Abstract
Most of the global seismicity occurs at these plate boundaries. However, large earthquakes of magnitude 6 or more occur in areas far from the plate boundaries. Quantifying this deformation and understanding the processes at its origin is a major scientific challenge but also a major societal issue because infrastructures and populations are not prepared for the occurrence of these large earthquakes. In the case of intraplate domains, the deformation is generally below 0.25 to 0.5 mm yr-1, which is close to the resolution limit of the geodetic measurement techniques, in particular of GNSS (Global Navigation Satellite System). Within this context, My PhD work focussed on characterizing and quantifying the current deformation in metropolitan France through new statistical analysis methods applied to GNSS data, velocities and strain rate fields. The first stage of my project was the calculation of the velocities of 1443 permanent GPS (Global Positioning System) stations between 1998 and the end of 2016. The second stage, using new statistical approaches, allowed quantifying the resolution of all the steps necessary to characterize the deformation. Firstly, the use of more than 25,000 synthetic data and statistical tools (regression trees) derived from machine learning techniques made it possible to quantify the precision of individual velocities resulting from the analysis of position time series. Secondly, I carried out a statistical cleaning of outliers on the slow (<0.5 mm yr-1 on average) and noisy raw GPS velocities. Then I used two independent statistical analyses (clustering and Gaussian smoothing methods) on the GPS velocity field to extract regional coherence. Owing to the extraction of regional coherence, the limit of interpretation of the GPS velocities can be lowered below the threshold of individual stations and reach, at 95% of confidence, 0.1-0.2 mm yr-1 for horizontal velocities and 0.35 × 10-9 yr-1 for strain rates. The third stage consisted in the study of the geodynamic implications associated with this characterization of the deformation.