ML16341E992

Preliminary Rept on Empirical Studies of Horizontal Strong Ground Motion for Diablo Canyon Site,Ca

ML16341E992
Person / Time
Site:	Diablo Canyon
Issue date:	10/31/1988
From:	Campbell K INTERIOR, DEPT. OF, GEOLOGICAL SURVEY
To:
Shared Package
ML16341E990	List: ML16341E989 ML16341E992 ML17083C154
References
NUDOCS 8902170148
Download: ML16341E992 (74)

Text

Report to the U.S; Nuclear Regulatory Commission PRELIHINARY REPORT ON EMPIRICAL STUDIES OF HORIZONTAL STRONG GROUND MOTION FOR THE DIABLO CANYON SITE, CALIFORNIA Prepared by Kenneth W. Campbell U.S. Geologicsl Survey

October, 1988.

+902'170>

< 050pp275 48 890207 PDR ADGC P

r>>

PRELIVINARY REPORT ON EMPIRICAL STUDIES OF HORIZONTAL STRONG GROUND MOTION FOR THE DIABLO CANYON SITE, CALIFORNIA Kenneth W. Campbell U.S.

Geological Survey

Golden, Colorado INTRODUCTION I

This report presents preliminary results for an empirical analysis of horizontal strong ground motion commissioned by the U.S. Nuclear Regulatory Commission (NRC) as part of their evaluation of the seismic safety of the Pacific Gas and Electric Company (PGEE) nuclear power plant at Diablo Canyon, California.

The study was initiated in 1986 to parallel similar studies conducted by PG8E as part of their Long Term Seismic Program.

This report is intended only as an interim progress report; a more comprehensive report will be submitted at the conclusion of the study, sometime in mid-1989. It should be noted that the preliminary results described in this report are subject to review, as part of the standard review procedures of the NRC.

Modifications, if any, required as part of this review process will be included in the final report.

The report provides a brief discussion of the data base and regression analyses used to develop near-source attenuation relationships for peak horizontal acceleration (PHA), peak horizontal velocity (PHV), and 5-percent

damped, horizontal pseudorelative velocity response (PSRVH) spectra.

This is followed by a presentation of site-specific estimates of ground motions at the

Diablo Canyon site for PGEE's prop:

'<<.3 Hosgri design earthquake as well as several other hypothetical earthquakes on the Hosgri fault.

The site-specific results indicate that the median and median-plus-one-I standard-deviation response spectra at Diablo Canyon for PGLE's proposed Hosgri design earthquake, weighted by fault type according to the weighting scheme proposed by PGEE, exceed the highest of three response spectra developed by PGSE at frequencies less than about 3.5 Hz and 8 Hz, respectively.

STRONG-MOTION DATA BASE The strong-motion data base has been updated from data bases developed by Campbell

(-1981, 1987).

The present data base has been enhanced with strong-motion recordings from nine additional earthquakes:

the 1972 Stone Canyon (M1=4.7), the 1976 Mesa de Andrade (M<=5.3), the 1976 Caldiran (Ms=7.3)> the 1984 Morgan Hill (Ms=6. 1), the 1985 Central Chile (Ms=7.8), the 1985 Michoacan (Ms=8. 1), the 1985 Michoacan aftershock (Ms=7.6), the 1986 North Palm Springs (Ms=6.0), the 1986 Chalfant Valley (Ms=6.2),

and the 1987 Whittier Narrows (M~=5.9).

Because Diablo Canyon is a rock site, priority was given to earthquakes with strong-'motion recordings on rock.

Strong-motion parameters selected for analysis were the horizontal components of peak acceleration (PHA), peak velocity (PHV), and 5-percent

~

~

~

~

~

~

~

~

~

damped pseudorelative velocity response (PSRVH) spectra at 15 periods 'ranging from 0.04 to 4.0 sec.

Strong-motion parameters were defined as the arithmetic average of the two horizontal components; magnitude was defined as either H~

for magnitudes less than 6.0 or Ms for magnitudes greater than or equal to 6.0; and source-to-site distance was defined as closest distance to seismogenic rupture (Campbell, 1987).

Strong-motion recordings were selected according to criteria adopted by Campbell (1987), with the following exceptions.

First, the magnitude 5.0 cutoff was relaxed to include the 1972 Stone Canyon (M~=4.7) earth'quake.

This earthquake was added because (1) a special study had been used to determine the rupture zone of this earthquake, and (2) processed recordings were vailable.

Second, rock sites were included in the present analyses in order to assess the differences in ground motion between soil and rock.

Only 3

recordings on soft rock similar to that found at the Diablo Canyon site were

used, since a preliminary analysis of ground motions indicated that there were large differences in both frequency content and amplitude between recordings on soft rock (primarily sedimentary rock) and hard rock (primarily crystalline J

rock).

Recordings vere limited to near-source distances in order to focus on ground motions of greatest relevance to Diablo Canyon.

Application of the selection criteria to the soil recordings compiled for this study resulted in the selection of 200 recordings from 25 earthquakes for the analysis of PHA, 152 recordings from 21 earthquakes for the analysis of l

PHV, and as fev, as 86 recordings from 15 earthquakes and as many as 144 recordings from 21 earthquakes for the analysis of PSRVH, depending on the period of the spectral component.

Application of the selection criteria to soft-rock recordings resulted in the selection of a substantially smaller number:

42 recordings from 11 earthquakes for PHA, 20 recordings from 5 earthquakes for PHV, and as few as 12 recordings from 2 earthquakes and as many as 20 recordings from 5 earthquakes for PSRVH, depending on the period of the spectral component.

The distribution of PHA with respect to magnitude and distance for both soil and soft-rock sites is displayed in Figures 1 and 2.

REGRESSION ANALYSES The functional form used to model the scaling relationship between strong-ground motion, magnitude, and distance is a variation of that proposed by Campbell (1987):

ln Y = a +

bM + d ln[R + cpxp(cg)]

+ eF + f~tanh(f~(M+f~)] +

+ g~tanh(g~D)

+ Z h K. + <

where Y is the specified strong-motion parameter; M is earthquake magnitude; R

is closest distance to seismogenic rupture in'ilometers; F is fault type (F =

0 for strike-slip faults and F =

1 for oblique/reverse/thrust faults);

D is depth'to basement rock in kilometers; K. represents soil-structure interaction (SSI) effects (K~ =

1 for embedded 3-11 story buildings, KZ =

1 for embedded

>12 story buildings, K~ =

1 for ground-level

>3 story buildings, and K~ = KZ =

K~ = 0 for all other sites);

e is a random error. term with mean of zero and standard deviation equal to o, the standard error of estimate; tanh(*) is the hyperbolic tangent function; and a, b,...,

h~ are coefficients to be determined by the data.

Equation (1) was modified from that proposed by Campbell (1987) to include the additional magnitude term, f~tanh[f~(M+f~)], required to model the magnitude dependence of PSRVH.

In addition, the SSI parameters were slightly redefined and revised to accommodate the expanded data base.

In particular, the prior dependence of K~ on distance was removed because the significance of this dependence was found to be equivocal in the present analysis.

The coefficients a,

b,...,

h~ were determined from a weighted nonlinear

~

~

~

~

~

least-squares regression analysis similar to that used by Campbell (1981,

~ 1987).

The weighting scheme was slightly modified to take into account multiple recordings at the same site.

In the revised

scheme, recordings from each earthquake that fall within a specified distance interval continued to be equally weighted with those from other earthquakes within the, same interval;

however, recordings obtained at the same site are now assigned the same weight as a single recording.

Th'is latter constraint is employed to reduce the bias associated with those recordings that share the same source,

path, and site effects (although they will not necessarily have the same SSI effects).

Coefficients were included in the relationship only if they were found to be'tatistically significant at a 90-percent level of confidence.

Because of the extremely limited number of soft-rock recordings, the regression analysis was restricted to soil. sites.

Once the regression coefficients for soil sites were established, an analysis of residuals was used to determine whether the soft-rock data differed substantially from predictions based on the soil data.

In all cases,'he differences were not found to be statistically significant, indicating that within the statistical variability of the soil and soft-rock data the predictions based on soil sites are indistinguishable from ground motions recorded on soft-rock.

Of course, this conclusion could change in the future as more soft-rock recordings are added to the data base.

This is particularly'true for PSRVH.

The results of the regression analyses are summarized in Table 1.

Although for some parameters the SSI coefficients, h~, were found to be statistically significant and were,

thus, included in the analysis, they are not presented in Table 1.

They were included in the analysis in order to reduce the potential bias associated with SSI effects

and, thus, provide a

more robust estimate of free-field ground motions, However, since it was not

the intent of this study to model SSI effects, the simple approach of representing SSI by a few simple scaling parameters is not considered

~

~

appropriate for predicting these complex effects.

Peak Horisontal hcceleration.

Results for PHA are similar to those derived by Campbell (1987).

As before, the data support the independence between PHA and magnitude at R = 0 (i.e., at the source of the earthquake).

This phenomenon is often referred to as ground-motion "saturation".,

.Furthermore, fault type was found to be statistically significant, with peak accelerations from oblique/re'verse/thrust faults being, on average, 47-percent higher than those from strike-slip faults (there were no recordings from normal faults).

Embedded buildings were again found to have lower-than-average peak accelerations; coefficients for both K~ (representing 3-11 story buildings) and K~ (representing

>12 story buildings) were found to be statistically significant.

The calculated standard error of cr = 0.421 (a

multiplicative factor of 1.52) is larger than that found previously; this is primarily a result of the increased dispersion of the added recordings.

The free-field attenuation relationship for PHA, plotted at four magnitudes ranging from 5.0 to 8.0, is displayed in Figure 3.

Peak Horizontal Velocity.

Results for PHV are somewhat different from those derived by Campbell (1987).

As before, the data support the independence between PHV and magnitude at R = 0 (ground-motion saturation),

as well as the dependence of PHV on depth to basement rock, although the latter dependence is ~eaker than found previously.

Fault type was again found to be signif'icant, with peak velocities from oblique/reverse/thrust faults being, on

average, 39-percent higher than those from strike-slip faults.

SSI coefficients were not found to be statistically significant in the present

analysis.

This departure from our previous results probably reflects the change in the definition of these parameters.

Consistent with the results for peak acceleration, the calculated standard error of o = 0.395, representing a

multiplicative factor of 1.48, is larger than that found previously.

The free-field attenuation relationship for PHV, plotted at four magnitudes ranging from 5.0 to 8.0, is displayed in Figure 4.

Zorisontal pseudorelative Velocity Response.

The analysis of PSRVH was much more involved than that for PHA or PHV.

Because of the spectral nature of the parameter, the analysis was carried out in several steps.

In the initial analysis, we performed an independent regression analysis on each spectral component using Equation (1), with fi = 0, as the regression model.

This analysis,

however, proved to be unsatisfactory, since extreme period-to-period variability in the regression coefficients resulted in predicted spectra that exhibited unreliable shapes for some combinations of parameters.

The reasons for this behavior are believed to be the result of (1) the large number of parameters included in the relationship, (2) the relatively small number of recordings available for PSRVH, and (3) the period-to-period variability in the number of. recordings and earthquakes used in the analysis of the various components.

Joyner and Boore (1982) and Joyner and Fumal (1985) simply smoothed their regression coefficients to obtain well-behaved spectra.

However, because of the correlation between some of the regression coefficients and because of the relatively large number of parameters included in Equation (1), this type of smoothing was=not feasible in the present study.

Instead, it was decided to defer to an alternate approach of estimating PSRVH from peak acceleration.

This approach, common among the engineering community (e.g.,

Newmark and Hall,

1982; '.ioyner and Boore, 1988),

was found to give more stable results than independent regressions on the spectral components.

The technique of estimating response spectra from peak acceleration has been seriously criticized in the literature (e.g.,

Joyner and Boore, 1988; Bender and Campbell, 1988);

however, the major criticism lies in the use of peak acceleration to scale a fixed spectral shape.

The technique used in the present study allays this criticism by allowing the spectral shape to vary with magnitude,

distance, and depth to basement rock.

Because of the relatively large number of independent variables in Equation (1), the regression analysis was carried out in several steps.

In the first step, the natural logarithm of the ratio PSRVH to PHA [in dependence of thi found for ln PHA.

This analysis indicated that the single most important (PSRVH/PHA)] was plotted against magnitude and distance to see if the

's ratio on each of these parameters was different from that difference between ln (PSRVH/PHA) and ln PHA was the dependence of the former on magnitude for periods greater than 0.3 sec.

For periods less than or equal to 0.3 sec, the analysis indicated that a simple constant scaling factor could' be used to predict ln PSRVH from ln PHA; in other words, the ratio is independent of both magnitude and distance.

The observed trend in ln (PSRVH/PHA) at periods greater than 0.3 sec suggested that a hyperbolic tangent functionthe function involving the f~

coefficients in Equation (1)--could be used to adequately model the magnitude dependence of this parameter.

The hyperbolic tangent function rather than a

simple polynomial was preferred because of its well-behaved characteristics; it has a value of zero when its argument is zero, it is nearly linear for small values of its argument, and it asymptotically approaches one at large

values of its argument.

In contrast, a polynomial has potentially unpredictable characteristics when extrapolated or evaluated at or near the limits of the data.

We also found a weak dependence of ln (PSRVH/PHA) on distance, especially at the longer periods.

However, since this dependence was negligible compared to that observed for magnitude, its consideration was delayed until a later stage of the analysis.

In step

two, a weighted nonlinear regression analysis was used to determine the constant a and the magnitude coefficients f'~ in Equation (1) ~

As a means of controlling excessive period-to-period variability i.n these coefficients, a joint regression of ln (PSRVH/PHA) on magnitude using all the data corresponding to those periods requiring the additional magnitude term (periods= of 0.4 sec or. larger) was used to determine f~ and f~.

As a result, f~ and f~ were constrained to be independent of period for periods of 0.4 sec and larger.

It was further necessary to constrain the absolute value of f3 to 4.7, the value of the smallest magnitude in the data set, because the unconstrained value resulted in a negative correlation between ln (PSRVH/PHA) and magnitude for small-magnitude earthquakes.

Careful analysis indicated that this physically undesirable trait was a result of low signal-to-noise ratios for the longer-period components of the small-magnitude spectra.

My experience indicates that this problem is pervasive in the standard processed strong-motion data set, even at larger magnitudes.

In order to obtain autonomy between the various spectral components, the coefficients a and fi were determined independently for each period in the above analysis.

In this way, both the absolute amplitude and the degree of magnitude scaling of individual spectral components are allowed to be 10

independent of period, resulting in a spectral shape that is free from unduly

~

~

~

restrictive constraints.

In step three, residuals of ln (PSRVH/PHA) calculated with respect to the attenuation relationship developed in step tw'o were plotted as a function of magnitude,

distance, and depth to basement rock.

Inspection of these plots indicated that the residuals were not.correlated with either magnitude or

distance, confirming the validity of the regression derived in step two.

However, there was an observed trend with depth to basement rock for periods greater than 1.0 sec.

This trend suggested that a hyperbolic tangent function--the function involving the g coefficients in Equation (1)could be adequately used to model the dependence of ln (PSRVH/PHA) on sediment depth.

In step four, a weighted nonlinear regression analysis was used to determine the sediment-depth coefficients 8~ and a new value for the constant a.

To simplify this analysis, the magnitude term derived in step two was subtracted from ln (PSRVH/PHA), thus removing the magnitude dependence of this parameter, before carrying out the analysis.

As a means of controlling excessive period-to-period variability in these coefficients, a joint regression of ln (PSRVH/PHA) on depth to basement rock using all the data corresponding to those periods requiring the additional seal.ing term (periods of 1.5 sec or larger) was used to determine g~.

As a result, g~ was constrained to be independent of period for periods of 1.5 sec and larger.

In order to obtain autonomy between the various spectral components, the coefficient g~ and a new value for the constant a were determined independently for each period in the above analysis.

In this way, both the absolute amplitude and the degree of sediment-depth scaling of individual 11

spectral components are allowed to be independen<

of, period, resulting in a spectral shape that is free from unduly restrictive constraints.

In step five,- residuals of ln (PSRVH/PHA) calculated with respect to the attenuation relationship developed in step four were plotted as a function of magnitude,

distance, and depth to basement rock; and significance tests were applied to determine whether ln (PSRVH/PHA) was dependent on either fault type or SSI.

These analyses indicated that the residuals were not correlated with either magnitude,

distance, depth to basement rock, or fault type, confirming the validity of the regression results derived in step four.

However, the significance tests did indicate that SSI effects were statistically significant at the longer periods.

In all cases, the residuals, and therefore ln (PSRVH/PHA), were found to be significantly higher for recordings from buildings

>3 stories high than for other recordings, with residuals for K~

(representing embedded 3-11 story buildings) and K~ (representing nonembedded

>3 story buildings) being significant at periods greater than 2.0 sec, and residuals for K~ (representing embedded

>12 story buildings) being significant at periods greater than 0.75 sec.

No other parameters were found to be statistically significant.

In step six, a weighted nonlinear regression analysis was used to establish a new value for the constant a and the SSI coefficients h~ for those periods where Kz was found to be significant in step five.

To simplify the

analysis, both the magnitude term developed in step two and the sediment-depth term developed in step four were subtracted from ln (PSRVH/PHA) before carrying out the analysis.

Autonomy between the various spectral components was"maintained by doing the analysis independently for each spectral component.

12

In the final step, attenuation relationships for ln PSRVH were derived by combining the relationships developed for ln PHA and ln (PSRVH/PHA) through

~

~

the mathematical relationship ln PSRVH = ln PHA + ln (PSRVH/PHA)

(2)

Coefficients for the resulting relationships are given in Table 1.

Using these coefficients together with Equation (1),

we calculated residuals for ln PSRVH and'lotted them against magnitude,

distance, and depth to basement rock.

Inspection of these plots indicated that the residuals were not correlated with either of these parameters, confirming the validity of the I

multistep regression procedure.

Furthermore, significance tests on subsets of the residuals, corrected for modeled SSI effects, revealed no statistically significant dependence of the residuals on either fault type or SSI.

The residuals were also used to estimate standard errors for ln PSRVH.

The standard errors reported in Table 1, however, have been adjusted to provide smoothed estimates of PSRVH spectra at the upper fractiles of their cumulative distributions.

Two criteria were used to make these adjustments.

First, no value of standard error was allowed to be less than that determined for ln PHA.

This is a mathematical constraint required by Equation (2).

Second, period-to-period variability in the standard errors was controlled by constraining them to be either a constant or a monotonically increasing function of period, consistent with their overall trend with period.

These criteria required several of the standard errors to be increased by as much as

0. 1; however, in no case were they reduced by more than 0.01.

13

GROUND-MOTION ESTIMATES FOR DIABLO CANYON The attenuation relationships developed in the previous section (Eq. (1);

Table

1) were used to develop site-specific estimates of free-field ground motion at the Diablo Canyon site for several earthquake scenarios.

One of these scenarios is PG&E's proposed design earthquake:

a moment magnitude (H) 7.2 earthquake on the Hosgri fault, whose closest point of rupture is located approximately 4.5 km from the site (PG&E, 1988).

Seismic velocity profiles at or near the site (PG&E, 1988, Figs.

2-9, 4-13, and 5-5) identify a strong velocity gradient to a depth of approximately 4 km.

Although rocks of the Franciscan Complex, usually considered to be basement rock, underlay the site at a depth of 1-2 km,'the velocity gradient in the upper 4 km is comparable to that typical of sedimentary rock (R. Wheeler and K. Campbell, unpublished

~

~

~

~

~

data).

Therefore, depth to basement rock was taken as 4

km for purposes of predicting ground motions at Diablo Canyon.

This assumption only affects estimates of PHV and estimates of PSRVH at periods larger than.1.0 sec.

Estimates of PHA and PHV for PG&E's proposed Hosgri design earthquake and four other Hosgri earthquakes of Hs = 6.6-7.8 are presented in Tables 2 and 3

(Hs is considered equivalent to H~ for this and later computations; Hanks and

Kanamori, 1979)

For convenience, the estimates have been segregated by fault type and uncertainty level.

The median and median-plus-one-standard-deviation (median+1e) estimates of PHA for the proposed Hosgri design earthquake are 0.50g and 0.77g for strike-slip faults and 0.74g and 1.13g for oblique/reverse/thrust faults, respectively.

Similarly, median and median+1'stimates for PHV are 65 and 96 cm/sec for strike-slip faults and 90 and 134 cm/sec for oblique/reverse/thrust faults.

If the estimates for the two fault types are combined according to the weighting scheme proposed by PG&E ( 1988),

14

the weighted median and median+1cr estimates are 0.59g and 0.89g for PHA and 74 1O9 cm/sec for PHV.

Other weighting schemes would result in either smaller or larger estimates depending on the weighting scheme used.

If more weight were given to the oblique/reverse/thrust

scenario, the weighted estimates would be higher than those given.

Estimated PSRVH spectra for PG&E's proposed Hosgri design earthquake for the strike-slip and oblique/reverse/thrust fault scenarios are presented in Figures 5 and 6, respectively.

Additional plots of the median spectra for five magnitudes ranging from Ms= 6.6-7.8 are given in Figures 7 and 8.

Similar plots for the horizontal components of 5-percent damped pseudoabsolute acceleration (PSAAH) spectra are given in Figures 9-12.

For convenience, all PSAAH spectra are plotted as a function of frequency at the same scale as similar plots provided by PG&E (1988).

Our estimated PSAAH spectra for PG&E's proposed Hosgri design earthquake, weighted by fault type according to the weighting scheme proposed by PG&E, is displayed in Figure 13.

PG&E's weighted spectra (their figs. 4-25 and 4-26) are presented in Figures 14 and 15.

Upon comparing Figures 13-15, we find that the median estimate of the higher of the PG&E spectra--that based on attenuation relationships from regression analyses is lower than the weighted median spectrum developed in the present study for frequencies less than about 3.5 Hz, but larger than this spectrum for frequencies larger than this value.

A similar comparison of the median+1cr spectra shows similar results, except that the frequency at which this transition occurs is about 8 Hz.

Note that because the dependence of PSAAH on fault type is substantially smaller for the PG&E spectra than for the spectra developed in this study, a weighting scheme that would give more weight to the oblique/reverse/thrust scenario would 15

result in transition frequencies somewhat higher than those found in the above comparison.

16

REFERENCES

Campbell, K.W. (1981).

Near-source attenuation of peak horizontal

~

~

~

acceleration'Seismological Society of America Bulletin, v. 71, p.

2039-2070.

Campbell, K.W. (1987). Predicting strong ground motion in Utah, in Gori, P.L.,

and Hays, W.W., editors, Assessment of Regional Earthquake Hazards and Risk Along the Vasatch Front, Utah: U.S. Geological Survey Open-File Report 87-585, v. II, p. L1-L90.

Hanks, T.C.,

and H. Kanamori (1979).

A moment magnitude scale:

Journal. of Geophysical

Research,

v. 84, p.

2348-2350.

Joyner, W.B., and D.M. Boore (1982). Prediction of earthquake response spectra:

U.S.

Geological Survey Ope'-File Report 82-977.

k

Joyner, W.B., and D.M. Boore (1988).

Measurement, characterization, and prediction of strong ground motion, in Von Thun, J.L., editor, Proceedings, Conference on Earthquake Engineering and Soil Dynamics II Recent Advances in Ground-Motion Evaluation:

A.S.C.E.

Geotechnical Special Publication No. 20, p.43-102.

Joyner, W.B., and T.E.

Fumal (1985). Predictive mapping of earthquake ground motion, in Ziony, J.I., editor, Evaluating Earthquake Hazards in the los Angeles Region An Earth Science Perspective:

U.S. Geological Survey Professional Paper

1360,

p. 203-220.

Newmark, N.M., and W.J. Hall (1982).

Earthquake spectra and design:

Berkeley, California, Earthquake Engineering-Research Institute.

17

PGaE i ',y88). Final report of the Diablo Canyon long term seismic program for the Diablo Canyon porkier plant: Pacific Gas and Electric Company report to the U.S. Nuclear Regulatory Commission:

Docket No.

50-275 and 50-323, W

July 31,

1988, San Francisco, California.

18

TABLE 1

Results of Regression hnalyses on PHh, PHV, and PSRVH Paraaeter, Period V

(sec) b cl c2 d

c fl 2

3 tl

~2 ao. of no. of r

records events PHA, g

-2.470 1.08 0.311 0.597

-1.81 0.382 PHV, ca/sec

-1.974 '.34 0.00935 1.01

-1.32 0.327 PSRVH, ca/sec 0.04

-0.648 1.08, 0.311 0.597

-1.81 0.382 0.05 4.379 1.08 0.311 0.597

-1.&l 0.382 200 25 0.421 152'1 0.395 86 15 0.42 142 20 0.44 0.075 0.251 1.08 0.311 0.597

-1.81 0.382 0.10 0.754 1.08 0.311 0.597

-1.81 0.382 0.15 1.424 1.08 0.311 0.597

-1.81 0.382 0.20 1.788 1.08 0.311 0.597

-1.81 0.382 0.30 2.170 1.08 0.311 0.597

-1.81 0.382 0.40 2.009 1.08 0.311 0.597

-1.81 0.382 0.425 0.570

-4.7 0.50 1.930 1.08 0.311 0.597

-1.81 0.382 0.685 0.570

-4.7 0.75 1.612 1.0&

0.311 0.597

-1.&l 0.382 1.266 0.570

-4.7 1.0 1.26&

1.08 0.311 0.597

-1.81 0.382 1.743 0.570

-4.7 144 21 0.46 144 21 0.48 144 21 0.50 144 21 0.50 144 21 0.50 144 21 0.50 lii 21 0.50 144 21 0.50 144 21 0.50 1.5 0.487 1.08 0.311 0.597

-1.81 0.3&2 2.425 0.570

-4.7 0.344 0.553 144 21 0.50 2.0 0.040 1.08 0.311 0.597

-1.81 0.382 2.827 0.570

-4.7 0.469 0.553 144 21 0.50 3.0

-0.576 1.08 0.311 0.597

-1.81 0.382 3.166 0.570

-4.7 0.623 0.553 4.0

-0.766 1.08 0.311 0.597

-1.81 0.382 3.079 0.570

-4.7 0.&57 0:553 144 21 0.55 127 20 0.59

TABLE 2 Predicted Values of Peak Horizontal Acceleration for Diablo Canyon, California (R = 4.5 kji)

Peak Horizontal Acceleration (g)

Magnitude, M

(M).

Strike-Slip Median Median+10 Oblique/Reverse/Thrust Median Median+1a 6.6 6.9 7.2 7.5 7.8 0.476 0.504 0.530 0.553 0.725 0.768 0.807 "0.842 0.446 0.'679

0. 653 0.698 0.739 0.776

0. 810

0. 994 1.06 1

13.

1. 18 1.23 21

TABLE 3 Predicted Values of Peak Horizontal Velocity for Diablo Canyon, California (8 = 4.5 ka, D = 4 tuI)

Peak Horizontal Velocity (cm/sec)

Magnitude, e, (z,)

Strike-Slip Median Median+1'blique/Reverse/Thrust Median Median+1n 6.6 6.9 7.2 7.5 7.8 50.6

58. 1 64.9 71.0

76. 1 75.1

86. 2 96.4 105.4 113.0 70.2 80.6

90. 1 98.5 105.6 104. 2 119. 6 133.7 146. 2 156.7

55 C

5o LLI 45 I-0 40 O

35 Z

LLI El~

3O V)

UJ 25 V)~

2O I

QJ>

>5 I-(A 10 o

x x

x x

x x

x x

<x N.

II x

x x

x x

x x

y x

x x

x x

x x

x X

xx X

0 4.5 5.0 5.5 6.0 6.5 NAGNITUOE 7.0 7.5 8.0 Figure 1.

Distribution of peak horizontal acceleration with respect to magnitude and distance for soil sites.

0

55 c

5o UJ 45 4o CL O

3 Z

ILJ~

3O 6)

V)

UJ 25 (A

I-~

2O UJO I5 Z

I-(A 10

'o X

x x

x X

x X

x x

x x

x x

x 0

4,5 5.0 5.5 6.0 6.5 NAGNITUOE 7.0 7.5 8.0 Figure 2.

Distribution of peak horizontal acceleration with respect to magnitude and distance for soft-rock sites.

0 0

6 S

O' OISTANCE T8 SEISMHGENIC RUPTURE (km) 6 8

P Pigure 3.

Attenuation relationship for peak horizontal acceleration (PHA) for strike-slip faults:

( bottom to top) H = 5.0, 6.0, 7.0 and 8.0.

O CV EV 0 I-o<

Gl IJJ)

< cv I-zQO N 0 CL g) tealx e UJ CL 0

0 4

6 8

O 2

DISTANCE TH SEISNHGENIC RUPTURE (km) 6 8

Figure 4.

Attenuation relationship for peak horizontal velocity (PHV) for strike-slip faults: (bottom to top) H = 5.0, 6.0, 7.0 and 8.0.

The assumed depth to basement rock is 4 km.

(Q O

0)

Ch 2

CV I

~ tV GO W m Ld I

~

cu I

K1O U)

CL (p CI LIJ CL

~n

~O lD C3

/

I I

I I

/

/

/

/

/

/

I

/

I

/

//

r/

//

p 2

p.l p

0 UNDANPEO NATURAL PERI80 (sec) p 1

Figure 5.

Predicted 5-percent damped horizontal pseudo-relative velocity response (PSRVH) spectra at Diablo. Canyon for the strike-slip fault scenario of the proposed Hosgri design earthquake (Ms=7.2, R=4.5 km): (solid line) median spectrum',

(dashed lines) median+1a spectra.

The assumed depth to basement rock is 4 km.

(Q O

Q) 0)

C C3

)-

cu I

+ co GO Ld Co

)

Ld)

I Q

K)

~~o

~ m (A

Q (p

CI LIJ Q

O

~O C) r

/

/

/

/

/

/

I

/

I

/

/

/

/

/

/

/

/

/

/

///

p 2

p 1

p 0

UNDANPED NATURAL PERISD (sec) p 1

Figure 6.

Predicted 5-percent damped horizontal pseudo-relative velocity response (PSRt/H) spectra at Diablo Canyon for the oblique/reverse/thrust fault scenario of the proposed Hosgri design earthquake (Hs=7.2, R=4.5 km): (solid line) median spectrum; (dashed lines) median+1cr spectra.

The assumed depth to basement rock is 4 km.

O CD CO E

0-cu I

~ fV-SO LLI -co

)

LLI)

I LY K)

O CA CL Cl W

CL

~O lC) 0 2

0-1 0

0 UNDAMPED NATURAL PERISD (sec) 0 1

'Figure 7.

Predicted median 5-percent damped horizontal pseudo-relative velocity response (PSRVH) spectra at Diablo Canyon for strike-slip faults:

( bottom to top)

Hs

= 6.6, 6.9, 7.2, 7.5 and 7.8; R = 4.5 km; D

=

4 km.

I

0 2

0 1

p 0

UNDANPED NATURAL PERIBD (sec) 0 Figure 8.

Predicted median 5-percent damped horizontal pseudo-relative velocity response (PSRVH) spectra at Diablo Canyon for oblique/reverse/thrust faults:

( bottom to top)

Ms

6.6, 6.9, 7.2, 7.5 and 7.8; R = 4.5 km; D

4 km.

3.0 2.5 2.0 K)

C)

LLJ U)

'.0

. 0.5 0.0 O

-1

//

O O

FREQUENCY (Hz)

Figure 9.

Predicted 5-percent "damped horizontal pseudo-absolute acceleration response (PSAAH) spectra at Diablo Canyon for the strike-slip fault scenario of the proposed Hosgri design earthquake (Ms=7,2, R=4.5 km):

( solid line) median spectrum; (dashed lines) median+1u spectra.

The assumed depth to basement rock is 4 km.

~

Q LU LLj 3.0 2.5 2.0

-/

/

CO IKlo CL 1.'5 1.0 0.5 0.0 p

-1

//

FREGlUENLY (Hz)

Figure 10.

Predicted 5-percent damped horizontal pseudo-absolute acceleration response (PSAAH) spectra at Diablo Canyon for the oblique/reverse/thrust fault scenario of the proposed Hosgri design earthquake (Hs=7.2, R=4. 5 km): (solid line) median spectrum; (dashed lines) median+le spectra.

The assumed depth to basement rock is 4 km.

l CL LLj LLj LLI Cl Ld V)

Cj 3.0 2.5 2.0 1.5 1.0 0.5 0.0 O

-I Q

0 p

1 FREQUENCY (Hzf Figure 11, Predicted median 5-percent damped horizontal pseudo-absolute acceleration response (PSAAH) spectra at Diablo Canyon for strike-slip faults:

( bottom to top)

Hs

6.6, 6 9, 7.2, 7.5 and 7.8; R = 4.5 km; D

4 km.

Z Cj Ltd LLj 3.0 2.5 2.0 40 Cl LLI U)

CL 1.5 1.0 0.5 0.0 O

-1 O

O FREQUENCY (Hz)

Figure 12.

Predicted median 5-percent damped horizontal pseudo-absolute acceleration response (PSAAH) spectra at Diablo Canyon for oblique/reverse/thrust faults:

( bottom Co top)

Ms

6.6, 6.9, 7.2, 7.5 and 7.8; R = 4.5 km; D

4 km.

Z 3'0 2.5 2.0 1.0

//

/

/

/

/-

0.5 0.0 p

-1 p

0 p

I FREQUENCY (Hz)

Figure 13.

Predicted 5-percent damped horizontal pseudo-absolute acceleration response (PSAAH) spectra at Diablo Canyon for the weighted fault scenario of the proposed Hosgri design earthquake (Ha=7.2, R=4.5 km):

( solid line) median spectrum; (dashed lines) median+1cr spectra.

The assumed depth to basement rock is 4 km.

The strike-slip and oblique/reverse/thrust fault scenarios have been weighted 0.65 and 0.35, respectively.

Median horizontal response spectra 6% Damping 2.5 EXPLANATION Based on statistics of neansooree recordings Based on attenuation relationships from regression analyses e e

~ ~

~

~ Based on numerical modeling studies C0 1.6

.5

///

/.

//'

~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~

0

.1

~ ~

~

~ ~

.5 2

5 Frequency (Hz) 10 20 50 100 Figure 14.

The Diablo Canyon site-specific median horizontal acceleration response spectra developed by PG&E, weighted for style of faulting as follows: 0.65, strike-slip; 0.30, oblique; 0.05, thrust (PGLE, 1988, Fig. 4-25).

. 84th Percentile horizontal response spectra 5% Damping 2.5 EXPLANATION

Based on ssasissiosof near.source recordings Based on attenuation relationships from regression analyses

~

~

~ ~ ~ ~ Based on numerical modeling studies 2

~ne 8

P.-

1.5

.5

/

0 ~

/

e

'e

/

//

/

/

e

//

e

/

~

//

e

/

~ ~ ~ ~

~

~ ~

~ ~ ~ ~ ~

~ ~

~

0

.2

.5 2

5 Frequency (Hz) 10 20 50 Figure 15.

The Diablo Canyon site-specific 84th percentile (median+1o) horizontal acceleration response spectra developed by PG8E, weighted for style of faulting as follows: 0.65, strike-slip; 0.30, oblique; 0.05, thrust (PGKE, 1988, Fig. 4-26).

0