The structure and composition of the upper mantle from joint inversion derived parameter relationships

Geophysical methods such as seismology, magnetotellurics and gravity are key to reconstructing the structure of the upper mantle and inferring its composition. However, the relationship between composition and geophysical parameters, e.g. seismic velocity, resistivity or density, is complex and depends on other factors such as temperature, for example. This makes it difficult to untangle the various effects from inversions based on single parameters. Joint inversion establishes quantitative relationships between different geophysical parameters and thus provides additional information that can be interpreted in terms of composition and temperature. I use 3D joint inversion of surface wave, gravity and magnetotelluric data to construct integrated models in a data driven way. The relationship between the different quantities is recovered as part of the inversion through a Variation of Information based constraint. This constraint aims at establishing a one-to-one relationship between each parameter pair. Results from the western United States, Germany, Southern Africa and Australia show that this approach can retrieve highly detailed and strongly coupled results that can be interpreted, for example, in terms of hydration of the lithosphere. Comparison of results from different geologic domains indicates significantly different relationships depending on formation age. I will discuss how we can use data driven parameter relationships to infer the state of the lithosphere. In addition, I will outline the road towards robust quantitative inference based on these relationships.