Generic earthquakes properties from the SCARDEC database : what can Source Time Functions tell us about the seismic process ?

Until now, our understanding of the earthquake process from seismological observation comes mostly from detailed studies of large earthquakes, or of a given region's seismicity. However, a more consistent way to describe the physics of earthquakes is to analyse the global seismicity in a systematic approach. Since the early 90'ies, global seismological networks have provided good records of the moderate (Mw>=5.5) to large earthquakes. This has allowed the developpement of automated analytic methods of earthquakes source parameters and their systematic analysis (Bilek et al, 2004, Allmann and Shearer, 2009). A new automated method called SCARDEC proposed by Vallée, 2011, retrieves the Source Time Function (STF, or Moment rate function) of an earthquake through deconvolution of the teleseismic body waves, together with seismic moment, focal depth, and focal mechanism. This method leads to a catalog of ~2700 earthquakes with Mw>=5.8, that occured in the past 20 years, and takes advantage of the deconvolution approach : STFs include the data complexity. STF describes how seismic moment is released with time. With the earthquake's average STF, one can measure a lot of properties, such as the source duration, related to stress drop and rupture velocity variations, and the radiated energy by the source. Moreover, the Relative STF at a given station is slightly modified by the directivity of the rupture ; with a well distributed set of stations this property gives access to the rupture propagation and velocity. The aim of this work is to search how to extract reliable physical parameters describing the earthquake rupture from the average and relative STFs, and to analyse their distribution in a global setting or in light of specific earth properties. In a first part, I will talk about the global relative variations of stress drop and radiated energy, for thrust subduction earthquakes, in comparison with other tectonic environments in a shallow range of depth (z<=70km). Our results show that subduction earthquakes are systematically lower in stress drop and radiations, which confirms previous observations but with a larger dataset. Moreover, we observe clear differences among subduction earthquakes properties ; we could locally relate poorly impulsive and radiative earthquakes to low coupled zones. In a second, I will focus on rupture propagation and velocity. We developed an automated tool to measure the rupture propagation and velocity from Relative STFs within a unilateral rupture model assumption. We apply this method to the whole dataset of SCARDEC and identify the earthquakes that exhibit unilateral rupture with significant rupture velocity. We then try to build a consistent catalog of worldwide earthquake rupture propagation and velocities, of which analysis can provide insights on the dynamics of rupture and its relation with tectonics and fault properties.