Geoscience Reference
In-Depth Information
(2004) are also depicted, although the classification of ground categories adopted in the
former do not coincide withthe present one.
The ratio B/A yielded by (2.2) is constant and equal to about 1.2 for
T
0s, while it
exhibitssignificantbroad-bandamplificationbetween0.2sand5.0s,reachingan(appar-
ent)peakbetween0.7sand0.8s.ThreemainindicationsmaybederivedfromFigure2.4:
(1) The difference between ground categories B and A is never negligible, since ampli-
fication ranges between 20% (at long period) and 75% (at
T
>
5
.
1s). (2) Whichever the
dataset used tocalibrate the attenuation model, spectral ratiosB/Atendtoagree interms
of median amplification value, with some differences in the peak period. (3) Amplifica-
tion factors obtained from CEN (2004) are too conservative for
T
<
>
2s, and vice versa
at lower periods.
Class C ground.
The C/A spectral amplification is portrayed on the rhs graph of
Figure 2.4, which contains results yielded by both (2.2) and (2.4); in the first case
thefactorshownasafunctionofperiodis
a
5
whileintheseconditis
b
6
10
b
7
M
.Astrong
peak occurs at 1s, with maximum amplification between about 2.7 and 4. For clarity,
amplification curves from Sabetta and Pugliese (1996) and Ambraseys et al. (2005) are
not shown in this case; the peak period of Sabetta and Pugliese (1996) for deep alluvium
sites is about the same as in this study, but peak amplification does not exceed 1.7.
The peak period fromAmbraseys et al. (2005) is close to 2.0s, withan ordinate of about
2.5. Median amplification curves yielded by (2.4), calibrated with the worldwide data-
base,areshowninthesamefigureforthree
M
values;theyshow,asexpectedonphysical
grounds, decreasing amplification when magnitude increases. The curves in question lie
fairly close to or within the s.e. band of coefficient
a
5
of (2.2), thus indicating that an
attenuation model non-linear in the site coefficients, such as (2.4), results in a marginal
improvement in statistical significance with respect to Eq. (2.2). Hence, the latter has
been generally preferredinthe applications.
H/V spectral ratios
. An important source of uncertainty in predictions of the ground
motion spectral amplitudes is likely to be due to errors in the attribution to ground cat-
egories based only on a geotechnical/geological site description or, ultimately, on
V
s
30
values. Following Zhao et al. (2006) a preliminary investigation was made on the feasi-
bility of subdividing the Japanese data subset into ground categories based on the H/V
ratio of 5% damped horizontal to vertical spectra. Thus, the H/V ratios were computed
for 0
2s for about 950 Japanese records, using for start the ground cate-
gories A, B, C, D, and picking from the ratio computed for each accelerogram the peak
value and the corresponding period. Finally, for each ground class, the average and stan-
dard deviation of the peak value and the dominant period associated to it were derived.
The evident aim was to detect correspondences between the geotechnical classification
adoptedandthepeakperiodsoftheH/Vspectralratios,inthehopeofgaininginsightsas
to the classification of stations with unknown soil conditions included in the worldwide
database. Results, listed in Table 2.3, show that the uncertainty associated to the average
dominantperiodandtoaveragepeakH/Vvalueofeachgroundclassislarge,suggesting
that a ground classification based on thesetwofactors would bring no improvement.
.
05s
≤
T
≤
