Imaging coseismic rupture in far field by slip patches

S U M M A R Y Since the end of the 1970s, teleseismic waves are used routinely to infer the first-order characteristics of the event (location, depth, duration, focal mechanism), but finding the second-order kinematic parameters (spatial distribution of slip, rupture velocity and more basically the discrimination between the fault plane and the other nodal plane) of distant events remains a difficult task. Classically, these events are studied by two different methods; either they are seen as a succession of subevents, each of which is considered as a point source, or like in near field, as extended sources where the kinematic parameters are retrieved on a grid. The first approach is not physically satisfactory and can lead to erroneous interpretations of the rupture process (Ihmlé 1998) while the second one often gives highly non-unique results. In this work, we present a method that aims to find a simplified source model able to explain the main features of the teleseismic data. The basic idea is to consider the event as an ensemble of slip patches, breaking at different rupture velocities. In order to obtain a simple and stable model, we use a maximum of two slip patches and model only periods longer than a third/fourth of the earthquake global duration. We use three different types of data: P and SH body waves through direct modelling, and surface waves through an empirical Green function (EGF) technique. We follow a recent efficient approach to stabilize the EGF deconvolutions with physical constraints. The method is applied to the 1999 Izmit (Turkey) and 1995 Jalisco (Mexico) earthquakes and gives results consistent with previous studies of these events. Thanks to its simplicity, this method can be applied almost routinely after each large earthquake, and can give interesting insights on the physical properties of the rupture (size, slip, rupture velocity) as well as important clues on the risk associated with the event (for example tsunami risk).