Geoscience Reference
In-Depth Information
3. Modelling site effects. The importance of the model
If site effects cannot be estimated from measurements, it may still be possible to model
them.Infact,manypaperspresentbothempiricalestimatesandnumericalresults.When
thetwocoincide,theirmutualvalidationstrengthenstheconfidencewithwhichtheresults
canbeusedinseismicriskreduction.Whenempiricalestimatesarelacking,itmightstill
bepossibletoevaluatelocalamplificationbycomputingtheseismicresponseofthelocal
structureresponsible for thisamplification.
The first question to address when modelling site effects is the dimensionality of the
model. We may choose among 1D, 2D, or even 3D models (recall that this paper passes
overthecontroversylinear/non-linear),whichrefertothepossibleirregularitiesthatmay
beincluded.1Dmodels,thosethatconsiderheterogeneitiesonlyintheverticaldirection,
have been the preferred choice because of their simplicity, reliability (which is taken
here to refer to the stability of the results and not necessarily to their trustworthiness),
and the possibility of generalising the results. Cities are usually built on alluvial valleys,
with more or less flat sediments overlying a rocky basement. If the boundaries of the
valleyarefarfromthesiteofinterestrelativetothedepthofthesediments,intuitionsug-
gests that 1D models are appropriate. Moreover, as it has been already mentioned, local
amplificationismostlyduetoimpedancecontrastbetweensoillayersandtheirbasement.
Whenthisisthegoverningphenomenon,1Dmodelsprovideanexcellentestimateofsite
effects, and one that needs the least number of parameters. However, a word of caution
is warranted. Intuition may be wrong here. For example, Euroseistest valley (Figure 3.4)
looksasanidealcasewherethe1Dapproximationwouldworkwell.Itis6kmwideand
the sediments are only 200m thick. Euroseistest has been extensively studied during the
last decade and its seismic response is well known, much more than what is the usual
case in Earthquake Engineering. For this reason, Euroseistest can be considered to be a
benchmark to evaluate different ideas concerning site effects. Makra et al. (2004) used
STI
250
200
PRO
150
100
TST
A
50
0
E
−
50
F
−
100
G*
G
−
150
−
200
−
16500
−
16000
−
15500
−
15000
−
14500
−
14000
−
13500
DISTANCE (m)
Fig. 3.4. 2Dmodel builtfor a NW-SE cross-section of Euroseistest valley
(After Raptakis et al., 2005.)
