Geoscience Reference
In-Depth Information
dictated by the results of probabilistic deaggregation analyses of seismic hazard at
many sites in Italy, showing that contributions of the
M
5 events to spectral dis-
placement hazard are negligible. On the other hand, the upper bound is close to the
largest magnitude (7
<
possibly overestimated) in the historical earthquake
catalogue in Italy (Gruppo di lavoro CPTI, 2004).
M
values of Japanese events,
representing the majority of the dataset, come from the F-Net broadband network
(www.fnet.bosai.go.jp/freesia/index-j.html), while most of those of the other events
are fromthe global CMT catalogue (www.globalcmt.org/CMTsearch.html).
.
41
±
0
.
12
,
Magnitudes of the selected crustal earthquakes in Japan do not exceed 6.8, so that all the
M
>
.
8 events in the dataset are from Europe, the Middle East, and the United States.
Theuppermagnitudeboundadoptedhasledtorejectingextensivelyrecordedeventssuch
as the 1999 (
M
7.6) Chi Chi earthquake.
6
Distance, depth, and focal mechanism
. The focal distance,
R
, was used, mainly for
consistency with the current model of seismic source zones (SSZ) for Italy, which asso-
ciatesan“effective”focaldepthtoeachsourcezone(zonesismiche.mi.ingv.it/documenti/
App2.pdf). The distance limit of 150km was dictated by the mentioned deaggregation
analyses.Inadditiontoexcludingallsubductionzoneeventsfromthecalibrationdataset
focal depths were restricted within about 22km, since the largest “effective depth” class
in the Italian SSZ model is 12-20km. For what concerns Japanese earthquakes, the K-
Net focal depths were chosen, after checking for a number of events that these are the
same as those reported in the Hi-Net database (www.hinet.bosai.go.jp), while most of
focal depths of the other events are taken from the global CMT catalogue. The smallest
value of
R
in the dataset is about 6km. Using a significant number of reverse and strike-
slip fault events with well identified source geometry in the selected set, the following
relationshipwasobtainedbetweenthefocaldistance
R
andthenearestdistance
R
f
from
the recording siteto theruptured fault:
r
2
R
(
km
)
=
10
.
70
+
0
.
99
R
f
(
km
)(
σ
R
=
6
.
37km
,
=
0
.
97
).
(2.1)
The distribution of the data used in terms of magnitude and focal distance is shown in
Figure 2.2. Events with reverse, normal and strike-slip fault mechanisms are all repre-
sented in the database; at this stage of the study, no dependence was yet introduced on
the type offocal mechanism.
Ground categories
. For consistency with CEN (2004), the four basic ground categories
A, B, C, and D defined therein mainly through the
V
s
30
propagation velocity index,
have been adopted for the classification of the recording sites. About 82% of the cali-
bration dataset derives from the K-Net database, with a velocity profile at the recording
sites known to a depth of 20m at most. This created an evident practical difficulty in
classifying the sites in terms of
V
s
30
. Four different methods were proposed in Boore
(2004) to estimate
V
s
30
from velocity data not reaching 30m depth. Three of them were
applied here, and a further one was independently developed and adopted. The simplest
method in Boore (2004) assumes that the lowermost measured velocity extends to 30m,
but this choice may lead to underestimating
V
s
30
because
V
s
generally increases with
