- Density structure of the lithosphere and upper mantle derived from multiple data
- Numerical modeling and inversion theory for the geodynamics of the asthenosphere
- Gravity field of terrestrial planets and its applications on the planetary interior
- Machine learning in the inversion of geophysical data
Density structure of the lithosphere and upper mantle derived from multiple data
Earth’s lithosphere and upper mantle are the most outer layers of the planet and display much stronger heterogeneities comparing to the deeper layers. Lithospheric and upper mantle studies help us understanding active tectonics as well as the dynamic processes within the asthenosphere. Considering the limitations of individual geophysical disciplines, I am interested in the integration of multiple geophysical data sets, including seismic, gravity, thermal and chemical data. The goal is to recover the physical properties of the outer layers of the Earth. Subsequently, lithospheric tectonic processes and mantle geodynamics can be deduced from such results.
Numerical modeling and inversion theory for the geodynamics of the asthenosphere
Sitting in between the rigid lithosphere and the convecting mantle, the asthenosphere plays an important role in the transformation of different mineral phases, thermal conditions and chemical properties between the deep interior of Earth and its outer parts. Through forward and inverse approaches, we can construct self-consistent models regarding the thermal and chemical states of the asthenosphere. In addition, we may study the interaction between the asthenosphere and the lithosphere, and the geodynamic influences on the deformation of the lithosphere and other plate tectonic features.
Gravity field of terrestrial planets and its applications on the planetary interior
Given the development of our space technologies, the gravity field gives us one of the best chances to investigate other terrestrial planets in our solar system. Most of these planets have experienced a very different geological evolution comparing to Earth. As a result, distinct features are observed in their gravity fields. Combined with other available data, some of their interior physical properties could be obtained. Additionally, insights of the early stage of Earth might be given. I am particularly interested in those findings derived from their gravity data and the corresponding techniques, as well as in the study of the time varying gravity field.
Machine learning in the inversion of geophysical data
The inversion of the geophysical data is usually solved by an iterative process. For large scale problems, this process can be quite time consuming, and the inverted geophysical properties cannot be easily reviewed until the process terminates. Due to the existence of multiple factors that could alter the outcome, the final result is more likely achieved through trial-and-error adjustment of the initial conditions or other assumptions. Meanwhile, large number of intermediate products of the iteration process have been overlooked in our current studies. I wonder if, by introducing machine learning techniques into the process of inversion to study the properties of the intermediate products, whether a better understanding of the problem or superior analysis methods could be achieved, and how should we do that.