Seismic Source Quantitative Parameters Retrieval from InSAR Data and Neural Networks

Tor Vergata University

The basic idea of this thesis is to exploit the capabilities of neural networks in a very new
framework: the quantitative modelling of the seismic source and the interferogram inversion for
retrieving its geometric parameters. The problem can be sum up as follows. When a moderateto-
strong earthquake occurs we can apply SAR Interferometry (InSAR) technique to compute a
differential interferogram. The latter is used to detect and measure the surface displacement
field. The earthquake has been generated by an active, seismogenic, fault having its own specific
geometry. Therefore each differential interferogram contains the information concerning the
geometry of the seismic source the earthquake comes from; its shape and size, the number of
fringes, the lobe orientation and number are the main features of the surface effects field. Two
problems have been analysed in this work. The first is the identification of the seismic source
mechanism. The second is a typical inversion exercise concerning the fault plane parameter. To
perform both exercises of the seismic fault a huge number of synthetic interferograms has been
computed. Each of them is generated by a different combination of such geometric parameters.
As far as the retrieval of the geometric parameters is concerned an artificial neural network has
been properly generated and trained to provide an inversion procedure to single out the
geometric parameters of the fault. Five among these latter, Length, Width, Dip, Strike, Depth,
have been simultaneously inverted. The result is in agreement with those results based on
different approaches. Furthermore the method seems very promising and leads to improve the
studies concerning the combined use of neural networks and InSAR technique.

Dr. PhD. Salvatore Stramondo
Istituto Nazionale di Geofisica e Vulcanologia
National Earthquake Center - Remote Sensing Lab.
Via di Vigna Murata 605
00143 Rome
Italy
e-mail: stramondo@ingv.it
Phone: +39 06 51860521
Fax: +39 06 5041181