Velocity structure, seismicity and seismotectonics of northwestern
Greece between the Gulf of Arta and Zakynthos
ETH Zürich, Zürich, Switzerland, (1998)DOI: 10.3929/ethz-a-002025706
Abstract
The Ionian region (western Mainland Greece and the Ionian islands)
plays an important role in the geodynanmics of the eastern Mediterranean.
At the triple junction of the African, Eurasian and Anatolian/Aegean
plates it is located in the transition zone between the Hellenic
oceanic subduction and the continental collision zone of the Hellenides/Dinarides.
This region exhibits a very high seismicity and one of the highest
crustal strain rates observed in Europe, Tn the framework of an ETH-Zürich
research project, GPS satellite geodesy, differential SAR interferometry,
and seismologic studies were combined in a detailed investigation
of the geodynamic processes in the Ionian region to detect temporal
variations of the stress and strain fields and to contribute to the
seismotectonic modelling in this region. This work reports on the
seismological part of the project, from data acquisition to final
3D structural models. A dense temporary seismic network of 70 digital
stations was operated during three months in the summer of 1995 in
an area from north of the Gulf of Arta to south of the Gulf of Patras
and on the lonian islands. Great efforts were undertaken to ensure
the accuracy of the timing system and station position determination.
For 1179 locatable events about 14300 P and 6100 S arrival times
were, determined, with an average onset time uncertainty less than
0.05 s for P arrivals . Well locatable events were used to calculate
minimum 1D models for subregions and for the whole study area. Resulting
station delays generally reflect the surface geology of the area.
The average P velocity of the upper 20 km varies between 6.2 - 6.5
km/s, with the higher values in the areas of thinner crust in the
southwest. S velocities are rather low with an average Vp/Vs ratio
of 1.85. Seismicity is confined to the upper 15 km in the west, whereas
in the northeast of the study area around the Katouna fault zone
(KFZ), hypocentral depths reach about 2.5 km. 7300 P and 3300 S observations
from 490 well locatable events were finally used in the inversion
for the 3D crustal velocity structure of the Ionian area. Careful
assessment of resolution and reliability, using synthetic tests,
standard resolution estimates, and the comparison of two different
ray tracers, allowed to reliably discriminate well resolved areas
in large parts of the study area to depths of 12-15 km. Between 12-N
km the central part of the network is still fairly well resolved.
The interpretation of the 3D velocity structure together with the
accurate hypocenter distribution and preliminary focal mechanisms
for selected events reveals that in the upper 5 km structures generally
correlate well with surface geology, results from seismic reflection
profiling, (7 and tectonic modelling of basin evolution in this area.
The KFZ is clearly imaged as a zone of low velocities in the upper
5 km, surrounded by zones of high velocities to the west, beneath
the limestones of the Pergandi Mts., and to the east, beneath the
foothills of the Pindus Mts, The majority of the deep seismicity
around the KFZ is connected to an eastward dipping body of relatively
low velocities which can be followed to 15-20 km depth. The deeper
structure beneath the island of Cefalonia, where slightly eastward
dipping hypocenters correlate with a general eastward dip of velocity
structures, can probably be linked to the ongoing subduction of the
Ionian oceanic crust beneath the Aegean. The diversity of focal mechanisms
observed at depths between 10 -1s km may tentatively be interpreted
as expression of internal deformations in the upper and lower plate.
Earthquake occurrence is confined to the east of the bathymetric
trench west of the Ionian islands, which is connected to the Cefalonia
fault. Hypocenters along or the eastern margin of that trench are
mainly confined to 8-15 km depth and predominantly show normal faulting
with NE-SW trending nodal planes. NO indication for strike-slip motion
along the Cefalonia fault, as it is inferred from large earthquakes
and geodetic data, was observed in this data set. This work presents
new information on the velocity structure in a rapidly deforming
region. From the results it may be inferred that a large part of
the geodetically observed NE-SW extension is taken up by aseismic
deformation of the middle and lower crust, at least in the northern
part of the study area. This deformation may lead to the complicated
pattern of internal def&mations inferred from the heterogeneous
velocity structure in the upper 10 km and the diversity of observed
focal mechanisms. From the very hererogeneous velocity structure
obtained by the tomographic inversion rose the question about the
accuracy of the implemented approximate 3D bending ray tracing scheme
when applied to a large area. To assess possible defects of the 3D
solution due to erroneous ray tracing, an alternative 3D shooting
method was implemented in the tomography code. A comparison of the
two ray tracing schemes revealed that for rays longer than around
60 km, ray paths can be significantly different, while the differences
in travel times may be as large as SO ms, which is within the average
onset time accuracy. Consequently, the different ray tracing schemes
affect more the resolution matrix than the velocity solution. On
one hand, these results reveal the ambiguity contained in resolution
estimates for 30 tomography and the difficulty to clearly define
a significance level for these estimates. On the other hand, the
similarity of the velocity solutions obtained with the different
ray tracers assures that adequate model parametrization and solution
display suppress artefacts due to algorithmic inaccuracies. The comparison
of different ray tracing schemes in local earthquake tomography (LET)
and their influence on solution and resolution estimates leads to
the conclusion, that future developments in LET are necessary to
better access the dependence of solution and resolution estimates
on model parametrization, forward, and inverse solution schemes.
Only then will it become, feasible to treat the increasing number
of high quality data sets covering large areas and to retrieve reliable
high resolution images of the earth's structure.