Geoscience Reference
In-Depth Information
sites. They advocate the use of seismic stations at the bottom of deep boreholes. Steidl
etal.(1996)showthatdeepboreholerecords,whencorrectedfortheeffectofdownward
propagating waves, are the more reliable records to use as reference. This conclusion
is well substantiated but impractical in most places, especially in developing countries,
becauseofitslargecost.Acheaperalternativeistokeepthoselimitationsinmindandto
restrict the analysis to frequencies not larger than 5 to 6Hz. There are several justifica-
tions for this limit. First, it would avoid problems with the possible appearance of local
amplification at the reference site alluded to above, whose consequence is the underesti-
mation of amplification at the soft soil sites. In addition to this, as frequency increases,
the distance between soil and reference sites also increases in terms of wavelength, and
this could invalidate the assumptions behind spectral ratios. A further argument is that,
as frequency increases anelastic attenuation in the sediments becomes more important,
possibly eliminating the need to estimate ground motion at those frequencies. Anyhow,
it is very likely, that for frequencies larger than 5 or 6Hz, in most cases it will be very
difficult to relate observed amplification to geological or geotechnical characteristics of
the site of interest. If shear-wave velocity is about 150m/s, wavelength will be 25m at
6Hz and 15m at 10Hz. It becomes very difficult, and probably not very useful, to detail
the subsoil structureand its lateral changes at this scale.
An unorthodox approach to a reference site was taken by Hruby and Beresnev (2003).
They proposed to use a synthetic record computed for the site of interest as the refer-
ence against which to measure amplification in an actual seismic record. The synthetic
recordwascomputedbythoseauthorsusingthestochasticfinite-faultmodellingmethod
(Beresnev and Atkinson, 1998). The idea is interesting and may be a way out of the
dilemma posed by the impossibility of finding an acceptable reference site. Hruby and
Beresnev(2003)validatedthisapproachfortheLosAngelesbasin,forwhichniceresults
were obtained. However, this procedure cannot be generally applied at present because a
modeloftheregionalstructuremodelforthesiteunderstudyisnecessary.Moreover,this
modelneedstohavebeenvalidated.Thisrequirementcannotbesatisfiedexceptforvery
few locations around the world. The limitations faced by the building of a model will be
discussed in moredepth in the following section.
Unfortunately, it is not infrequent that an adequate reference site is impossible to find.
Forthisreason,twoothertechniqueshavebeenapplied,whichdonotrequireareference
site. The first one is the inversion of a set of Fourier spectra for many stations and events
(Boatwright et al., 1991; Field and Jacob, 1995; Raptakis et al., 2005). This technique,
similarly to spectral ratios, has the advantage of having solid physical bases. Moreover,
the results include estimates of the magnitude of the seismic events and of the regional
Q factor that bests fits all the data. However, this method is really reliable only when
the dataset is large; a significant number of events recorded simultaneously by several
stations is necessary toobtain areliableresult.
The second technique that does not require a reference site uses spectral ratios of hori-
zontal components relative to the vertical component motion recorded at the same site.
This technique is a frequency domain application of the receiver functions proposed by
