Probabilistic modeling of earthquakes

Friday, 18 December 2015: 13:55
305 (Moscone South)
Zacharie Duputel1, Romain Jolivet2, Junle Jiang3, Mark Simons3, Luis A Rivera1, Jean-Paul Ampuero3, Baptiste Gombert1 and Sarah E. Minson4, (1)Institut de Physique du Globe de Strasbourg, Strasbourg, France, (2)Ecole Normale Supérieure Paris, Department of Geosciences, Paris, France, (3)California Institute of Technology, Pasadena, CA, United States, (4)U.S. Geological Survey, Earthquake Science Center, Menlo Park, CA, United States
By exploiting increasing amounts of geophysical data we are able to produce increasingly sophisticated fault slip models. Such detailed models, while they are essential ingredients towards better understanding fault mechanical behavior, can only inform us in a meaningful way if we can assign uncertainties to the inferred slip parameters. This talk will present our recent efforts to infer fault slip models with realistic error estimates. Bayesian analysis is a useful tool for this purpose as it handles uncertainty in a natural way. One of the biggest obstacles to significant progress in observational earthquake source modeling arises from imperfect predictions of geodetic and seismic data due to uncertainties in the material parameters and fault geometries used in our forward models - the impact of which are generally overlooked. We recently developed physically based statistics for the model prediction error and showed how to account for inaccuracies in the Earth model elastic parameters. We will present applications of this formalism to recent large earthquakes such as the 2014 Pisagua earthquake. We will also discuss novel approaches to integrate the large amount of information available from GPS, InSAR, tide-gauge, tsunami and seismic data.