Periodic Reporting for period 3 - RhEoVOLUTION (Micro-scale dependent, time- and space-evolving rheologies: the key for generating strain localization in the Earth)
Okres sprawozdawczy: 2023-11-01 do 2025-04-30
Strain localization is essential for the dynamics of the solid Earth. It is the rule rather than the exception in the lithosphere (the external layer of the solid Earth). Its first order expression is Plate Tectonics. However, after >50 years of the establishment of Plate Tectonics as the paradigm in Geodynamics, modelling spontaneous strain localization in ductile regime, which prevails in ~90% of the lithosphere, remains a challenge. As a consequence, we cannot predict: (1) when and where strain localization will develop, (2) the number and thickness of the shear zones accommodating this localized deformation, or (3) how the strain distribution will evolve through time.
Observations of ductile strain localization at various spatial scales in nature and experiments shows that heterogeneity in the mechanical behaviour is key for strain localization. This heterogeneity exists at all scales, but it is particularly well-developed at small scales, and it evolves in response to the mechanical fields. In the ERC RhEoVOLUTION, we posit that a poor representation of this heterogeneity and of its evolution with deformation is the locking point for generating strain localization in geodynamical models. The tools we design and develop in RhEoVOLUTION will bridge scales and unravel how heterogeneity and anisotropy in the mechanical behavior of rocks control strain localization in the Earth from the cm to the tens of km scale. To do so, we will:
1. describe the heterogeneity of mechanical behavior of rocks deforming by dislocation creep using stochastic parameterizations of the rheology;
2. constrain these parameterizations using experiments with in-situ follow-up of the microstructure and strain evolution and mesoscale models;
3. accelerate by orders of magnitude the calculation of the evolution of mechanical anisotropy during deformation using supervised machine-learning;
4. quantify feedbacks between the main processes producing strain localization by comparing the predictions of models parameterized to simulate these processes to observations in natural shear zones.
The aim of RhEoVOLUTION is to empower the geodynamics community with a robust framework and modelling tools for predicting self-consistent ductile strain localization and evolution of anisotropy over the large range of scales that characterize the deformation of the solid Earth, but also of ice caps and glaciers.
Observations of ductile strain localization at various spatial scales in nature and experiments shows that heterogeneity in the mechanical behaviour is key for strain localization. This heterogeneity exists at all scales, but it is particularly well-developed at small scales, and it evolves in response to the mechanical fields. In the ERC RhEoVOLUTION, we posit that a poor representation of this heterogeneity and of its evolution with deformation is the locking point for generating strain localization in geodynamical models. The tools we design and develop in RhEoVOLUTION will bridge scales and unravel how heterogeneity and anisotropy in the mechanical behavior of rocks control strain localization in the Earth from the cm to the tens of km scale. To do so, we will:
1. describe the heterogeneity of mechanical behavior of rocks deforming by dislocation creep using stochastic parameterizations of the rheology;
2. constrain these parameterizations using experiments with in-situ follow-up of the microstructure and strain evolution and mesoscale models;
3. accelerate by orders of magnitude the calculation of the evolution of mechanical anisotropy during deformation using supervised machine-learning;
4. quantify feedbacks between the main processes producing strain localization by comparing the predictions of models parameterized to simulate these processes to observations in natural shear zones.
The aim of RhEoVOLUTION is to empower the geodynamics community with a robust framework and modelling tools for predicting self-consistent ductile strain localization and evolution of anisotropy over the large range of scales that characterize the deformation of the solid Earth, but also of ice caps and glaciers.
Among the scientific achievements in the first 42 months of activity of the project, we can highlight:
• Development and implementation in a geodynamical modelling code of a time-evolving stochastic formulation of the rheology of upper mantle rocks, the first one in this domain, which can produce spontaneous strain localization (WP1).
• Based on these developments, establishment of the necessary conditions and a regime diagram for viscous strain localization and quantification of its effect on the mechanical behaviour, at both the local and system scales (WP1 - preprint in publication list)
• Data (experiments and nature observations) providing new constraints on the consequences of dynamic recrystallization on the strain distribution (WP2 and WP4 - articles 2 and 3 in publication list).
• Experimental data on ice documenting for the first time decoupling between strain localization and the microstructural record of deformation (WP2 – article 4 in publication list).
• Successful experiments with in situ follow up by electron microscopy of the deformation and dynamic recrystallization on Mg alloys (WP2 - article 9 in publication list).
• Experimental data with in situ follow up of the strain field showing a systematic variation of the bulk strength and strain localization evolution as a function of the mechanical viscoplastic anisotropy in Zn alloys (WP2).
• Development of an innovative full-field approach, based on an attractor formulation, to simulate the role of dynamic recrystallization on the evolution of texture and associated mechanical anisotropy in ice deforming by dislocation creep (WP2 - article 6 in publication list).
• New observational data highlighting strong feedbacks between fluids and strain localization in the lithospheric mantle (WP4 – article 7 in publication list and articles in preparation).
• Deep-learning approaches that predict the evolution of elastic anisotropy in response to deformationt in 2D (WP3)
• Deep-learning approaches that predict predict thermodynamically self‐consistent rock properties at arbitrary PTX conditions between 1–28 GPa and 773–2,273 K, and dry mantle compositions ranging from fertile (lherzolitic) to refractory (harzburgitic) end‐members (WP3- article 8 in publication list)
• Development and implementation in a geodynamical modelling code of a time-evolving stochastic formulation of the rheology of upper mantle rocks, the first one in this domain, which can produce spontaneous strain localization (WP1).
• Based on these developments, establishment of the necessary conditions and a regime diagram for viscous strain localization and quantification of its effect on the mechanical behaviour, at both the local and system scales (WP1 - preprint in publication list)
• Data (experiments and nature observations) providing new constraints on the consequences of dynamic recrystallization on the strain distribution (WP2 and WP4 - articles 2 and 3 in publication list).
• Experimental data on ice documenting for the first time decoupling between strain localization and the microstructural record of deformation (WP2 – article 4 in publication list).
• Successful experiments with in situ follow up by electron microscopy of the deformation and dynamic recrystallization on Mg alloys (WP2 - article 9 in publication list).
• Experimental data with in situ follow up of the strain field showing a systematic variation of the bulk strength and strain localization evolution as a function of the mechanical viscoplastic anisotropy in Zn alloys (WP2).
• Development of an innovative full-field approach, based on an attractor formulation, to simulate the role of dynamic recrystallization on the evolution of texture and associated mechanical anisotropy in ice deforming by dislocation creep (WP2 - article 6 in publication list).
• New observational data highlighting strong feedbacks between fluids and strain localization in the lithospheric mantle (WP4 – article 7 in publication list and articles in preparation).
• Deep-learning approaches that predict the evolution of elastic anisotropy in response to deformationt in 2D (WP3)
• Deep-learning approaches that predict predict thermodynamically self‐consistent rock properties at arbitrary PTX conditions between 1–28 GPa and 773–2,273 K, and dry mantle compositions ranging from fertile (lherzolitic) to refractory (harzburgitic) end‐members (WP3- article 8 in publication list)
We expect that the tools developed in RhEoVOLUTION will allow modeling the most evident expression of strain localization on Earth: Plate Tectonics and the memory effects in this process. By allowing systematic investigation of relations between changes in mechanical properties and strain distributions, the tools we are developping will provide explanations for localized deformation at different scales in Earth. They will allow predicting the evolution of shear zones in extensional and convergent plate margins, thereby enhancing our understanding of intraplate strain localization, rifting initiation, or the architecture of passive margins and mountain belts. The development of IA tools allowing for fast prediction of the evolution of seismic anisotropy during deformation by dislocation creep in the mantle, which is a byproduct of WP3, will allow to develop forward models constraining the interpretation of seismic anisotropy measurements in terms of flow patterns. All these tools amy also be extended for predicting ductile strain localization and mechanical anisotropy evolution in ice and metals with applications in glaciology and metallurgy.