Geoscience Reference
In-Depth Information
used for European regions either do not cover the entire period range of interest (e.g.
Ambraseys et al., 1996; Sabetta and Pugliese, 1996; Bommer et al., 2000; Fukushima
et al., 2003; Ambraseys et al., 2005), or were found not to be sufficiently reliable at long
periods(e.g.Berge-Thierryetal.,2003),presumablybecauseofthehigh-passfilteringof
the analog accelerograph recordings making up most of their calibration data sets.
Therefore,anewworldwidedatabasewasassembledconsistingonlyofdigitallyrecorded
accelerogramsofshallowcrustalearthquakes,carefullyselectedasregardsgroundcondi-
tion at the accelerograph station, minimal long period disturbances in the recording, and
fullestpossible coverage inthe required magnitude and distance ranges.
Thefollowingsubsectionsdescribethedataselectionandtheuniformprocessingapplied,
the features of the empirical predictions obtained for, the
DRS
of both vertical and hori-
zontal motion, and the formulationof a simplifiedspectral model for design.
2.1. DATA SELECTION
Asiswellknown,inselectingthecalibrationdataforstrongmotionattenuationrelations
one should avoid to adopt both excessively loose criteria, because they lead to unreliable
estimations, and the use of overly restrictive criteria that will result in lack of robustness
of the predictions.
Strong motion data sources and format
. By far the largest contributing source to the
data set was the Japanese K-Net strong motion network (www.k-net.bosai.go.jp), while
the rest of the data are from California, Europe, Iran, and Turkey. Additional data
from Japan are those of the 1995 Hyogo-ken Nanbu earthquake, that were provided
at the time of the Simultaneous Simulation Experiment of the Second International
Symposium on ESG, in 1998 (ESG98 data distribution CD-ROM for the Kobe Simul-
taneous Simulation), and 10 records at rock sites, taken from the Kik-Net network
database (www.kik.bosai.go.jp). The source of Iran data was the Iran Strong Motion
Network, ISMN (www.bhrc.ac.ir/Bhrc/d-stgrmo/D-StGrMo.htm); for California we
mainly used data available from the US National Strong-Motion Network, NSMN
(nsmp.wr.usgs.gov) and, for the 1999 Hector Mine event, digital data from both the
Engineering Strong Motion Data Center, CISN (www.quake.ca.gov/cisn-edc) and from
the Southern California Seismic Network, SCSN (www.scsn.org). Concerning Europe
and Turkey, most data come from the strong-motion data archive at the Imperial College
of London (Ambraseys et al., 2002). A few (about 10) Italian records were provided by
the National Accelerograph Network (RAN).
All data were acquired in the form of uncorrected acceleration histories, including nine
recordsofthe1980Irpinia(Italy)
M
6.9earthquake,whicharetheonlyanalogrecordings
present in the database and were inserted in it only after a careful scrutiny of their long
period characteristics.
Magnitude
.Crustalearthquakesworldwidewithindependentlydeterminedmomentmag-
nitude were considered, in the 5
.
0
≤
M
≤
7
.
2 range. The lower magnitude bound was
