corso di costruzioni in zona simica modulo di ... › data › earthquake_engineering › sei… ·...

FACOLTA’ DI INGEGNERIA DELL’UNIVERSITA’ DEGLI STUDI ROMA III

Laurea specialistica in protezione del territorio

dai rischi naturali

Corso di costruzioni in zona simica

Modulo di determinazione della pericolosità sismica

F.Sabetta

7. Microzonazione

Attenuazione e GMPEs

Anno accademico 2017/18

Corso di Sismologia 7. Microzonazione e Attenuazione

F. Sabetta 7 2

SUMMARY

SEISMIC MICROZONATION ................................................................................................. 3 Coseismic effects .......................................................................................................................... 3 Liquefaction ................................................................................................................................... 4 SEISMIC MICROZONATION: GOAL AND PURPOSE ................................................................ 5 Guidelines and Criteria for Seismic Micro-Zoning ..................................................................... 7 2010-2017 SM studies in Italy ....................................................................................................... 9 GMPES: ATTENUATION RELATIONS ................................................................................10 INDEPENDENT VARIABLES SELECTION AND ADJUSTMENT. .................................................10 1. Magnitude ................................................................................................................................. 11 2. Distance .................................................................................................................................... 11 3. Style of faulting ........................................................................................................................ 12 4. Local Site Conditions .............................................................................................................. 13 VARIABLES SELECTION ...................................................................................................14 MODEL SELECTION .........................................................................................................14 Magnitude dependence ............................................................................................................... 14 Distance dependence .................................................................................................................. 15 Style of faulting and site dependence ....................................................................................... 15 Variance and uncertainty ............................................................................................................ 16 SP96 Sabetta & Pugliese (1996) model ..................................................................................... 16 New Models .................................................................................................................................. 17 Modification of the spectral shape according to earthquake magnitude .............................. 17 Examples ...................................................................................................................................... 18 SELECTION AND USE OF ATTENUATION RELATIONS ...........................................................21 Recent s.m. databases used in GMPEs .................................................................................... 21 COMPARISON OF RECENT ATTENUATION RELATIONS ........................................................22 ALEATORY UNCERTAINTY: STANDARD DEVIATION ............................................................25


F. Sabetta 7 3

Seismic microzonation

Coseismic effects


F. Sabetta 7 4

Liquefaction


F. Sabetta 7 5

Seismic microzonation: Goal and Purpose


F. Sabetta 7 6


F. Sabetta 7 7

Guidelines and Criteria for Seismic Micro-Zoning


F. Sabetta 7 8


F. Sabetta 7 9

2010-2017 SM studies in Italy


F. Sabetta 7 10

GMPEs: Attenuation Relations

Independent variables selection and adjustment.

It is common practice by now to use more than one attenuation relation in hazard studies.

As different relations use different definitions both of the dependent and independ-ent variables, empirical conversions rules or adjustments are need to achieve compatibility amongst the relations.

Consistency among different attenuation relations need to be checked in terms of:

• Distance source-to-station definition

• Magnitude scale

• Style-of-faulting

• Site conditions (definition of rock site)

• Selection of the two horizontal components

ATTENUATION RELATIONS

DATA

SELECTION

PARAMETERS

SELECTION

MODEL

SELECTION

Dependent Variables

(PGA, PGV, Arias Intensi-ty, Spectral Values, Dura-

tion,etc.)

Independent Variables

(Magnitude, Distance, Site

Geology, etc.)

They probably represent the most used tool in Engineering Seismology and Earthquake Engineering for estimating the values of response spectral accelerations for future earth-quake scenarios.

Strong-motion attenuation rela-tions are empirical equations that can be used to estimate the values of strong-motion parameters as functions of independent parame-ters that characterise the earth-quake and the site of interest. The name ‘attenuation relationships’ arises from the fact that the equa-

tions describe the decay of the strong motion with distance from the source. A more suitable term would be ground motion scal-ing relationships or ground mo-tion prediction equations.


F. Sabetta 7 11

1. Magnitude

2. Distance

The conversion among different distance measures is the most critical an difficult to select because it clearly depends on the dimension and orientation of the causative fault.

Several empirical conversions rules are available in literature to con-vert among different magnitude scales.

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0

MsM

w

Ambraseys & Free 1997 h<30 km

Bungum et al. 2003

Ekstrom & Dziewonski 1988

The difference between Mw and Ms becomes of some rel-evance only for M<5.5 and the distinction between different Mw-Ms conversions is absolutely negligible up to magni-tudes lower than 4.

The magnitude scales used in most attenuation relationships are Ms and Mw, although some studies use a combination of ML for smaller earth-quakes and Ms for larger earth-quakes.

The moment magnitude Mw is the most physically meaningful meas-ure of earthquake size but the seismic moment is not available for all earthquakes, particularly smaller and older events.

The selection of a distance source-to-station is one of the most critical steps in developing and comparing attenuation relations because many different definitions can be, and have in fact been em-ployed.

In the far field the choice of dis-tance measure makes little differ-ence because the dimension of the fault is generally small compared to the distance of the station. Howev-er, closer to the earthquake source and for high magnitudes, the choice of distance measure is very im-portant and consistency is needed between the way the regression equation was derived and the

way is being used.

Near field = distance comparable with the source dimensions

Far field = distance much higher than the source dimensions


F. Sabetta 7 12

3. Style of faulting

The influence of the fault mechanism on the nature of the strong motion is a subject of some de-bate, and it is included only in most recent attenuation relationships.

In case of reverse and thrust faulting there is a clear evidence ground motion shows higher amplitudes respect to strike-slip faulting. The difference is period-dependent showing the higher values (1.2-1.3) at short-to-mid periods and. The difference is nearly magnitude and distance independent.

In case of normal faulting ground motions tend to be smaller than those from strike-slip earthquakes by about 10% In this case the difference is

period-independent.

Bommer et al. (2003)

Adj. fact. for reverse

Adj. fact. for normal


F. Sabetta 7 13

4. Local Site Conditions

Last factor to be taken into account is the presence of topographic features such as ridges, basins and valleys. However, no attenuation relationship has included topography explicitly, partly because of the complexity of the effects and the difficulties in defining simple parame-ters to model them.

Due to the very broad categories and large uncertainties in site classification of most of strong motion stations, it is generally more appropriate to use attenuation relations for a rock site condition and apply subsequently a site specific response calculation.

The issue is that also the definition of “rock” changes considerably among different re-lations.

The surface geology is accounted for or by broad site categories ("rock", "stiff-soil","soft-soil“, “alluvium” etc.) or by the velocity of shear waves over the up-permost 30 m at the site, VS30.

Why 30 m? Simply because is the depth to which a borehole can be drilled in one day.

This may be determined from direct measurements in boreholes or inferred from empirical relations between the shear modulus, μ and the SPT blow count, N.

Another feature that can influence the na-ture of strong motion is the presence of large structures, which can modify the ground motion through interaction with the soil and foundations. This is only included implicitly in attenuation relationships through the exclusion of records from the basement or ground floor of buildings with more than three storeys in some attenua-tion studies

East-west components of ground velocity from a magnitude 4.3 aftershock of the 1983 Coalinga earthquake in California. The influence of the alluvial layers on amplitude and duration can be appreciated from the records (Reiter, 1990).

One feature of the site that does have a very pronounced influence on the na-ture of the ground motion is the sur-

face geology.

a) b)a) b)

Shear wave velocity profiles of s.m. stations on rock: a) 9 Italian stations; b) mean values of about 50 U.S. stations. Black=observed data; blue=mean profile; red= standard deviation.

In case of most of the European re-lations the VS30 for rock can be considered around 800 m/s,

whereas in case of Western U.S. is generally lower and around 600

m/s.


F. Sabetta 7 14

Variables selection

Model Selection

Magnitude dependence

The basic functional form for the ground motion regression analysis is:

where Y is the ground motion parameter (peak acceleration or response spectral value), M is themagnitude, R is the distance measure, Pi are other parameters which may be used to characterize the source (e.g. style of faulting) and the local site conditions, and e is an error term representing the uncertainty in Y.

Because the distribution of Y is closely approximated by a lognormal distribution, the regression is usually performed on the logarithm of Y:

Forms of the functions f1(M), f2(R), and f3(Pi) can be determined according to statistical procedures generally referred to as Exploratory Data Analysis. This has been done for ex-ample by Breiman and Friedman (1985) and Sabetta & Pugliese (1987) who adopted the procedure called "Alternating Conditional Expectation" providing estimates of the optimal transformations of the independent variables which may be used graphically to select specif-ic functional forms.

Another way to select forms for predictive relations is based on theoretical considerations:

1) Earthquake magnitude is typically defined as the logarithm of some peak motion and con-sequently lnY should have a linear dependence from M and f1(M) becomes:

f1(M)=C2M

Some authors (e.g. Boore et al. 1997) adopted also a quadratic term in the function of M

f1(M)=C2M + C3M2

1. DESCRIPTIVE PARAMETER

• PGA

• PGV

• Response spectral values

• Arias/Housner intensity

• Duration

2. SELECTION OF THE HORIZONTAL COMPONENTS

• max value

• both components

• arithmetic mean

• geometric mean

3. STRONG MOTION DATA PROCESSING

"uncorrected" data (only PGA)

"corrected" data

• analog or digital recordings (instrument and digitization noise)

• processing procedure (time or frequency domain; type of filter)

• selection of the band-pass filtering fre-quencies (large effect on PGV and PGD)

• selection of the spectral ordinates included in the band-pass filtering range

)()()( 3211 iPfRfMfCY

YiPfRfMfCY ln3211 )()()(ln


F. Sabetta 7 15

Distance dependence

2) There are two terms in distance because there are two mechanisms through which energy is dissipated with increasing distance from the source: the first term represents the geomet-rical spreading of the seismic wave front and the second term the anelastic attenuation due to absorption. Consequently f2(R) becomes:

Style of faulting and site dependence

3) To include the effects of local site conditions, most attenuation relations use broad site categories such as "rock", "stiff-soil" and "soft-soil". Recently, there has been a move toward using quantitative site classifications based on the average shear-wave velocity through the top 30 m, VS30.

Consequently f3(Pi) could be for example:

f3(Pi)=C5F+C6S1+C7S2

Where F is a flag for the style-of-faulting (reverse, strike-slip,normal) and S1 and S2 are two dummy variable which can take the value of 1 or 0 according to the site category (rock or soil).

The parameter h is a fictitious depth determined by the regression and in-corporates all of the factors that tend to limit the motion near the source, a property normally referred to as satu-ration with distance (Joyner and Boore, 1981; Campbell, 1985).

This effect is only relevant at short distances (< 10

km) and high magnitudes (> 6.5)

An alternative form allows the shape of the attenuation relation to depend on magnitude with the curves pinch-ing together at short distances.

This type of saturation with magni-tude (Campbell, 1997; Sadigh, 1997) of the ground motion, implying that at short distances moderate magnitude earthquakes produce sim-ilar levels of shaking to those of large magnitude, can be parameterized replacing h with a magnitude de-

pendent h

after Campbell 1985

f2(R) = C4log(R2 + h 2)1/2 + C5 R (Y=C1eC2MRC4eC5R)


F. Sabetta 7 16

Variance and uncertainty

SP96 Sabetta & Pugliese (1996) model

Grouping the formulae shown in the previous slides, the most general functional form for the ground motion regression analysis is:

27165411 SCSCFCRChCY 2h22

4

2

32 )e(RlogCMC MCln

An example of application of this model to the Italian data is given in Sabetta and Pugliese (1996).SP96

f2(R) = c log(R2 + h 2)1/2 + [d ] R h = h1 exp([h2] M)

log (Y) = a + b M - log (R2 + h 2)1/2 + e1 S1 + e2 S2 ± lny

S1 S2

= 1 for shallow soil sites

= 0 otherwise

= 1 for deep soil sites

= 0 otherwise

f1(M) = b1 M +[b2]M2 [ ] not statistically significant

Geometrical spreading Anelastic attenuation

Pseudo-depth

saturation with distance

saturation with M c = -1


F. Sabetta 7 17

New Models

Modification of the spectral shape according to earthquake magnitude

0,0

0,5

1,0

1,5

2,0

2,5

3,0

0,0 0,5 1,0 1,5 2,0

T(sec)

Spectr

al a

mplif

icatio

n

M=7 R=80 km PGA=0,06 g

M=5 R=15 km PGA=0,06 g

Two M -R pairs giving the same predicted PGA with different spectral shapes (Sabetta & Pugliese, 1996 attenuation rela-tion)


F. Sabetta 7 18

Examples

0.01

0.10

1.00

1 10 100 1000

Fault distance (km)

PG

A (

g)

M=5 stiff and deep soilM=5 shallow soilM=6 stiff and deep soilM=6 shallow soilM=7 stiff and deep soilM=7 shallow soil

0.2 g

0.10

1.00

10.00

100.00

1 10 100 1000

Fault distance (km)

PG

V (

cm

/s)

M=5 stiffM=5 soilM=6 stiff M=6 soilM=7 stiff M=7 soil

30 cm/s

SP96

Distance = Repi and Rjb

M=Mw if 5.5

M= ML if < 5.5

This dual magnitude scale avoids the sat-uration effects of ML for the larger earth-quakes and assures, over the entire range of magnitudes examined, a linear relation-ship between the logarithm of the peak values and the magnitude, as suggested by the definition of the Richter magnitude scale and by the Alternating Conditional Expectation (ACE) statistical procedure used in this study.

The Sabetta & Pugliese GMPE is, by now, long standing and overtaken by newer models but it maintains importance because is one of the two GMPEs used in the Italian seismic code. Fur-thermore is one of the few GMPEs including predictive rela-tions for Arias Intensity, duration, and vertical components

PGA and PGV attenuation Arias Intensity and duration attenuation


F. Sabetta 7 19

Albeit recording stations are very close (see following figure) and the soil is of the same type, the spectra are very different (example of intra-event aleatory variability)

PGA and PGV of the L’Aquila 2009 earthquake plotted as a func-tion of the distance from the sur-face projection of the fault (Rjb) and compared with attenuation equa-tions for stiff soil and normal fault conditions (AB07 =Akkar & Bom-mer, 2007; SP96 Sabetta & Pugliese, 1996; BA08 Boore & At-kinson, 2008) 1.E-03

1.E-02

1.E-01

1.E+00

0 1 10 100 1000Rjb (km)

PG

A -

larg

er

ho

riz c

om

p (

g)

58 S.M.AB07 stiff normal fltAB07+sigmaAB07-sigmaSP96BA08 Vs=800 normal flt

0.1

1.0

10.0

100.0

0 1 10 100 1000Rjb (km)

PG

V -

la

rge

r h

ori

z c

om

p (

cm

/s)

58 S.M.AB07 stiff normal fltAB07-sigmaAB07+sigmaSP96 BA08 Vs=800 normal flt

The data at zero distance are un-der-predicted by all the GMPEs. In the intermediate and large dis-tance range (10-300 km) the rec-orded data are overestimated, with BA08 providing the best fit at large distance because it includes an anelastic attenuation coeffi-cient.

GMPEs and L’Aquila data

0

200

400

600

800

1000

1200

1400

1600

1800

2000

0.0 0.5 1.0 1.5 2.0 2.5 3.0T (s)

PS

A (

cm

/s2

)

AQV EW soil=BAQV EW deconvol bedrockAQG EW soil=AAQU EW soil=BAQK EW soil=BSP96 Mw=6.3 Repi=5 kmSP96 -sigmaSP96 +sigma

Response spectra, at 5% damping, for the stations closest to the L’Aquila 2009 epicen-tre compared with the attenuation relationship of Sabetta and Pugliese (SP96), calculat-ed for the same magnitude and distance and for a soil A site, together with its confi-dence limits

SP96 and L’Aquila response spectra


F. Sabetta 7 20


F. Sabetta 7 21

Selection and use of attenuation relations

Reviews of attenuation relationships are provided by several authors and in particular by J. Douglas (2003-2017). There are many equations available in the literature (may be too many the last review from Douglas report 432 GMPEs developed in different parts of the world) and the selection of an appropriate equation can be a daunting task for the engi-neer.

A common requirement is in generally to select a GMPE derived from data obtained in the region of interest. However, provided that the tectonic environment, the geology and the seismicity are similar, the GMPEs obviously do not recognize the national borders

There are many parts of the world for which there are few or no strong-motion records and no attenuation relationships derived specifically for the region. In these cases it is nec-essary to adopt equations from elsewhere, trying to use equations from regions with compa-rable geology and seismicity. Another possibility is the Hybrid Empirical Method as sug-gested by Campbell (2003). Scaling factors for the empirical attenuations are derived from the ratio of theoretical simulation models developed for a “host region” (where the seismological parameters, as stress drop, are well constrained by strong motion data) and for a “target region” (where the seismological parameters can be derived from weak motion data).

In employing an attenuation relationship, it is important to ensure that the definitions of pre-dicted and independent variables are stated clearly and that appropriate adjustments of those variables are performed.

Main items to be considered in selecting attenuation relations

• geographical distribution of data;

• tectonic regime;

• recording instruments;

• processing technique;

• independent variables;

• definition of ‘rock’;

• dependent variable;

• functional form;

• regression method;

Differences in each of the above items lead of course to different estimates

Recent s.m. databases used in GMPEs

A new databank has been compiled, from the subset of European data included in EU pro-ject SHARE after metadata revision and uniform re-processing of waveforms (Akkar et al. 2013). The RESORCE databank includes 5.882 waveforms from 1814 earthquakes oc-curred in Europe and Middle East from 1967 to 2011 in the magnitude range from 2.8 to 7.8

NGA-West2 is an initiative coordinated by the Pacific Earthquake Engineering Research Center (PEER) to develop Next Generation Attenuation models (Phase 2) for shallow crustal earthquakes in active tectonic regions. http://peer.berkeley.edu/ngawest2/databases/

It includes worldwide ground motion data: 21.336 records from 599 events in the mag-nitude range from 3.0 to 7.9. Ground motions and metadata for source, path, and site con-ditions were subject to quality checks by ground motion prediction equation developers and topical working groups. [DOI: 10.1193/070913EQS197M].The database has one of the most comprehensive sets of meta-data, including different distance measure, various site charac-terizations, earthquake source data, etc.

http://peer.berkeley.edu/ngawest2/databases/


F. Sabetta 7 22

Comparison of recent Attenuation Relations

8 relationships: 5 Europe, 2 West. N. America (WNA), 1 Worldwide

Attenuation Relations selected in the GEM Project

GEM - GLOBAL EARTHQUAKE MODEL 2009-2014

http://www.globalquakemodel.org/


F. Sabetta 7 23

Comparison of PGA attenuation with distance for different GMPEs at Mw=5 and Mw=6.5.


F. Sabetta 7 24

Comparison of response spectra for different GMPEs at Mw=5 and Mw=6.5.

The differences among various GMPEs (epistemic uncertainty) are always inside the ± 1 sigma bound of a single GMPE (aleatory uncertainty > epistemic)


F. Sabetta 7 25

Aleatory uncertainty: Standard Deviation


F. Sabetta 7 26

corso di costruzioni in zona simica modulo di ... › data › earthquake_engineering › sei… ·...

Documents