B. Sket Motnikar, and P. Zupancic

Figure 8.

Peak ground acceleration map of Slovenia.

Statistical modeling of seismogenic faults allows the incorporation of general seismotectonic knowledge and the derivation of the seismicity model. The inuence of large historical events on seismic hazard is enhanced by introducing the calculation of the activity rate from the released seismic energy. The attenuation models based on the closest fault distance may be used even when faults are not determined. With statistical inclusion of seismotectonic data, point sources may be regarded as the centers of fault sources. Using a grid of cells, seismic hazard is calculated from the simplied equation for point sources. An arbitrary geometry of epicenter and fault distribution is allowed.

References
Bath, M. (1973). Introduction to Seismology, Birkhauser, Basel, 395 pp. Bender, B., and D. M. Perkins (1987). Seisrisk III: a computer program for seismic hazard estimation, U.S. Geological Survey Bulletin 1772, Washington, D.C., 48 pp. Bernreuter, D. L., J. B. Savy, R. W. Mensing, and J. C. Chen (1989). Seismic hazard characterization of 69 nuclear power plant sites east of the Rocky Mountains, Lawrence Livermore National Laboratory, NUREG/CR-5250. Berrill, J. B., and R. O. Davis (1980). Maximum entropy and the magnitude distribution, Bull. Seism. Soc. Am. 70, 18231831. Cao, T., M. D. Petersen, and M. S. Reichle (1996). Seismic hazard estimate from background seismicity in southern California, Bull. Seism. Soc. Am. 86, 13721381. Cornell, C. A. (1968). Engineering seismic risk analysis, Bull. Seism. Soc. Am. 58, 15831606. Cornell, C. A., and E. H. Vanmarcke (1969). The major inuences on seismic risk, in Proc. of the Fourth World Conference on Earthquake Engineering, January 1969, Santiago, Chile, A-1, 6993.

Cornell, C. A., and S. R. Winterstein (1988). Temporal and magnitude dependence in earthquake recurrence models, Bull. Seism. Soc. Am. 78, 15221537. dePolo, C. M., and D. B. Slemmons (1990). Estimation of earthquake size for seismic hazards, in Neotectonics in Earthquake Evaluation, Reviews in Engineering Geology VIII, E. D. Krinitzsky and D. B. Slemmons (Editors), Geological Society of America, Boulder, Colorado, 128. Electric Power Research Institute (EPRI) (1988). Seismic hazard methodology for the central and eastern United States, EPRI NP-4726-A, revision 1, Final Report, Palo Alto, California. Fajfar, P., J. Lapajne, I. Perus, B. Sket Motnikar, P. Zupancic, and M. Sostaric (2000). Projektni parametri za potresno odporno projektiranje gradbenih objektov po Eurocode 8, Ljubljana, Univerza v Ljubljani, FGG, IKPIR and MOP, URSG (in Slovene). Frankel, A. (1995). Mapping seismic hazard in the central and eastern United States, Seism. Res. Lett. 66, 821. Frankel, A., C. Mueller, T. Barnhard, E. Leyendecker, R. Wesson, S. Harmsen, F. Klein, D. Perkins, N. Dickamn, S. Hanson, and M. Hopper (2000). USGS national seismic hazard maps, Earthquake Spectra 16, 120. Grunthal, G., and D. Stromeyer (1986). Stress pattern in central Europe and adjacent areas, Gerlands Beitr. Geophys. 95, no. 5, 443452. Gutenberg, B., and C. F. Richter (1944). Frequency of earthquakes in California, Bull. Seism. Soc. Am. 34, 185188. Gutenberg, B., and C. F. Richter (1956a). Earthquake magnitude. intensity, energy, and acceleration, Bull. Seism. Soc. Am. 46, 105145. Gutenberg, B., and C. F. Richter (1956b). Magnitude and energy of earthquakes, Ann. Geos. 9, 115. Hamdache, M., J. K. Lapajne, and B. Sket Motnikar (2001). Seismic hazard assessment in North Algeria, in Proc. of the Workshop Sur la sismicite et la gestion du risque sismique dans la region Euro-Mediterra neenne, Agadir, Morocco, 1819 June, 33. Kafka, A. L. (1999). Testing the hypothesis that locations of larger earthquakes can be forcasted based on the spatial distribution of smaller earthquakes, in Proc. of the 71st Annual Meeting of the Eastern Section of the Seismological Society of America, 1620 October, Memphis, Tennessee.

Probabilistic Seismic Hazard Assessment Methodology for Distributed Seismicity


Kijko, A., and G. Graham (1998). Parametrichistoric procedure for probabilistic seismic hazard analysis. Part I: Estimation of maximum regional magnitude mmax, Pure Appl. Geophys. 152, 413442. Kijko, A., and G. Graham (1999). Parametrichistoric procedure for probabilistic seismic hazard analysis. Part II: Assessment of seismic hazard at specied site, Pure Appl. Geophys. 154, 122. Kijko, A., and M. A. Sellevoll (1989). Estimation of earthquake hazard parameters from incomplete data les. Part I. Utilization of extreme and complete catalogs with different threshold magnitude, Bull. Seism. Soc. Am. 79, 645654. Lapajne, J. K. (2000). Some features of the spatially smoothed seismicity approach, in Proc. of the Workshop Seismicity Modeling in Seismic Hazard Mapping, Poljce, Slovenia, 2224 May, 2733. Lapajne, J. K., and B. Sket Motnikar (1996). Estimation of upper-bound magnitude in earthquake hazard assessment, in Earthquake Hazard and Risk, V. Schenk (Editor), Kluwer, Hingham, Massachusetts, 39 48. Lapajne, J. K., B. Sket Motnikar, B. Zabukovec, and P. Zupancic (1997). Spatially smoothed seismicity modelling of seismic hazard in Slovenia, J. Seism. 1, 7385. McGuire, R. K. (1993). Computation of seismic hazard, Ann. Geos. 36, 181200. Page, R. (1968). Aftershocks and microaftershocks of the great Alaska earthquake of 1964, Bull. Seism. Soc. Am. 58, 11311168. Poljak, M., M. Zivcic, and P. Zupancic (2000a). The seismotectonic char acteristics of Slovenia, Pure Appl. Geophys. 157, 3755. Poljak, M., P. Zupancic, J. K. Lapajne, and B. Sket Motnikar (2000b). Seismotectonic input for spatially smoothed seismicity approach, in Proc. of the Workshop Seismicity Modeling in Seismic Hazard Mapping, Poljce, Slovenia, 2224 May, 117124. Pugliese, A., and F. Sabetta (1989). Stima di spettri di risposta da registrazioni di forti terremoti italiani, Ingegneria Sismica 6, no. 2, 314. Reiter, L. (1990). Earthquake Hazard Analysis: Issues and Insights, Columbia U Press, New York, 254 pp. Richter, C. F. (1958). Elementary Seismology, W. H. Freeman, New York, 768 pp. Risk Engineering (1988). FRISK88 Users Manual, Version 1.2, Golden, Colorado. Sabetta, F., and A. Pugliese (1996). Estimation of response spectra and

2515

simulation of nonstationary earthquake ground motions, Bull. Seism. Soc. Am. 86, 337352. Sket Motnikar, B., J. K. Lapajne, P. Zupancic, and B. Zabukovec (2000). Application of the spatially smoothed seismicity approach for Slovenia, in Proc. of the Workshop Seismicity Modeling in Seismic Hazard Mapping, Poljce, Slovenia, 2224 May, 125133. Slejko, D., L. Peruzza, and A. Rebez (1998). Seismic hazard map of Italy, Ann. Geos. 41, 183214. Stirling, M. W. (2000). A new probabilistic seismic hazard model for New Zeland, in Proc. of 12 WCEE, Wellington, New Zealand 2362. Weichert, D. H. (1980). Estimation of the earthquake recurrence parameters for unequal observation periods for different magnitudes, Bull. Seism. Soc. Am. 70, 13371346. Wells, D. L., and K. J. Coppersmith (1994). New empirical relationships among magnitude, rupture length, rupture width, rupture area, and surface displacement, Bull. Seism. Soc. Am. 84, 9741002. Veneziano, D., C. A. Cornell, and T. OHara (1984). Historic Method for Seismic Hazard Analysis, Electric Power Research Institute, Report NP-3438, Palo Alto, California. Willmore, P. L. (Ed.) (1979). Manual of Seismological Observatory Practice, World Data Center A for Solid Earth Geophysics, Washington, D.C., 165 pp. Woo, G. (1996). Kernel estimation method for seismic hazard area source modeling, Bull. Seism. Soc. Am. 86, 353362. Youngs, R. R., and K. J. Coppersmith (1985). Implication of fault slip rates and earthquake recurrence models to probabilistic seismic hazard estimates, Bull. Seism. Soc. Am. 75, 939964. Zabukovec, B. (2000). OHAZ: A computer program for spatially smoothed seismicity approach, in Proc. of the Workshop Seismicity Modeling in Seismic Hazard Mapping, Poljce, Slovenia, 2224 May, 135140. Environmental Agency of the Republic of Slovenia Seismology Ofce Dunajska 47/VII 1000 Ljubljana, Slovenia (J.L., B.S.M., P.Z.) Manuscript received 22 August 2002.

Potrebbero piacerti anche