Seismic Microzonation via PSHA Methodology and Illustrative Examples

Transcription

1 Seismic Microzonation via PSHA Methodology and Illustrative Examples I.D. Gupta Central Water and Power Research Station, Khadakwasla, Pune A Workshop on Microzonation Interline Publishing, Bangalore Introduction The aim of a seismic zoning study is to delineate the zones of equal seismic hazard, characterized more commonly in terms of the peak ground acceleration. But, for different practical applications, it needs to be defined directly in terms of the response spectrum amplitudes or the site response effects. Depending upon the size of the area under study, the zoning can be performed on two scales, viz. macro or micro scales. Seismic macrozoning maps for a large region or whole of a country are generally prepared on a very broad scale. This shows only the overall distribution of seismic hazard in different parts of a country (e.g., GSHAP, 1999; Adams and Atkinson, 2003; Das et al., 2006), and reflect the presence of major fault systems or of large seismically active zones The characteristics of strong earthquake ground shaking at a site, however, depend on numerous soil and geological features surrounding the site, along with that on the distribution and the level of seismic activity. The microzoning maps of a metropolitan area are able to include such fine details on a local scale (Trifunac, 1990; Todorovska et al., 1995; etc.). The scenario based deterministic zoning (Anderson, 1997) is unable to envelop with any confidence the hazard due to all the earthquakes expected to occur during a given lifetime. This is because different combinations of earthquake magnitude and distance may produce significantly different effects in different frequency ranges at a site. The probabilistic seismic hazard analysis (PSHA) approach provides a means to integrate the effects of all the expected earthquakes by considering their random and uncertain nature and also that of the strong-motion parameter needed to quantify the hazard. As the so called deterministic zoning also cannot ensure absolute safety, it would be more rational to accept a level of the seismic risk that we are prepared to accept in the everyday life (e.g., risk of traffic accident or plane crash, for example). To this end, the PSHA approach can be used to prepare the zoning maps by quantifying the seismic risk. The PSHA approach can be used to prepare the zoning maps on both macro and micro scales by computing the values of a strong-motion parameter of interest at a grid of closely spaced sites covering the entire area of interest. The salient features of the PSHA approach have been described briefly in this paper by drawing attention to several issues of importance. This approach is based on defining the compound probability distribution of a strong-motion parameter at a site due to the total expected seismicity in the region around the site (Cornell, 1968; McGuire, 1977; Anderson and Trifunac, 1978). The basic PSHA approach considers the inherent random uncertainties, also termed as aleatory uncertainties, involved in all the governing input elements. The effect of the various contributing factors can thus be considered simultaneously and in a balanced way. However, this requires knowing the probability distributions of earthquake magnitude, interevent time, and distance from the site, and also that of the strong-motion parameter for any given earthquake magnitude and distance combination. The paper provides the necessary guidelines for defining all of these probabilistic functions in practical applications.

2 108 Microzonation The illustrative examples of microzonation maps presented in the paper show that the probabilistic estimates of ground motion may differ significantly within the same metropolitan area. These differences are related to the variations in distance from various faults and the local geology. It has been shown that a variety of microzoning maps can be prepared using PSHA approach for different applications, such as design and retrofitting, loss estimation, insurance, urban planning, and prediction of earthquake induced soil failures, for example. The PSHA Formulation The basic PSHA formulation considers the inherent random uncertainties in the input quantities and parameters to estimate the expected occurrence rate, υ ( Z > z ), of exceeding a specified value, z, of a random parameter, Z, which is defined by the following generalized expression N > υ ( Z z) = N ( M ) q ( Z z M, R) f ( M ) g ( R dmdr (1) n= 1 n min n > n n ) In this expression, N n ( M min ) is the occurrence rate of earthquakes above a selected threshold magnitude M min, and f n (M ) and g n (R) are the probability density functions of magnitude and distance from a site for the nth seismic source, and the summation is taken over all the N number of sources. Quantity q n ( Z > z M, R) represents the probability of exceeding level z due to magnitude M at distance R. The occurrence rate of earthquakes for a small magnitude intervals ( M j δm j, M j + δm j ) and small distance interval ( R i δri, Ri + δri ) in the nth source zone can simply be defined as M j + δm j Ri + δri λ ( M, R ) = N ( M min ) f ( M ) g ( R) dmdr (2) n j i n n M j δm j Ri δri This can be expressed in terms of the probability distribution functions F(M) and G(R) of magnitude and distance as λn ( M j, Ri ) = Nn ( M min ) { F( M j + δm j ) F( M j δm j )} { G( Ri + δri ) G( Ri δri )} (3) With this, the expression of equation (1) can be written in a fully discrete form suitable for numerical computation as follows N ( Z > z) = J I υ q ( Z > z M, R ) λ ( M, R ) (4) n= 1 j= 1 i= 1 n Assuming the occurrence rate λ n ( M j, Ri ) to follow a Poisson probability distribution, the occurrence rate υ ( Z > z), which is a linear combination of λ n ( M j, Ri ), can also be described by a Poisson probability distribution. Thus, the probability of Z > z due to all the earthquakes in all the sources during an exposure period of Y years can be defined by P Z > z Y = 1 exp Y υ( Z > z) (5) ( ) { } The Poisson assumption is generally violated in that the large characteristic earthquakes in seismically active areas may follow a long-term cyclic behavior with time varying occurrence rate. Such events are required to be described by a real-time renewal model, wherein the occurrence rate is small soon after a large earthquake and increases with the lapse of time since the last such event (Rikitake, 1976). Their occurrence rate has to be defined using a hazard function based on the probability distribution of inter-event times (Jara and Rosenblueth, 1988). The PSHA formulation of equation (5) for the j i n n j i

3 Seismic Microzonation via PSHA 109 stationary Poisson processes has been shown to be applicable to such events also, if their average occurrence rate is obtained using a time-dependent hazard function (Lee, 1992). But, even after considering the time dependence to estimate their average occurrence rate, the expression of equation (5) is not applicable for the dependent events like aftershocks and sequential earthquakes. To include the contribution of aftershocks in the PSHA, it is necessary to decluster the available earthquake catalog (e.g. Maeda, 1996) and describe the aftershocks by a Poisson process with time varying occurrence rate or by some other suitable model (e.g., Corral, 2004). The Monte Carlo simulation is then used to generate a large number of earthquake catalogues of Y years duration and perform the time-taking hazard calculations for each catalogue (Beauval et al., 2006). For convenience and computational efficiency in real applications, it is proposed to account for the effect of aftershocks by assuming them to occur in a literal way. If η( M l, R k ; Y ) is the total number of aftershocks in Y years in a small magnitude interval around central magnitude M l and in a small distance interval around central distance R k, the probability of Z > z due to these events to occur in a deterministic way can be defined as (Todorovska et al., 1995) K L * P ( Z > z Y ) = 1 exp ln(1 q( Z > z M l, Rk ) η ( M l, Rk ; Y ) (6) k = 1 l= 1 Any other type of events, not following the Poisson distribution and occurring in a literal way (e.g., earthquake prediction), can also be included in P * ( Z > z Y ) by including their numbers in η ( M l, R k ; Y ). Thus, by carrying out the hazard analysis for the declustered catalog of the main shocks using the expression of equation (5), the combined probability of Z > z from both the Poisson and non Poisson type of earthquakes can be defined as * P ( Z > z Y ) = 1 exp Yυ ( Z > z) 1 P ( Z > z Y ) (7) { } { } The plot of the probability P ( Z > z Y ) versus z is commonly known as the hazard curve, which can be used readily to estimate the value of a hazard parameter with any desired probability of exceedance in a specified exposure period. The 10% probability of exceedance in 50 years is used commonly for preparing the zoning maps for design of ordinary residential buildings. Whereas, 2% in 50 years is recommended for more important structures (FEMA, 2004). The hazard curve are finally calculated by integrating over all the magnitudes and distances in all the source zones using the various inputs described as above. To carry out the PSHA for frequency dependent quantities like response spectra, one hazard curve is required to be defined for each frequency. Implementation of PSHA The foregoing basic PSHA formulation is able to consider the effects of only the random aleatory uncertainties in specifying the magnitude, location and occurrence time of earthquakes, and that in the estimate of the strong-motion parameter of interest at a site for specified magnitude and distance combination. Though it refers to the seismic hazard evaluation for a single site, seismic zoning maps can be prepared easily by estimating the values of a strong-motion parameter at a large number of closely spaced sites covering an entire area. The PSHA method accounts for the effect of all the expected earthquakes within about 300 km distance from any location in the area to be zoned. The various steps involved in implementation of PSHA for seismic zoning can broadly be listed as: identification of all the source zones which can affect any of the locations in the area; and for each source zone, estimation of the total expected seismicity with the probability distribution of the magnitude and distance, and selection of a suitable probabilistic attenuation relationship for the strongmotion parameter as a function of magnitude and distance.

4 110 Microzonation In an ideal situation, all the seismic sources have to be individual faults. However, due to insufficient knowledge about the faults and poor correlation of the past seismicity with the known faults, broad area sources (encompassing several faults) and the tectonic provinces (covering large geographic areas of diffused seismicity with no identifiable active faults) are used commonly in real practice. The observed seismicity is sometimes seen to be highly concentrated in a very small area. If located far away from a site of interest, this can be defined by a point source. The commonly used types of seismic source include: the Point, Line, Dipping Plane, and Area sources (Gupta, 2007). Expert knowledge, detailed familiarity with the geology in the area, interpretation and judgment play important role in defining the seismic sources. A typical example of delineation of two area sources A1 and A2 of diffused seismicity and two line sources L1 and L2 is shown in Figure 1. L1 L2 A1 Latitude, N Site A2 Longitude, E Figure 1 Typical examples of area and line types of seismic source. To estimate the total seismicity for a seismic source, defined by the number N n ( M min ) of earthquakes above a threshold magnitude M min, the log-linear frequency-magnitude relationship due to Gutenberg and Richter (1944) is fitted for each source zone using the available past earthquake data (Weichert, 1980), or using the information about geological slip-rates (Wallace, 1970; Gupta, 2002). A magnitude distribution function, F(M), is also defined for each source to distribute these numbers among different magnitude intervals between M min and a maximum magnitude M max. The exponentially decaying magnitude distribution is generally found suitable for this purpose. However, a characteristic earthquake model (Youngs and Coppersmith, 1985) is found more suitable for individual faults. Both these distributions can be defined with a constant seismicity rate or a constant moment release rate (Gupta, 2007). To obtain N n ( M min ) and F(M), it is necessary to convert the available data into a common magnitude scale, remove the dependent events, and to account for the incompleteness of lower magnitude earthquakes for older periods in the available catalogue. Typical examples of the exponential and characteristic magnitude distribution are shown in Figure 2.

5 Seismic Microzonation via PSHA 111 Figure 2 Probability distribution function of earthquake magnitude Next, the probability distribution function, G(R), of the source-to-site distance, R, is defined for each seismic source. Though analytical expressions are available for the distribution G(R) for some simple fault geometries and idealized area type of sources (Cornell, 1968), numerical estimation is necessary for more complicated real sources. For this purpose, a source is divided into a large number of small size elements, and the geometric center of each element is assumed to be the epicentral location. Any desired measure of the source to site distance (epicentral distance, closest distance to fault rupture, closest distance to surface projection of fault rupture, etc.) from each source element can be defined from knowledge of the fault geometry or by assuming random distribution for the fault rupture (Gupta, 2006). Distances from all the source elements are then used to obtain the observed probability distribution using suitable kernel function (Silverman, 1986). The distribution of the distance can also be defined by suitable spatial averaging of the distribution of observed past seismicity in a source zone (Frankel, 1995; Das et al., 2006). Figure 3(a) shows an example of the distributions for the epicentral distance and the closest distance to fault rupture for the line source L1 in Figure 1 using uniformly distributed seismicity. Also, the distance distributions for the area source A1 in Figure 1 using uniform and actual spatial distribution of seismicity are shown in Figs. 3(b) and 3(c), respectively. (a)

6 112 Microzonation (b) (c) Figure 3 Probability distribution of distances for (a) line source with uniform spatial distribution, (b) area source with uniform spatial distribution, and (c) area source with actual spatial distribution of seismicity Lastly, a suitable attenuation relationship providing a probabilistic description of the strong-motion parameter is required to be selected for the estimation of the probability q n ( Z > z M, R) for each source. Generally, the same attenuation relation is applicable for all the source zones. But, different attenuation relations may be needed in some situations (source zones representing shallow crustal and deep subduction earthquakes, for example). The attenuation relation should provide the median estimate and the corresponding probability distribution of the residuals for specified earthquake magnitude, source-to-site distance, and site geologic and soil conditions. An attenuation is commonly developed by fitting a simple equation in terms of a limited number of earthquake and site parameters to the values of strong-motion parameter observed during past earthquakes. The median attenuation relation thus fitted is seen to be associated with large scattering of the observed data around it. The residuals between the observed and the median estimates are described by suitable probability distribution function, making the attenuation relation of probabilistic nature. However, due to limitation of data and the simplifying assumptions made, it is difficult to define the median attenuation relation as well as the probability distribution of the residuals in a unique way (Gupta, 2005). The uncertainties leading to multiple choices for the various inputs to the basic PSHA are termed as epistemic type of uncertainties. Due to such uncertainties, there could be several possible choices for the definition of seismic source zones and distribution of distance, type of earthquake recurrence model and the maximum magnitude for each source, as well as for the attenuation relationship for the hazard parameter of interest (Gupta et al., 2005). The various possible sets of input to the PSHA are subjectively assigned appropriate weights on the basis of the confidence imposed in each of them such the total weight for all the sets considered for a particular input is unity. The logic-tree method (Kulkarni et al., 1984) is then utilized to account for the effect of the epistemic uncertainties. A typical logic-tree depicting the possible uncertainties in the various elements of the basic PSHA is shown in Figure 4 for the purpose of illustration.

7 Seismic Microzonation via PSHA 113 Source Zones Seismicity Data Recurrence Relation Moment Rate Maximum Magnitude Spatial Distribution Attenuation Model Set-1 Set-2 Set-3 (0.2) Catalog-1 (0.6) Catalog-2 Constant Seismicity (0.5) Constant Moment (0.5) Large Average (0.6 small (0.3) preferred large (0.3) Uniform Past Seismicity (0.6) model-1 model-2 model-3 (0.2) Figure 4 Logic-tree to organize the epistemic uncertainties in the PSHA approach. The basic principle to be followed in setting up a logic-tree is that the branches emanating from a single node should cover only the physically realizable distinct possibilities, which may lead to significantly different estimate of the hazard. In the logic-tree of Figure 4, three sets of source zones with different weights may result from different interpretations and subjective judgments for a given database on seismotectonics and geological features in the region of interest. Two different sets with weights of 0.6 and 0.4 for the past earthquake catalog is the second element of the logic-tree, which may result from the availability of several catalogs prepared by different organizations or use of different methods for homogenization of magnitudes in a given catalog. Two options with equal weights are shown for the two different types of recurrence relationship. Further, two different moment release rates are considered in the recurrence relationship with constant moment rate. The next element in the logic tree is the maximum magnitude, for which three options as small, large and preferred with weights of 0.3, 0.3 and 0.4 are considered for each source zone. The spatial distribution of seismicity in a source zone is considered in two different ways as uniform distribution and that based on spatially smoothed past seismicity. Finally, there are three different options for the ground motion attenuation model with weights equal to 0.4, 0.3 and 0.3. The example logic-tree in Figure 4 has a total of 324 end branches, which is given by the product of the number of different options for each input element. The weight for an end branch is given by the product of the weights of all the intermediate branches leading to that branch. To account for the effect of the epistemic uncertainties, the basic PSHA is performed for all the combinations of the input leading to various end branches, and the resulting hazard curves are assigned the corresponding weights. These can be used to define the mean or the median hazard curve, as well as the hazard curves with any desired confidence level. As the result of the basic PSHA with a suitable confidence level (e.g., 98% in 50 years) is already able to ensure the required conservatism, it is proposed to use only the median hazard curve to arrive at the most probable decision in face of the epistemic uncertainties. Examples of Microzonation The PSHA method can be used to prepare the microzonation maps in terms of any strong-motion parameter of engineering importance, including the non-instrumental MMI intensity (Gupta, 1991). For the purpose of illustration, example maps are presented in this section for the response spectrum

8 114 Microzonation amplitudes and the occurrence of liquefaction. The area considered is the Los Angles metropolitan, California, USA. The seismic sources for Los Angeles area are defined by 29 fault segments and a rectangular area of diffused seismicity as shown in Figure 5. Each fault segment is labeled by a serial number followed by two values within parenthesis. The first value is the estimate of the moment release rate, M & 0, in dyne-cm/year, and the second value is the activity rate, a, for an exposure period of 50 years. The b-value for all the faults is taken as 0.86, and the M min and M max as 2.75 and 7.0, respectively. The expected seismicity on each fault segment is assumed to be distributed uniformly over its entire length. The expected number of earthquakes for different magnitude intervals in the diffused rectangular source over a period of 50 years are also indicated in Figure 5, which are assumed to occur uniformly over the entire source area. The major faults in the Los Angeles metropolitan area and the local geological condition in terms of the depth of sediments in kilometers at 5 5 grid points are shown in Figure 6, with grey areas indicating the rock outcrops (8.3E+25, 5.29) (2.8E+24, 4.14) 16 (1.8E+25, 4.94) 24 (2.2E+24, 4.03) 25 (4.6E+23, 3.35) 19 (2.4E+23, 3.07) L atitud e, N orth M n(m) (6.6E+23, 3.51) (1.3E+24, 3.80) 12 (2.5E+22, 2.09) (6.4E+22, 2.50) 8 (6.4E+22, 2.50)11 27 (10.4E+23, 2.71) (2.3E+23, 3.05) 9 10 (4.0E+24, 4.29) 4 (1.7E+23, 2.92) 5 (2.5E+23, 3.09) (1.2E+24, 3.77) 1 23 (9.7E+24, 4.68) 22 (10.0E+24, 3.69) 20 (4.5E+23, 3.34) 26 (4.3E+23, 3.32) 3 (3.9 E + 2 3, 3.2 8) 7 (1.5E+25, 4.55) 18 (9.5E+23, 3.67) 29 (1.5E+25, 4.55) 6 (2.3E+25, 5.05) 2 (8.3E+21, 1.60) 17 (3.6E+23, 3.25) Longitude, West Figure 5 Seismic sources and their potential in the region of Los Angeles metropolitan.

9 Seismic Microzonation via PSHA 115 Figure 6 Rock outcrops (grey areas) and thickness (km)of sediments in Los Angeles metopolitan area. Microzoning in Terms of Response Spectrum Amplitudes The microzonation maps for Los Angeles metropolitan area in terms of the pseudo relative velocity (PSV) spectrum amplitudes at four different natural periods are shown in Figure 7. These maps are based on the empirical scaling relations due to Trifunac and Lee (1985), who have expressed the spectral amplitude, PSV(T), at period T in terms of earthquake magnitude, source-to-site distance, and the site geological condition defined by the depth of sediments in kilometers. Using the probability distribution of the residuals, these scaling relations can be used to obtain the conditional probability, q ( PSV ( T ) M j, Ri ), that spectral amplitude PSV(T) will be exceeded at a site due to earthquake magnitude M j at distance R i from the site. Using these probabilities and the seismicity associated with the 29 fault segments and the diffused rectangular area source, PSHA has been carried out to compute the PSV(T) amplitudes with different confidence levels at each 1 1 grid points. The maps in Figure 6 are for a confidence level of 0.50 and natural period equal to 0.04, 0.34, 0.90 and 2.8 sec. The microzoning maps for many other natural periods and confidence levels are available in Lee and Trifunac (1987). The maps in Figure 7 correspond to the horizontal component of PSV spectrum with 5% damping. By reading the spectral amplitudes from such maps for a series of natural periods and for a particular confidence level and damping value, one can readily construct the complete response spectrum for any site in the area. Examples of such spectra for two typical sites, one on rock and the other on about 7.5 km thick sediments (shown by solid triangles in Figure 6), are presented in Figure 8. It is seen that the difference between the two spectra is drastic, although both the sites are in the same metropolitan area. Similar situation may exist in the Delhi metropolitan area in India, where the seismicity can be associated with several fault segments and the site soil and geological conditions vary widely.

10 116 Microzonation Figure 7 Microzoning maps for Los Angeles metropolitan in terms of 5% damped PSV(T) for T=0.04, 0.34, 0.9 & 2.8 sec with confidence level of 0.50 in 50 years.

11 Seismic Microzonation via PSHA 117 Microzoning for the Occurrence of Liquefaction The study by Trifunac (1995) provides a basis for using the PSHA methodology for preparing the zoning maps for occurrence of liquefaction at various sites in an area of interest. Using 90 worldwide observations, Trifunac (1995) has developed an empirical model to obtain the standard penetration test (SPT) value, N, corrected for the overburden pressure, σ 0, which separates on average the observed cases of liquefaction from those of no liquefaction. Thus the model prediction can be viewed as a critical value, N crit, of N such that liquefaction will occur at a site if the actual N value is smaller than the estimated Ncrit value. Due to the uncertainties associated with the ground motion and the site characterization, the observed data points were found to be scattered randomly around the mean predictions by the model. This scattering is defined using a Gaussian probability distribution as the probability of N crit being greater than the actual N, which can be considered as the probability of liquefaction, In this expression, standard deviation. The Figure 8 Response spectra obtained from the microzoning maps at two different sites shown by solid triangles. in Fig. 6 2 x μ 1 1 N Pr ob. { Ncrit > N} = dx πσ exp (8) 2 2 σ N N N μ is the mean value of the corrected SPT value and σ is the associated N N μ can be obtained from the model of Trifunac (1995) as follows N 1/ υmax dur μ N = (9) 3/ 2 σ0 where, υ max is the peak ground velocity in cm/sec and dur is the strong motion duration, which can be obtained from the empirical relationship due to Trifunac (1976) and Novikova and Trifunac (1993), respectively. For the model of equation (8), the value of σ N is specified by Trifunac (1995) as 5.5.

12 118 Microzonation Thus, the probability distribution defined by equation (8) is equivalent to the conditional probability, q ( liquefaction occurs M j,ri ), that liquefaction will occur at a site with specified N and σ 0 values due to earthquake magnitude M j at distance R i. This provides a basis to use the PSHA formulation for preparing the microzoning maps in terms of the average recurrence periods (reciprocal of annual occurrence rate) for the occurrence of liquefaction for given N and σ 0 values. Two such maps for the Los Angeles Metropolitan for N =10 & 20 and σ 0 = 40 kpa are shown in Figure 9 (Todorovska and Trifunac, 1999). If a site is actually characterized by the N and σ 0 values for which the map is prepared, the corresponding return period gives the period for the liquefaction to occur at that site. By preparing such maps for a large number of N and σ 0 values, it is possible to identify the average recurrence period for liquefaction to occur at any site in the area. Figure 9 Microzoning maps for Los Angeles Metropolitan in terms of return period (years) for occurrence of liquefaction for N of 10& 20 and σ 0 of 40 kpa Figure 10 Microzoning maps for Los Angeles Metropolitan with a confidence level of 0.50 in 50 years in terms of N crit for σ 0 = 40 kpa.

13 Seismic Microzonation via PSHA 119 For a given value of the overburden pressure σ 0, the microzoning maps for the occurrence of liquefaction can also be prepared in terms of the critical SPT value for which liquefaction may initiate at a site with a specified probability during a specified exposure period. A typical microzonation map of this type for Los Angeles metropolitan area is shown in Figure 10 (Todorovska and Trifunac, 1999). By preparing such maps for several different values of σ 0, the occurrence of liquefaction with desired probability and exposure period can be found readily at a site with known N and σ 0 values. Conclusions The seismic microzoning studies based on deterministic scenario earthquake approach cannot be considered to ensure the intended conservatism due to random nature of earthquake events and the frequency dependent nature of the resulting effects at each site. Due to the various uncertainties, the choice of a fixed earthquake scenario may be questionable and difficult to justify. Scenario zoning maps are unable to provide the distribution of hazard over an area with a uniform confidence level. Also, the confidence level associated with the level of strong-motion parameter at different locations remains unknown. The probabilistic seismic hazard analysis (PSHA) is a powerful tool to prepare the microzoning maps in terms of any of the strong-motion parameters of engineering importance by accounting for the randomness and uncertainties in the various input elements. This approach is able to quantify the hazard in a more scientific and rational way by including both the life-period and the probability of occurrence in the formulation. In view of the fact that even without earthquakes some risk is inevitable in day-to-day life, the quantification of seismic risk by PSHA approach can be used to bring the earthquake hazards at par with the risk due to other hazards. Examples of microzonation maps presented for the response spectrum amplitudes and the occurrence of liquefaction have illustrated that PSHA approach is able to exhibit in a physically realistic way the effects of the spatial distribution of seismicity and site geological condition. In reality, the ground motion amplification effects at different frequencies is a trade off among the effects of earthquake magnitude, source-to-site distance, fault rupture directivity, local site condition (top 30 m), and the site geologic condition (deep sediments). A possible indication in support of this fact is provided by the lack of a universal matching between the amplification effects inferred from microtremor studies (which incorporates only shallow site effects) and those observed during real earthquakes. The PSHA approach only is able to account for the amplification/deamplification effects due to all the factors simultaneously and in a balanced way. Several studies present microzoning maps in terms of only site response parameters like amplification factor or the predominant period. Such maps are of no direct use for any practical engineering applications, and they are unable to provide an accurate and reliable estimate of the strong motion parameters required for characterizing the seismic hazard. The PSHA approach, on the other hand, is able to provide directly the estimate of any of the desired strong motion parameters needed for structural response analysis or the risk estimation. Though, the illustrative examples are shown for response spectrum amplitudes and the occurrence of liquefaction, the PSHA formulation can readily be used to consider many other parameters like frequency-dependent strong motion duration, Fourier amplitude spectrum, surface dislocation across faults, and also the non-instrumental MMI levels required for vulnerability analysis. References 1. Adams, J. and Atkinson, G.M. (2003). Development of Seismic Hazard Maps for the 2005 National Building Code of Canada, Canadian Jour. of Civil Eng., Vol. 30, pp Anderson, J.G. (1997). Benefits of Scenario Ground Motion Maps, Engg. Geol., Vol. 48, pp

14 120 Microzonation 3. Anderson, J.G. and Trifunac, M.D. (1978). Uniform Risk Functionals for Characterization of Strong Earthquake Ground Motion, Bull. Seism. Soc. Am., Vol. 68, No. 1, pp Beauval, C., Hainzl, S. and Scherbaum, F. (2006). Probabilistic Seismic Hazard Estimation in Low-Seismicity Regions Considering Non-Poissonian Seismic Occurrence, Geophys. Jour. Int., Vol. 164, No. 3, pp Cornell, C.A. (1968). Engineering Seismic Risk Analysis, Bull. Seism. Soc. Am., Vol. 58, pp Corral, A. (2004). Long-Term Clustering, Scaling and Universality in the Temporal Occurrence of Earthquakes, Phys. Rev. Lett., Vol. 92, Das, S., Gupta, I.D. and Gupta, V.K. (2006). A Probabilistic Seismic Hazard Analysis of Northeast India, Earthq. Spectra, Vol. 22, No. 1, pp FEMA (2004). NEHRP Recommended Provisions for Seismic Regulations for New Buildings and Other Structures (FEMA 450), Part 1: Provisions, 2003 Edition. 9. Frankel, A. (1995). Mapping Seismic Hazard in the Central and Eastern United States, Seism. Res. Lett., Vol. 66, No. 4, pp GSHAP (1999). The Global Seismic Hazard Assessment Program (GSHAP) , Annali Di Geofisica, Vol. 42, No. 6, pp (Summary volume edited by D. Giardini). 11. Gupta, I.D. (1991). A Note on Computing Uniform Risk Spectra from Intensity Data on Earthquake Occurrence, Soil Dyn. and Earthq. Eng., Vol. 10, No. 8, pp Gupta, I.D. (2002). The State of the Art in Seismic Hazard Analysis, ISET Jour. of Earthq. Tech., Vol. 39, No. 4, pp Gupta, I.D. (2005). Probabilistic Seismic Hazard Analysis With Uncertainties, Procs. Symp. Seismic Hazard Analysis and Microzonation, September, Roorkee, Vol. I, pp Gupta, I.D. (2006). Defining Source-to-Site Distances for Evaluation of Design Earthquake Ground Motion, Procs. 13 th Symp. on Earthq. Eng., December 2006, IIT Roorkee, Vol. I, pp Gupta, I.D. (2007). Probabilistic Seismic Hazard Analysis Method for Mapping of Spectral Amplitudes and Other Design Specific Quantities to Estimate the Earthquake Effects on Manmade Structures, ISET Jour. of Earthq. Tech., Vol. 44, No. 1, Paper No Gutenberg, B. and Richter, C.F. (1944). Frequency of Earthquakes in California, Bull. Seism. Soc. Am., Vol. 34, No. 4, pp Jara, J.M. and Rosenblueth, E. (1988). Probability Distribution of Times Between Characteristic Subduction Earthquakes, Earthq. Spectra, Vol. 4, pp Kulkarni, R.B., Youngs, R.R. and Coppersmith, K.J. (1984). Assessment of Confidence Intervals for Results of Seismic Hazard Analysis, Procs. 8 th World Confs. on Earthq. Eng., San Francisco, Vol. 1, pp Lee, V.W. (1992). On Strong Motion Uniform Risk Functionals Computed From General Probability Distributions of Earthquake Recurrences, Soil Dyn. and Earthq. Eng., Vol. 11, pp Lee, V.W. and Trifunac, M.D. (1987). Microzonation of a Metropolitan Area, Report No , Dept. of Civil Eng., Univ. of Southern California, Los Angeles, USA 21. Maeda, K. (1996). The Use of Foreshocks in Probabilistic Prediction Along the Japan and Kuril Trenches, Bull. Seism. Soc. Am., Vol. 86, No. 1A, pp McGuire, R.K. (1977). Seismic Design Spectra and Mapping Procedures Using Hazard Analysis Based Directly on Oscillator Response, Earthq. Eng. and Strut. Dyn., Vol. 5, pp Novikova, E.I. and Trifunac, M.D. (1993). Duration of Strong Ground Motion: Physical Basis and Empirical Equations, Report No , Dept. of Civil Eng., Univ. of Southern California, Los Angeles, USA.

15 Seismic Microzonation via PSHA Rikitake, T. (1976). Recurrence of Great Earthquakes at Subduction Zones, Tectonophysics, Vol. 35, pp Silverman, B.W. (1986). Density Estimation for Statistics and Data Analysis, Chapman and Hall, London. 26. Todorovska, M.I., Gupta, I.D., Gupta, V.K., Lee, V.W. and Trifunac, M.D. (1995). Selected Topics in Probabilistic Seismic Hazard Analysis, Report CE 95-08, Dept. of civil Eng., Univ. of Southern California, Los Angeles, USA. 27. Todorovska, M.I. and Trifunac, M.D. (1999). Liquefaction Opportunity Mapping Via Seismic Wave Energy, Jour. of Geotech and Geoenvironmental Eng., ASCE, Vol. 125, No. 12, pp Trifunac, M.D. (1976). Preliminary Analysis of the Peaks of Strong Earthquake Ground Motion Dependence of Peaks on Earthquake Magnitude, Epicentral Distance and Recording Site Conditions, Bull. Seism. Soc. Am., Vol. 66, pp Trifunac, M.D. (1990). A Microzonation Method Based on Uniform Risk Spectra, Soil Dyn. and Earthq. Eng., Vol. 9, No. 1, pp Trifunac, M.D. (1995). Empirical Criteria for Liquefaction in Sands Via Standard Penetration Tests and Seismic Wave Energy, Soil Dyn. and Earthq. Eng., Vol. 14, No. 6, pp Trifunac, M.D. and Lee, V.W. (1985). Preliminary Empirical Model for Scaling Pseudo Relative Velocity Spectra of Strong Earthquake Acceleration in Terms of Magnitude, Distance, Site Intensity and Recording Site Conditions, Report No. CE 85-04, Univ. of Southern California, Los Angeles, USA. 32. Wallace, R.E. (1970). Earthquake Recurrence Intervals on the San Andreas Fault, Bull. Geol. Soc. Am., Vol. 81, Weichert, D.H. (1980). Estimation of Earthquake Recurrence Parameters for Unequal Observation Periods for Different Magnitudes, Bull. Seism. Soc. Am., Vol. 70, No. 4, pp Young s, R.R and Coppersmith, K.J. (1985). Implications of Fault Slip Rates and Earthquake Recurrence Models to Probabilistic Seismic Hazard Estimates, Bull. Seism. Soc. Am., Vol. 75, No. 4, pp