As our seismic acquisition system becomes denser and records broad-band data, reconstruction of parameters related to the fluid circulation in the crust becomes feasible. Biphasic structures with double porosity (one related to the porosity of the medium and the other one related to the fractures of this medium) can be tackled for the reconstruction of parameters as porosity and/or saturation, although one must be aware of the non-unicity of such reconstruction.

Hugh numbers of data and parameters should be considered and optimization techniques should be frugal when computing new images: the adjoint method is one of the paths to do so. The goal is the extraction of the most pertinent information from the whole seismogram. These approaches are strongly related to the temporal inversion (reverse time strategy or backprojection operator) of recorded signals inside the model. at each location, they are cross-correlated with the incident signal from the selected source in order to locate and estimate anomalies of physical parameters.

We shall apply these tools for crustal structures as well as for faulted structures of earthquakes without forgetting global structures. We may envision that fluid migration is the key for the variability of response times to mechanical transferts either for an earthquake or for an extraction/injection of fluids in the crust.

Modelling and interpreting seismic signals for studying both seismic sources (essentially occurring inside the crust) as well as for imaging crustal structures has been at the center of my research activity.

The dynamic rupture is a highly non-linear problem as the rupture propagation depends on criteria based on the singular stress field nearby the fault. My current interest is related to the geometrical complexity of the fault surface which cannot be restricted into a plane. The energy partitioning includes radiation and dissipation effects (fluids, brine zone).

Numerical techniques which have been improved recently allows us to consider these local effects on powerful computers while modelling seismic wave propagation. These modelling techniques could be integrated as well in the seismic imaging procedures through reverse-sensitivity analysis. They have to be tuned for fast performance either by considering them in the frequency domain or by speeding them in the time domain.

The present target is moving from 3D acoustic seismic imaging to 3D elastic seismic imaging : this often used technique for oil exploration needs to be improved for deeper targers and large-scale structures with recent deployments of seismic arrays at the lithospheric scale. The trend of the high resolution seismic imaging is the complete interpretation of the whole seismic signal.