Page 287 - Numerical Analysis and Modelling in Geomechanics

268 THE 2-D CHEBYSHEV SPECTRAL ELEMENT METHOD
for the Defence Against Earthquakes (CNR-GNDT) and devoted to evaluating
the seismic risk of a highly urbanised area, such as that of Catania, located in a
seismically active region. All details about the study are reported in Priolo
(1998, 1999, 2000).
The ground motion was calculated along four vertical transects (t01, t02, t03,
and t05 in Figure 9.2), which span the Catania municipal area along four different
directions and provide a good spatial sampling of the northern part. The model
along these transects represent the upper 20 km of the Earth’s structure in terms
of seismic velocities, density, and attenuation. Particular emphasis was given to
the definition of the surface structure, with the finest local detail. Data consist of
three geological profiles, several pre-interpreted seismic lines, data relative to
deep wells, a complete geotechnical survey of the area, and several studies
regarding Eastern Sicily and the area surrounding Catania, in particular. On a
regional scale, the main units of the crustal structure are: (i) the carbonatic
basement of the Hyblean Foreland, (ii) the sedimentary formations of the
Northern Chain, (iii) the volcanic body of Mt. Etna, (iv) the Ibleo-Maltese
escarpment running offshore in the NNW-SSE direction, and, on a smaller scale,
(v) the Gela-Catania Foredeep, with the sedimentary basin of the Catania Plain.
Table 9.1 summarises the parameter values adopted for the main formations in
the models.
The reference earthquake simulates the M7 event of January 11, 1693. This
destructive event is commonly associated to rupture with normal mechanism
along the Ibleo-Maltese system of faults. This is a system of sub-vertical normal
faults, NNW-SSE oriented, which runs for about 70–100 km offshore along the
Ionian coast of Sicily. The reference earthquake of this study is associated to the
northern part of the system, which is simplified in a segment about 25 km long
(s. IBM in Figure 9.2). Table 9.2 summarises the values of the main source
parameters. The source mechanism is of pure normal faulting. The wavefield
amplitude is scaled by the value assumed for the fault-slip D.
Two groups of sources are considered. The first group uses a point source
model with the aim of studying the effect induced by a variation in the fault
orientation and source position. With the second group, the purpose is to
simulate an extended source. To do this, the fault is discretised into three
elementary point sources, aligned along the fault dip direction (Figure 9.3). In
this way, three different directions of rupture propagation are reproduced.
The main results of this study are synthetic seismograms, peak ground
acceleration (PGA) envelopes, and response spectra. Seismograms are computed
up to a maximum frequency of 7.5 Hz and for a total propagation time of 25 s.
The time step is 15 ms. On average, the size of the computational models is 45
km×25 km and the meshes contain 170,000–200,000 nodes (Figure 9.3).
Figure 9.4 shows snapshots of wave propagation through transect t02. The
main pressure and shear wavefronts—P −P − and S −S , respectively—can be
+
1
2
clearly distinguished, as well as the reflected wavefronts ( ) Figure 9.5
summarises the spatial distribution of PGA. On average, values range between 0.