 |
| Titel |
Imaging the crust of the western Alps using travel times from 20,000 earthquakes. |
| VerfasserIn |
Bertrand Potin, Bernard Valette, François Thouvenot, Vadim Monteiller |
| Konferenz |
EGU General Assembly 2014
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 16 (2014) |
| Datensatznummer |
250095610
|
| Publikation (Nr.) |
 EGU/EGU2014-11073.pdf |
|
|
|
|
|
| Zusammenfassung |
| Located between France, Italy and Switzerland, the western Alps are a complex area were
dense seismic networks have recorded a few ML < 2 crustal earthquakes per day in the past
25 years. By merging data from more than 200 stations distributed in 5 different local
networks, we built up a 20,000-event database over a 400 x 430km2 area. This
database provides a unique opportunity to study the details of the western Alps
structures, such as the Ivrea upper-mantle body (Italy) or the Digne sedimentary nappes
(France).
Our method to carry out a high-resolution tomographic imaging of the Alpine crust and
upper mantle is based on the inversion of Pg, Pn, Sg and Sn-wave arrival times from crustal
events, for V P and V P-V S fields as well as for event locations and station site-effect
residuals. Because of the large size of the model box, we take the ellipticity of the Earth into
account, and of course its surface topography. In order to model the Moho discontinuity, we
build up an a priori Moho topography based on a selection of different studies published in
the past 30 years. Using two different velocity models for the crust and the mantle separated
by this a priori Moho discontinuity, we perfectly model the velocity jump and we also
decorrelate those two domains. Local adjustments to this a priori Moho topography are
performed, based on the velocity model obtained after a few iterations of the inversion
process.
To invert this 600,000 data set we use a non-linear least-squares approach based on a
stochastic description of data and model. The forward computation of travel times in the 3D
model is performed by integrating slowness along the rays, which are determined by the
Podvin-Lecomte algorithm (basically a finite-difference resolution of eikonal equation). The
regularization of the velocity fields is achieved through a covariance norm on V P and
V P-V S, the kernel of which is of exponential type. The smoothing and damping parameters
are adjusted by means of L-curves analysis.
In the past 25 years, a few events have caused damages in the western Alps and local
networks have monitored a few swarms. Some faults have also been discovered
just by looking at hypocentre alignments of low magnitude events, such as the
90-km-long Belledonne fault near Grenoble (France). Accurate locations are crucial to
understand and describe this kind of fault. This new 3D velocity model, where a
realistic Moho discontinuity allows us to work with both direct and refracted P
and S waves, can be used to locate hypocentres using a probabilistic approach. |
|
|
|
|
|
|