Back to Jean Paul Ampuero's Homepage

Research: earthquake dynamics

The research in our team combines theoretical, computational and seismological approaches to improve our understanding of the physics of earthquakes.
Our research topics include:

Earthquake nucleation

How do earthquakes start? This is one of the fundamental questions of earthquake dynamics, with implications on earthquake predictability, early warning systems and time-dependent seismic hazard.
I have been investigating the different styles of earthquake nucleation implied by usual fault friction laws. Recent results, in sustained collaboration with Allan Rubin (Princeton), include the characterization of quasi-static pulses during nucleation under rate-and-state friction with the slip evolution law.
Dynamic nucleation under linear slip-weakening friction
(Ampuero, Vilotte and Sanchez-Sesma, 2002)


Earthquake nucleation under rate-and-state friction with the "aging" state evolution law
(Rubin and Ampuero, 2005)


Earthquake nucleation under rate-and-state friction with the "slip" state evolution law
(Ampuero and Rubin, 2008)

Complexity of 3D earthquake dynamics

Seismic ground motions in the vicinity of active faults are strongly affected by the spatio-temporal details of the earthquake source. Unfortunately, near-field recordings that can nourish empirical approaches for ground motion prediction are notoriously scarce. An emerging complementary approach is based on direct simulations of earthquake dynamics scenarios, with the aim to integrate our best current knowledge about source complexity and earthquake physics.
In collaboration with Martin Mai (ETH Zurich) and former graduate student Johannes Ripperger (ETH Zurich, now in Oxford), I have been studying the effect of initial stress heterogeneities on the statistical properties of dynamic ruptures and the induced near-field ground motions. Post-doc Javier Ruiz has recently joined the team to work on this topic.


Rupture "percolates" through a heterogeneous initial stress
(Ampuero, Ripperger and Mai, 2006; Ripperger, Ampuero, Mai and Giardini, 2007)


(a) a set of stochastic initial stress realizations
(b) dynamic rupture fronts and final slip
(c) slip velocity at the center of the fault
(Ripperger, Mai and Ampuero, 2008)


Dynamic rupture on a heterogeneous initial stress
featuring dynamic triggering of secondary events ahead of the main rupture front
(courtesy of Johannes Ripperger)

Earthquake source seismology

Most of the information that can discriminate between different models of earthquake physics is contained in high frequency waves. Unfortunately these are hard to analyze, due to the opacity of the Earth crust. We are exploring new techniques to extract deterministic and statistical information from strong ground motion records, based on wavelet and array techniques.
The very frequent small magnitude earthquakes can also provide a wealth of information about earthquake physics. However they require specific processing methods. We are applying empirical Green's functions techniques to study very early aftershocks and to analyze the directivity of small magnitude events. This research started in collaboration with Allan Rubin (Princeton) and continues with post-doc Javier Ruiz.

Detection of composite microearthquake in Parkfield,
based on empirical Green's function deconvolution
(Ampuero and Rubin)

Earthquakes on bimaterial faults

Mature faults usually juxtapose rocks of different physical properties. Dynamic ruptures on such bimaterial faults feature interesting phenomena induced by the coupling between slip and normal stress changes: a tendency for macroscopic source asymmetry, preferred rupture direction, preferred aftershock triggering direction and asymmetric off-fault damage.
In collaboration with Allan Rubin (Princeton) and with Yehuda Ben-Zion (USC), I have been investigating the different styles of bimaterial ruptures predicted by various fault friction laws. Recent findings include the macroscopic asymmetry induced by the bimaterial effect on rupture pulses generated by fast velocity-weakening friction, and the competing effect of fault heterogeneities.

Dynamic rupture on a bimaterial fault with slip-weakening friction
(Rubin and Ampuero, 2007)


Dynamic rupture styles on a bimaterial fault with velocity-weakening friction
(Ampuero and Ben-Zion, 2008)

Earthquakes on non planar faults

Earthquakes are classically modeled as slip on faults that are planar or have a few large-scale geometrical features like steps, jogs, kinks and branches. However, the geometrical complexity of natural faults extends over a broad range of length scales and this might have an effect on the propagation of dynamic rupture fronts, and on high-frequency seismic radiation.
In collaboration with Raul Madariaga (ENS Paris) I have been investigating the dynamics of faults on rough faults. Post-doc Jean Elkhoury joined us to work on this topic.

Mode III dynamic rupture on a fault with multiple kinks
(Madariaga, Ampuero and Adda-Bedia, 2006)

Earthquake rupture with off-fault damage

Earthquakes are classically modeled as frictional instabilities on fault surfaces. However, natural faults are surrounded by damaged fault zones characterized by dense micro-fracturing and low elastic moduli. Dynamic rupture fronts concentrate stresses that can overcome the yield strength of the surrounding rock and trigger anelastic processes.
In collaboration with Yehuda Ben-Zion (USC) and Vladimir Lyakhovsky (GS Israel), I am exploring the interaction between frictional ruptures and a continuum representation of off-fault damage.


Pulse-like dynamic rupture with velocity-weakening friction
and off-fault continuum damage
(Ampuero, Ben-Zion and Lyakhovsky, SSA 2008)

Slow slip and non-volcanic tremor

Recently discovered recurrent slow slip events, low and very low frequency earthquakes and non-volcanic tremor, primarily in subduction zones, extend the known spectrum of earthquake phenomena.
In collaboraton with Allan Rubin (Princeton) and with Hugo Perfettini (IRD France) I have been studying the implications of rate-and-state friction on slow slip transients. Our simplest models require fine tuning to match the ensemble of observations, and we are now exploring additional physical ingredients: the effect of fault heteroegeneities and fluid-related processes. Post-doc Gregor Hillers recently joined the team to work on this topic.

Slow pulse propagating on a rate-and-state fault
with the "slip" state evolution law
(Ampuero and Rubin, 2008)


A periodic slow slip event on a creeping fault (velocity-strengthening rate-and-state)
containing one brittle asperity (velocity-weakening rate-and-state)
(Ampuero and Perfettini, 2008)


A periodic slow slip event on a creeping fault (velocity-strengthening rate-and-state)
containing one large and several small scale brittle asperities (velocity-weakening rate-and-state)
The slow slip event triggers fast slip on the secondary asperities. (Ampuero and Perfettini, 2008)

Numerical methods for earthquake dynamics

The emergence of physics-based ground motion prediction approaches, the current trend towards dynamically-constrained source inversion and our increasing interest on stochastic aspects of earthquake rupture are imposing high demands on computational methods for large scale earthquake dynamics simulations. Modern methods should reach high performance and high accuracy, and be able to accommodate geometrical complexity, heterogeneous media, non-linear off-fault rheologies and multiple physics in the fault zone.
Owing to continued collaboration with Jean-Pierre Vilotte (IPG Paris) and graduate student Yoshihiro Kaneko, the application of the Spectral Element Method (SEM) to earthquake dynamics has reached maturity. The SEM code SEM2DPACK has been applied to rupture on non-planar faults and rupture with off-fault damage and plasticity. Fault dynamics features are now implemented in the SPECFEM3D code.
Recent methodological contributions to SEM include the application of high-order symplectic time integration schemes to improve the accuracy of numerical wave propagation over long distances, developed in collaboration with post-doc Tarje Nissen-Meyer.
In collaboration with Josep De La Puente (LMU Munich) we have recently introduced fault dynamics in the ADER-DG (Discontinuous Galerkin) method, a high-order method that can take advantage of automated unstructured meshing tools to overcome a major bottleneck of SEM simulations.

Horizontal slip rate in the SCEC TPV5 problem, solved by SPECFEM3D

Dynamic rupture on a rate-and-state fault with a shallow velocity-strengthening layer
(Kaneko, Lapusta and Ampuero (2008)

Seismic hazard along the Peruvian subduction zone

Peru is located along the Nazca-South America subduction zone, one of the most seismically active regions in the world. Earthquakes are the first cause of human and economic loss by natural disasters in Peru. Lima, the capital city, lies closer to subduction seismicity than any other megacity of the western hemisphere.
Efforts are being coordinated with institutions in Peru to create a new generation of earthquake risk management tools. Our team is contributing with a portable seismic network (15 Guralp CMG-40T sensors with Nanometrics Taurus recorders), operated in collaboration with the Instituto Geofisico del Peru as an "Earthquake Task Force Unit" for seismological studies and earthquake engineering applications, including seismic tomography, seismicity monitoring, aftershock recording, and small array microtremor methods for site effect quantification.

Current seismic experiment in the North of Peru.
Red: our portable stations. Green: IGP's permanent stations

Back to Jean Paul Ampuero's Homepage