A classic conceptual model of the earthquake cycle (Ohnaka, 2013) states that, after the interseismic loading phase, the mainshock is preceded by a nucleation phase, where the fault gradually unlocks over a growing slip patch which, when large enough, allows reaching suitable conditions for the propagation phase. During the latter, the fault sees high slip rates behind the propagating rupture front, radiating damaging seismic waves.

The observation that no two earthquakes have similar nucleation phases continues to puzzle earth scientists to the day. Indeed, recent seismological observations highlighted that both aseismic silent slip and/or foreshock sequences can precede large earthquake ruptures (Tohoku-Oki, 2011, Mw 9.0 (Kato et al., 2012); Iquique, 2014, Mw 8.1 (Ruiz et al, 2014; Socquet et al., 2017); Illapel, 2015, Mw 8.3 (Huang and Meng, 2018); Nicoya, 2012, Mw 7.6 (Voss et al., 2018)). However, the evolution of such precursory markers during earthquake nucleation remains poorly understood.

Regarding earthquake propagation, the discovery of lubrication mechanisms in the laboratory has shed light on many physical mechanisms that can enhance fault weakening and result in large co-seismic slips which could not be physically explained before. Two of the most well-documented mechanisms are flash heating and thermal pressurization of pore fluids (Rice, 2006). These mechanisms are activated when the fault heats up as co-seismic slip at high velocities increases. Nevertheless, flash heating has, until now, been considered mostly under dry conditions while thermal pressurization has been considered only to affect fault mechanics through the effective stress law.

Here, we report for the first time, experimental results regarding both the nucleation and propagation phases of laboratory earthquakes (stick slip events) conducted on Westerly Granite saw-cut samples and under stress conditions representative of the upper continental crust, i.e confining pressures from 50 to 95 MPa; fluid pressures (water) ranging from 0 to 45 MPa.

During nucleation, slight changes in fluid pressure (at a given effective confining pressure of 70 MPa) drastically changed the temporal evolution of laboratory earthquake precursors. In dry conditions, precursory slip evolved exponentially up to the main instability and was escorted by an exponential increase of acoustic emissions. With pressurized fluids (pf=1 MPa), precursory slip evolved first exponentially then switched to a power law of time. There, precursory slip remained silent, and this silent precursory phase was observed independently of the fluid pressure level. The temporal evolution of precursory fault slip and seismicity are controlled by the fault's environment, limiting its prognostic value. Nevertheless, we show that, independently of the fault conditions, the total precursory moment release scales with the co-seismic moment of the main instability. The relation is demonstrated using a newly derived semi- empirical scaling law and seems to hold in several natural earthquakes (Acosta et al., 2019).

During earthquake propagation, it is observed that, at a given effective confining pressure (σ'3=70 MPa) co-seismic slips are small (~30-100 micron) at high fluid pressure (pf=25 MPa) while they are very large (~160-300 micron) under both dry (pf= 0 MPa) and low fluid pressure conditions (pf= 1 MPa). Microstructural analysis on post-mortem surfaces revealed that the Flash Heating mechanism operated in both dry (pf= 0 MPa) and low fluid pressure conditions (pf= 1 MPa) but no evidence of melted asperities was found at high fluid pressure (pf=25 MPa). Through an analytical model of flash heating in presence of pressurized fluids we show that this mechanism can operate at low fluid pressure because water can vaporize when enough heat is produced (at slip velocities >10 cm.s-1). In opposition, at pf=25 MPa, water's liquid-to supercritical transition acts as an extremely efficient heat buffer due to large increases in its specific heat at the transition. At those high fluid pressures, it is thermal pressurization the main driving mechanism during propagation (Acosta et al., 2018).

In summary, we showed that fluid pressure can drastically change the temporal evolution of precursors during nucleation but the amount of moment released prior to an earthquake is directly related to its magnitude, increasing therefore the detectability of large earthquakes. During propagation, the efficiency of Flash Heating and Thermal Pressurization are controlled by fluid thermodynamics, and therefore physics based models of earthquake propagation should not neglect fluid pressure through the effective stress law.

References
Acosta, M., Passelègue, F.X., Schubnel, A. and Violay, M., 2018. Dynamic weakening during earthquakes controlled by fluid thermodynamics. Nature communications, 9(1), pp.1-9.
Acosta, M., Passelègue, F.X., Schubnel, A., Madariaga, R. and Violay, M., 2019. Can precursory moment release scale with earthquake magnitude? A view from the laboratory. Geophysical Research Letters, 46(22), pp.12927-12937.
Huang, H. and Meng, L., Slow Unlocking Processes Preceding the 2015 Mw 8.4 Illapel, Chile, Earthquake. Geophys. Res. Lett, 45(9), 3914-3922. (2018)
Kato, A., et al., Propagation of Slow Slip Leading Up to the 2011 Mw 9.0 Tohoku-Oki Earthquake. Science, 335(6069), 705–708. (2012) https://doi.org/10.1126/science.1215141
Ohnaka, M., 2013. The physics of rock failure and earthquakes. Cambridge University Press.
Rice, J.R., 2006. Heating and weakening of faults during earthquake slip. Journal of Geophysical Research: Solid Earth, 111(B5).
Socquet, A., et al., An 8-month slow slip event triggers progressive nucleation of the 2014 Chile megathrust. Geophys. Res. Lett., 44(9), 4046–4053. (2017) https://doi.org/10.1002/2017GL073023
Voss, N., et al., Do slow slip events trigger large and great megathrust earthquakes? Science advances, 4(10), (2018)

Earthquakes are difficult to predict with certainty, but progress in forecasting their likelihood using probabilistic models based on stress changes has been made. However, challenges remain in understanding how earthquakes start and the initial conditions of faults. Here, we analyzed the Groningen gas field as a natural laboratory, where production is seasonal and seismic activity began 34 years after gas production started. By studying how the earthquakes respond to rapid changes in stress, we could better understand how they start and develop models to forecast their temporal occurrence. By considering factors like the initial strength of the faults, the finite duration of earthquake initiation, and seasonal variations in gas production we could accurately match the observed seismic activity. We introduced a new measure to evaluate how well the models captured the dampened strength and timing of seismic activity in response to short-term stress changes (such as seasonal variations), which helped refine the model's parameters. 

M. Acosta, J-P. Avouac, J.D. Smith, † K. Sirorattanakul, † H. Kaveh, S.J. Bourne; (2023 ) Earthquake nucleation characteristics revealed by seismicity response to seasonal stress variations induced by gas production at Groningen. Geophysical Research Letters.

We've proposed a method to quantify the uncertainty associated to the forecasts which combines epistemic and aleatoric variability. Using the method, we observe that an early deployment of a good seismic network, combined with the use of various timescales of seismicity response allows very early calibration of the model'sH parameters.

Kaveh, P. Batlle, M. Acosta, P. Kulkarni, J-P. Avouac; (2023) Induced Seismicity forecasting with Uncertainty Quantification: Application to the Groningen Gas Field. Seismological Research Letters.

As part of the main goal of understanding the two-way coupling between mechanical rock deformation and fluid transport in EGS reservoirs,  I developped complex experimental protocols to study fluid transport in deep geothermal reservoirs. The experimental observations, coupled with analytical and numerical models, and with natural observations, allowed for the unravelling of some fundamental processes at play during EGS stimulations. 

I showed that, in anisotropic rocks, the foliation orientation towards the stress field controls its mechanical and hydraulic transport properties. With ongoing deformation, anisotropic micromechanical models accurately predict the onset of damage, ultimate strength, porosity evolution, and fracture structure towards foliation orientation. Permeability is controlled by foliation orientation, fracture structure, and the applied stress. This shows that the full permeability tensor needs to be used with care on failed anisotropic rocks (Acosta and Violay, 2020).

In rock fractures with customized roughness, increasing the normal load decreases fracture transmissivity and thus fluid transport capacity. The transmissivity decrease is controlled by the contact geometry, as predicted by numerical models. Further, reversible shear loading and irreversible shear displacements (up to 1mmoffset) have little effect on transmissivity. Transmissivity evolution with shear displacement can be predicted by simplemodels only at low normal stress, where wear is not too prominent. These results question the concept of hydro-shear stimulations with small fault displacements (Acosta, Maye, and Violay, 2020).

References: 

M.Acosta, R.Maye, M.Violay; (2022)

Hydraulic transport through calcite bearing faults with customized roughness: Revisiting Hydro-shear stimulations.

Proceedings of the Stanford Geothermal Workshop 

Acosta, M., Maye, R., & Violay, M. (2020). Hydraulic transport through calcite bearing faults with customized roughness: effects of normal and shear loading. Journal of Geophysical Research: Solid Earth, 125(8), e2020JB019767.

Acosta, M., & Violay, M. (2020). Mechanical and hydraulic transport properties of transverse-isotropic Gneiss deformed under deep reservoir stress and pressure conditions. International Journal of Rock Mechanics and Mining Sciences, 130, 104235.

During my time at EPFL, I developped, installed and calibrated a novel HP-HT rock deformation apparatus called TARGET (TriAxial Rig for GEoThermal energy applications ). It's design is based on the Paterson apparatus but this device can reach higher pressures, deform larger samples across the Brittle and Ductile domains, and measure physical properties of the rock samples during deformation.  The apparatus can reach a maximum confining pressure of 400 MPa and a temperature up to 800 °C. Its main application is to deform rocks at high P–T and to measure associated hydraulic, elastic (Vp, Vs) and mechanical properties. TARGET can also be used for hot isostatic pressing (HIP), materials synthesis or other applications requiring a high-pressure, high-temperature environment, such as experimental petrology. However, its main application is to provide an environment for mechanical testing and associated studies at  HP–HT. 

References:

G. Meyer, M. Acosta, H. Leclère, L. Morier, M. Teuscher, G. Garrison, A. Schubnel, and M. Violay; (2023)

A new high-pressure high-temperature deformation apparatus to study the brittle-to-ductile transition in rocks. Rev. Sci. Instr. - Link 

Understanding the preparation phase of moderate to large earthquakes is of crucial importance to reduce seixsmic hazard. 

Using borehole strain-meter data within the 100 km region surrounding the mainshock epicenter, we identify significant transient deformation starting in February 2014. We use independent component analysis on daily GPS data to isolate the signal corresponding to the transient surface deformation identified on the BSM data. We combine these observations with a high resolution seismicity catalog generated through machine-learning phase detection algorithms (Moussavi et al., 2018), non-linear earthquake location (Lomax et al., 2013), and cross-correlation based earthquake relocation (Trugman and Shearer, 2017). 

The high resolution seismicity catalog allows a clearer understanding of the faults active during the deformation transient, their spatio-temporal evolution, and a better constraint on the fault's geometries. Finally, using the combined high-resolution dataset, we invert for the precursory slip in a multi-fault inversion procedure and resolve the aseismic deformation transient identified on the BSM data. This study highlights the uses of high-resolution seismic and geodetic data for a better understanding of earthquake initiation and deformation transients. 

Among other projects, I've actively worked on:

• The evolution of acoustic wave velocity and amplitudes during co-seismic slip. We observe that the acoustic wave velocities are fully controlled by the stress state of the rock, thus the co-seismic drop in wave velocities is only due to the co-seismic stress drop in a reversible manner. However, the acoustic attenuation (amplitude of the transmitted wave) is sensitive to on-fault processes such as damage and the generation of wear. See top panel in the figure.
  
  F. Paglialunga, F. Passelègue, M. Acosta, M. Violay; (2021) Origin of the co-seismic variations of elastic properties in the crust: insight from the laboratory. Geophysical Research Letters

• The initiation and evolution of a seismic swarm in California using geodetic and seismic observations. See bottom left panel in the figure. We showed by the use of high frequency GPS, InSAR data and an enhanced seismicity catalog that the swarm initiates as aftershocks of an aseismic slip transient that ocurred on two sub-perpendicular faults near the Westmorland area. The swarm was then sustained by fluids propagating on en-echelon faults.  

K. Sirorattanakul, Z.E. Ross, M. Khoshmanesh, E. Cochran, M. Acosta, J-P. Avouac; (2022) The 2020 Westmorland, California earthquake swarm as aftershocks of a slow slip event sustained by fluid flow. Journal of Geophysical Research: Solid Earth.

• The evolution of roughness and wear processes during laboratory fault slip. We've shown that young fault's roughness is strongly dependent on the fault morphology and that more wear is generated when the direction of the shear stress vector is perpendicular to the major morphological features of the faults. The faults have a strong tendency to go toward self affinity as wear processes develop. Manuscript in preparation.