Investigation of - and Post-Seismic Signals in GRACE Satellite Gravity Data Using Mcmc Approach

Earthquakes cause mass shifts and stress-dependent density changes. The corresponding gravity field variations can be measured by the GRACE satellites what was demonstrated for recent Andaman-Sumatra (2004), Maule, Chili (2010) and Tohoku-Oki (2011) earthquakes. In contrast to other data, satellite gravity regularly covers continental and oceanic areas, providing important information about the seismic cycle, especially in subduction zones. We make use of the characteristic temporal behavior and spatial scale of the earthquake signals, in order to separate them from other contributions to the gravity field. This is realized by wavelet transform and application of a time evolution model. The later consists of a step function followed by an exponential decay, representing co- and post-seismic variations. As the problem is nonlinear, parameters were fitted by determining their posterior distribution by means of a Markov-Chain Metropolis-Hasting sampling. As a result, we clearly separate the co- and post-seismic gravity variations for all three abovementioned earthquakes. We also apply the method to separate post-seismic signals from the Andaman-Sumatra and Nias-Sumatra events. The later occurred 3 months later to the south of the main megathrust event. These results are compared with other geophysical data and models. This allows to discriminate between candidate models for the co-seismic gravity variations using GPS data and seismology, and to better understand the physical processes involved in the post-seismic deformation.