Combining InSAR and GNSS to Track Magma Transport at Basaltic Volcanoes
Résumé
The added value of combining InSAR and GNSS data, characterized by good spatial
coverage and high temporal resolution, respectively, is evaluated based on a specific event:
the propagation of the magma intrusion leading to the 26 May 2016 eruption at Piton de la Fournaise
volcano (Reunion Island, France). Surface displacement is a non linear function of the geometry
and location of the pressurized source of unrest, so inversions use a random search, based on a
neighborhood algorithm, combined with a boundary element modeling method. We first invert
InSAR and GNSS data spanning the whole event (propagation phase and eruption) to determine the
final geometry of the intrusion. Random search conducted in the inversion results in two best-fit
model families with similar data fits. Adding the same time-period GNSS dataset to the inversions
does not significantly modify the results. Even when weighting data to provide even contributions,
the fit is systematically better for descending than ascending interferograms, which might indicate
an eastward flank motion. Then, we invert the GNSS time series in order to derive information
on the propagation dynamics, validating our approach using a SAR image acquired during the
propagation phase. We show that the GNSS time series can only be used to correctly track the magma
propagation when the final intrusion geometry derived from InSAR and GNSS measurements is
used as an a priori. A new method to extract part of a mesh, based on the representation of meshes
as graphs, better explains the data and better accounts for the opening of the eruptive fissure than
a method based on the projection of a circular pressure sources. Finally, we demonstrate that the
temporal inversion of GNSS data strongly favors one family of models over an other for the final
intrusion, removing the ambiguity inherent in the inversion of InSAR data.
Domaines
VolcanologieOrigine | Fichiers éditeurs autorisés sur une archive ouverte |
---|
Loading...