FIELD MAPPING AND FAULT MODELING OF THE Mw=5.9, SEPTEMBER
7, 1999 ATHENS EARTHQUAKE BASED ON ERS-2 SATELLITE RADAR INTERFEROMETRY
P. Elias(1), O. Sykioti(1),
P. Briole(2), D. Remy(2), M. Sachpazi(3)
, G. Veis(4), I. Kotsis(4)
Research Letters, Vol. 27, No. 24, 3989-3992, December
(1) National Observatory
of Athens, Institute for Space Applications and Remote Sensing,
Metaxa & Vas. Pavlou, 152 36, Palea Penteli, Athens, Hellas,(email:
(2) Institut de Physique du Globe, Paris, 4 Place Jussieu,
75005 Paris, France,(email: email@example.com).
(3) National Observatory of Athens, Institute for Geodynamics,
Lofos Nimfon, 118 10 Athens, Hellas,(email: firstname.lastname@example.org).
(4) National Technical University of Athens, Department
of Surveying Engineering, 15780, Athens, Hellas,(email: email@example.com).
National Observatory of Athens
CO-SEISMIC SURFACE DEFORMATIONS
MODELING OF THE OBSERVED DEFORMATION FIELD
On September 7th, 1999 at 11h
56m 50s UT a magnitude Mw=5.9
earthquake struck the area of Attica. It was strongly
felt by the Athenian population, caused heavy damages and
made several buildings collapse killing 143 people. Its epicenter
is located at 38.10oN; 23.56oE which
is about 20km NNW from the Athens center, between the main
rifts of Evia and Corinth, at the oriental extremity of the
In the area around the epicenter, several normal
faults dipping northward or southward are clearly expressed
on the seismotectonic map (Plate
1a). However, the absence of any seismic rupture at
the surface after the earthquake does not allow a direct field
identification of the seismogenic fault. Seismological operations
for the purpose of recording the aftershock activity were
launched immediately by several state and research organizations.
At the same time geodetic data were collected with terrestrial
and satellite methods. These data are currently at the stage
Soon after the earthquake a parallel operation
using radar interferometry, was initiated at the
Institute for Space Applications and Remote Sensing of the
National Observatory of Athens (ISARS/NOA) in order to measure
the surface deformation and its characteristics. The study
was done in collaboration with the National Technical University
of Athens/Department of Surveying Engineering, Institut de
Physique du Globe de Paris (IPGP) and the Geodynamics Institute
of NOA (NOAGI).
To capture the deformations caused by the main
shock and aftershocks, interferograms were constructed from
the phase difference recorded on ERS-2 satellite images. Interferometric
analysis of SAR images has demonstrated potential to monitor
and measure surface deformations associated with earthquakes
[Massonet et al., 1993, Zebker et al., 1994,
Murakami et al., 1996]. In Greece the same technique has
been used for the study of Kozani (13/5/1995. Mw=6.6) and
Aigion (15/6/1995, Mw=6.1) earthquakes [Meyer et al.,
1996; Bernard et al., 1997].
In order to isolate surface deformations from
the interferograms, it was necessary to remove the topography.
The elimination of the topography was done by subtracting
a synthetic fringe pattern produced by using the Digital Elevation
Model (DEM) of the area of interest from the interference
pattern resulted by the two SAR images. The DEM was produced
by digitizing contour lines and spot height data shown on
existing 1 :5000-scale topographic maps. The accuracy of the
DEM was tested with a set of independent control points of
known elevation and the estimated RMS error was of the order
The selection of the interferometric pairs
to process was based on their sensitivity to the topography,
expressed by the altitude of ambiguity (ha).
Only image pairs with ha values greater
than 5 times the expected DEM error, were processed in order
to avoid residual fringes due to topography. The magnitude
of the expected topographic artifact expressed in phase cycles
is of the order of 0.2 phase cycles or 5.6mm in range, for
the worst co-seismic pair (ha = 50 m) and 0.075
cycles or 2.1 mm in range for the best co-seismic pair (ha
Table 1 illustrates the
time spanning of the interferometric pairs used and their
corresponding ha values. The images collected for
interferometric calculations were acquired at the descending
pass of the ERS-2 satellite.
From the interferograms it is apparent that
the main shock and aftershocks have induced a co-seismic surface
deformation, which appears with at least two concentric, but
not symmetric, fringes centered at 38.10oN; 23.60oE.
This point is located in a distance of less than 3 km away
from the main shock epicenter. The fringes indicate a change
of 56mm in range. The concentric fringes are apparent
in all co-seismic interferograms spanning the periods [December
28th, 1995 to September 23rd, 1999],
[November 27th, 1997 to September 23rd,
1999], [September 19th, 1998 to October 9th,
2a shows one of the calculated interferograms, spanning
the period [September 19th, 1998 to October 9th,
It is important to notice that the interferograms
composed from images acquired before the seismic event (Table
1), do not show fringes in the affected by the earthquake
area. Unfortunately we could not find appropriate SAR images
just before the Athens earthquake to create interferograms,
in order to investigate for any pre-seismic deformation signals.
It is worth-noting that the difference
of co-seismic interferograms belonging to the same track do
not show fringes, indicating that the surface displacements
are rather constant in shape, position and magnitude.
THE OBSERVED DEFORMATION FIELD
The calculated co-seismic interferograms were
used to sample slant-range displacements along the two observed
fringes. The sampled slant range data were used to fit
a simple inversion model, assuming for the earthquake a dislocation
of a rectangular fault in homogeneous elastic half-space.
The forward algorithm used is that developed by Okada,
. The inverse algorithm is that developed by Briole
et al.,  using a least squares approach proposed
by Tarantola and Valette, .
The model predicted two faults. A main one which
is responsible for more than the 90% of the energy released
and a secondary smaller fault, which was added to fit the
fringe asymmetry to the east. Due to the large predominance
of the main fault segment, the parameters of this secondary
structure are not well constrained by the data and are less
reliable than those relative to the main segment.
The parameters of the two modeled fault segments
are shown in Table 2.
2b shows the synthetic interferogram derived by the model.
Based on the model results the main
fault segment has experienced a slip of 50cm at the depth
of 8km resulting in the observed surface deformations.
No slip above the depth of 8km is predicted for the main fault
segment. The fault segments intersect the Earth surface
at points 38.17oN; 23.53oE and 38.14oN;
23.76oE, defining a trace, which lies about 5 km
northern to the Thriassion depression, crosses the Fili mountain
and shows a WNW-ESE trending.
The resulted fringes are bounded by the Fili
mountain in the NE and the Aegaleo mountain in the SE (Plate
2a). They define an active zone of more than 20km
in the E-W and 10km in the N-S directions. The area defined
by the fringes encompasses the vast majority of the located
epicenters, extending S-W to the Fili fault with a striking
along a WNW-ESE axis (Plate
1b) (Papadopoulos et al., 1999). However, a high concentration
of aftershocks is also observed at the eastern border of the
Thriassion depression along the Aegaleo mountain. If this
is not due to the geometry of the seismograph array, it indicates
the activation of other faults parallel to the direction of
the Aegaleo mountain. This is consistent with the results
obtained from the focal mechanism study done by Tselentis
and Zahradnik, . It should be noted that the observed
fringes follow the same direction in the south. Also the fact
that no fringes are observed beyond the root of the Aegaleo
mountain, it corroborates the opinion of other researchersthat
this mountain acted as a boundary, preventing further activation
to the south-east direction (V. Papazachos, oral communication).
The absence of shallow seismic activity
prevented the rupture propagation towards the Earth surface
and hindered the direct identification of the seismogenic
fault. N.N. Ambraseys (internal field report) after his visit
at the earthquake site confirms that the two faults located
in the epicentral area, that is the Aspropyrgos and the Fili
1a), nowhere showed any field evidence of reactivation.
This makes Ambraseys to draw the conclusion that the fault
associated with the Athens earthquake it could well has been
one of the many not mapped, not clearly visible, or most probably
blind faults. Currently several studies dealing with aftershock
recordings are in progress (Papanastasiou et al., 1999). However,
the first results of this study suggest that the activation
of a fracture zone located at the Fili mountain, is more likely
associated with the Athens earthquake resulting in the observed
Thanks are due to prof. N.N. Ambraseys
for making available to us his internal field report on the
Athens earthquake and for his comments on the paper. We are
grateful to the European Space Agency for providing ERS SAR
data. Sincere thanks to Dr. George Stavrakakis, director of
the Geodynamics Institute of NOA (NOAGI) for providing the
aftershock data recorded by the local seismograph array in
the period September 8th to October 5th
1999. Thanks are also due to Helene Vadon and Kurt Feigl for
their support in using DIAPASON sw tool (CNES), and Antonio
Avallone (IPGP) for his contribution to the fault modeling.
Bernard P., P. Briole, B. Meyer, H. Lyon Caen,
J.-M. Gomez-Gonzalez, C. Tiberi, R. Cattin, D. Hatzfeld, C.
Lachet, B. Lebrun, A. Descamps, F. Courboulex, C. Larroque,
A. Rigo, D. Massonnet, P. Papadimitriou, J. Kassaras, D. Diagourtas,
K. Makropoulos and G. Veis, The Ms=6.2, June 15, 1995 Aigion
earthquake (Greece): Evidence for low normal faulting in Corinth
rift, J. of Seismology, 1, 131-150, 1997.
Briole P., G. De Natale, R. Gaulon, F. Pingue
and R. Scarpa, Inversion of geodetic data and seismicity associated
with the Friuli earthquake sequence (1976-1977), Annales
Geophysicae, 4, 481-492, 1986.
Massonnet D., M. Rossi, C. Carmona, F. Adragna,
G. Peltzer, K. Feigl, T. Rabaute, The displacement field of
the Landers earthquake mapped by radar interferometry, Nature,
364, 138-142, 1993.
Meyer B., R. Armijo, J. B. Chabalier, C. Delacourt,
J. C. Ruegg, J. Acache, P. Briole, D. Papanastassiou, The
1995 Greneva (Northern Greece) Earthquake: Fault model constrained
with tectonic observations and SAR interferometry, Geophysical
Research Letters, 23(19), 2, 677-2, 680, 1996.
Murakami M., M. Tobita, S. Fujiwara, T. Saito,
H. Masaharu, Co-seismic crustal deformations of the 1994 Northridge,
California, earthquake detected by interferometric JERS 1
Synthetic Aperture Radar, Journal Geophysical Research, 101,
Okada Y., Surface deformations due to shear
and tensile faults in a half-space, Bull. Seism. Soc. Am.,
75, 1135-1154, 1985.
Papadopoulos G.A., Baskoutas I.,Chouliaras G.,
Drakatos G., Kalogeras I., Karastathis V., Kourouzidis M.,
Latoussakis I., Makaris D., Melis N., Panopoulou G., Papanastassiou
D., Pappis I., Tassos S., Plessa A. and G. Stavrakakis, 1999:
Seismological aspects of the Athens earthquake of 7th
September, 1999: Preliminary Results. "1st Conference
Advances on Natural Hazards Mitigation - Experiences from
Europe & Japan, Athens 3-5 Nov. 1999, Abstracts &
Papanastassiou,D., Stavrakakis,G. Drakatos,G.
and Papadopoulos,G., 1999. The Athens, September 7, 1999,
Ms=5.9, Earthquake: First Results on the Focal
Proprerties of the Main Shock and the Aftershock Sequence.
"Ann. Geol. des Pays Helleniq.", submitted.
Tarantola A. and B. Valette, Generalized nonlinear
inverse problem solved using the least squares criterion,
Rev. Geophys. Space Phys., 20, 219-232, 1982.
Tselentis G.A., J. Zahradnik, The Athens earthquake
of September 7, 1999, paper submitted to Bull. Seism. Soc.
Zebker H.A., P.A. Rosen, R.M. Goldstein,
A. Gabriel, C.L. Werner, On the derivation of co-seismic displacement
fields using differential radar interferometry: The Landers
earthquake, J. Geophys. Res., 99, 19, 617-19, 634, 1994.
|Table 1. ERS-2 Image Combinations
Used for Interferometric Calculations. Shaded Lines
Correspond to Co-seismic Frame Acquisitions. The correspo-
nding Altitude of Ambiguity Values are Illustrated.
Acq. Date Image No1
Acq. Date Image No 2
Alt. of Ambig.
|Table 2. Parameters of the Two Modeled
Faults. The RMS Error for Data Fit is 6mm (0.2 fringes).
East location of the center of
the upper edge of the fault
38° 6' 29''
38° 6' 30''
North location of the center
of the upper edge of the fault
23° 35' 47''
23° 42' 23''
Depth of the upper edge of the
Half length of the fault d(km)
Width of the fault
Deep slip (positive
normal) D (mm)