Inversion conjointe des donnees electriques et de radar en forage

We present two joint structural inversion algorithm for cross-hole electrical resistance tomography (ERT) and cross-hole radar travel time tomography (RTT).;In the second algorithm, we propose to combine a zonal cooperative inversion (ZCI) scheme with a hierarchical Bayesian approach, in order to invert cooperatively cross-hole ERT data and cross-hole radar travel time data. The basic idea of ZCI is to use cooperatively cluster analysis and separate inversion algorithm. For each iteration cluster analysis of separate inversion results is used to construct models that contain the parameter characteristics of dominant subsurface structures. These constructed models are then used as starting model in the next iteration of separate inversion. The resulting models are then biased to starting models which are a function of the number of clusters. To overcome this problem, we formulate the inverse problem within a hierarchical Bayesian framework where the hierarchical prior distribution is based on the a priori models constructed from cluster analysis. The advantage of such a formulation is to avoid undesirable bias towards the starting model and leads to significantly improved spatial resolution for consistent prior information. To validate our methodology and its implementation, a few experiments using three simple synthetic models are performed using different number of clusters. The results show that our cooperative inversion approach provides effective means to constrain resistivity and radar velocity models without biasing the solution. Hence, the choice of number of cluster to create the a priori model is not very important. However, this algorithm seems to be sensitive de noise level.;Finally, the proposed algorithms were applied for Sherwood sandstone vadoze zone characterisation to evaluate their performance. The results were compared to joint inversion with cross-gradient constraint algorithm. The first algorithm presents the best results that are in accordance with hydrogeological information and geophysical logs.;In addition to joint inversion algorithms, we introduce a new traveltime picking schemes developed specifically for crosshole ground-penetrating radar (GPR) applications. The approach is based on the Akaike information criterion (AIC) and continuous wavelet transform (CWT). It is not tied to the restrictive criterion of waveform similarity that underlies crosscorrelation approaches, which is not guaranteed for traces sorted in common ray-angle gathers. It has the advantage of being automated fully. Performances of our algorithm and a few crosscorrelation approaches are tested with synthetic and real data. The results show that the AIC-CWT approach is more versatile and performs well on all data sets. Only with data showing low signal-to-noise ratios is the AIC-CWT superseded by the modified crosscorrelation picker.;The first algorithm proceeds by combining the exchange of structural information and a regularization method that consists of imposing an L1-norm penalty in the wavelet domain. The minimization of the L1-norm penalty is carried out using an iterative soft-thresholding algorithm. The thresholds are estimated by maximizing a structural similarity criterion, which is a function of the two (ERT and RTT) inverted models. Besides, the regularization in the wavelet basis allows for the possibility of sharp discontinuities superimposed on a smoothly varying background. Hence the structural information is extracted from each model using a Canny edge detector. The detected edge serves to construct a weighting matrix that is used to alter the smoothness matrix constraint. To validate our methodology and its implementation, three synthetic models were created. Experiments demonstrate that the proposed approach improves the spatial resolution and quantitative estimation of physical parameters. In addition, it seems to be more robust in high noise level condition.