Final Report Summary - AUGURY (Reconstructing Earth’s mantle convection)
The AUGURY team has pursued the goal of reconstructing the past motions of the Earth’s mantle and tectonics combining geological information and a new generation of 3D spherical models of mantle motions. To do so, the project aimed at pushing the boundaries of dynamic models of the mantle so they could be predictable enough to be used in a framework compared to weather forecast. The second goal was to develop this time of framework for that specific case. Such methodology would be designed to reconstruct the past and present-day structure of the Earth beneath our feet, not to predict the future off course.
Considering these two aspects, AUGURY was successful in producing a model that has the major characteristics of the observed Earth’s mantle and global tectonics: topography, gravity field, heat budget, tectonic speeds, supercontinent cycles, plate tectonic layout, dynamic features as subducting slabs and rising hot plumes, thermal heterogeneity, core-mantle boundary structures, multi-scale convective flows. The subsequent PhDs of Claire Mallard and Maëlis Arnould helped reach that point where a full calculation of a high resolution model with Earth’s like characteristics was performed. Throughout the process, we used the generations of models to analyse the underlying physics of plate tectonics, which had never been done before with global and fully dynamic strategies. We solved a growing issue at the time about conflicting observation and models of the topography of the Earth. We understood how large and small plates coexist on Earth, and how the driving mechanisms for plate tectonics evolve through time and space. Within that frame, we had to develop new tools, and among them, we collaborated with computer vision scientists to propose ADOPT, an innovative tool to automatically detect tectonic plates within convection models. Such tool opens a way to make structural geology analysis on numerical models.
The second aspect of the project was to develop methods to retrieve the internal structure of the mantle and past tectonics based on geological, geophysical observations and sometimes plate tectonic models. This attempt was a start from scratch, because there has not been any such attempt in the past. The PhD work of Marie Bocher established sequential data assimilation methods as proof of the concept. With surface information only, it is possible to reconstruct the missing information about the thermal structure of the mantle in simple models. This was key to drive the development of a methodology called 4D-var, which uses the adjoint of the code to realize the inversion combining the dynamic model and observations. The specificity of our development was to generate automatically the adjoint code, through automatic differentiation. Such process allows the co-evolution of the code and adjoint code, changing the observations to be used in the inversion. It is adaptable and sustainable. It took the whole project time to reach the automatic generation of the adjoint.
Therefore, at the end of the AUGURY project, the model and the framework are set to realize in the near future the reconstruction of mantle and tectonic evolution without using the theory of plate tectonics, hence releasing some unnecessary constraints of plate tectonics and giving access to the links between the depth and the surface of our planet.
To finish with, AGURY was a platform for a team mostly composed of women to work on the issue of gender inequalities in academia. With the support of ERC, we initiated the initiative didthisreallyhappen.net which became popular amongst scientists over the world.
Considering these two aspects, AUGURY was successful in producing a model that has the major characteristics of the observed Earth’s mantle and global tectonics: topography, gravity field, heat budget, tectonic speeds, supercontinent cycles, plate tectonic layout, dynamic features as subducting slabs and rising hot plumes, thermal heterogeneity, core-mantle boundary structures, multi-scale convective flows. The subsequent PhDs of Claire Mallard and Maëlis Arnould helped reach that point where a full calculation of a high resolution model with Earth’s like characteristics was performed. Throughout the process, we used the generations of models to analyse the underlying physics of plate tectonics, which had never been done before with global and fully dynamic strategies. We solved a growing issue at the time about conflicting observation and models of the topography of the Earth. We understood how large and small plates coexist on Earth, and how the driving mechanisms for plate tectonics evolve through time and space. Within that frame, we had to develop new tools, and among them, we collaborated with computer vision scientists to propose ADOPT, an innovative tool to automatically detect tectonic plates within convection models. Such tool opens a way to make structural geology analysis on numerical models.
The second aspect of the project was to develop methods to retrieve the internal structure of the mantle and past tectonics based on geological, geophysical observations and sometimes plate tectonic models. This attempt was a start from scratch, because there has not been any such attempt in the past. The PhD work of Marie Bocher established sequential data assimilation methods as proof of the concept. With surface information only, it is possible to reconstruct the missing information about the thermal structure of the mantle in simple models. This was key to drive the development of a methodology called 4D-var, which uses the adjoint of the code to realize the inversion combining the dynamic model and observations. The specificity of our development was to generate automatically the adjoint code, through automatic differentiation. Such process allows the co-evolution of the code and adjoint code, changing the observations to be used in the inversion. It is adaptable and sustainable. It took the whole project time to reach the automatic generation of the adjoint.
Therefore, at the end of the AUGURY project, the model and the framework are set to realize in the near future the reconstruction of mantle and tectonic evolution without using the theory of plate tectonics, hence releasing some unnecessary constraints of plate tectonics and giving access to the links between the depth and the surface of our planet.
To finish with, AGURY was a platform for a team mostly composed of women to work on the issue of gender inequalities in academia. With the support of ERC, we initiated the initiative didthisreallyhappen.net which became popular amongst scientists over the world.