Optimizing the information content available in geodetic data to jointly estimate co-seismic and early afterslip models

Description

When analyzing the rupture of a large earthquake, geodetic data are often critical. Yet, these data are generally characterized by either a good temporal (continuous GNSS) or a good spatial (InSAR and subpixel image correlation) resolution, but rarely both. As a consequence, many studies analyze the co-seismic rupture with data also including days of early post-seismic deformation, usually corresponding to afterslip. This approximation implies that the co-seismic slip models can be biased, and that the early afterslip process is disregarded. Here, we propose a new and simple approach to improve the use of the information contained in the data: we invert simultaneously for the co-seismic slip and the early afterslip with the condition that the sum of the two models remains compatible with data covering encompassing the two slip episodes (e.g. InSAR). Meanwhile, each individual model is also constrained by geodetic data covering its own specific time frame (e.g. continuous GNSS). We validate the benefits of our approach with a toy model and an application to the 2009 Mw6.3 l'Aquila earthquake. We find that if early afterslip deformation is acknowledged as co-seismic signal, co-seismic models may be biased for a third of their amplitude while longer term post-seismic models may overlook up to 300% of the total afterslip amplitude. This example illustrates how the proposed approach could improve our comprehension of the seismic cycle, of the fault frictional properties, and how the co-seismic rupture, afterslip and aftershocks relate to one another.

Abstract

International audience

Additional details