Curtin Applied Geology Special Seminar

Tues 11th March

2 – 3 pm

Rm 312.222

Boris Kaus

Professor, Geophysics and Geodynamics, Johannes-Gutenberg Universität. Mainz

Insights in the physics of the Earth by combining models and observations

Abstract

Better understanding the physics of the Earth is remains one of the goals of the solid Earth sciences. Whereas geophysical methods give a fair idea of the structure of the present-day Earth, geological data indicate that most processes occur over millions of years and thus tell us something on how the Earth behaved on a much longer timescale. If we want to reconcile both types of data, we need models that describe how the lithosphere deforms and results in mountain belts such as the Himalayas. However, most geological tectonic models are just cartoon models drawn in Illustrator. They satisfy the geological (and often geophysical) data, but they do not necessarily tell us all that much about the underlying physics of the lithosphere and the mechanics of deformation. As a result, there are many competing geological models that can explain the same data sets.

A different approach is to use thermo-mechanical numerical models to simulate collision scenarios on the computer. Over the last decade, such models have become quite sophisticated and we can now take realistic rock rheologies into account that vary from brittle (or elastoplastic) under low temperatures to ductile (or viscous) at higher temperatures. I will give an overview of recent progress in this field using examples from my research group. Typically, geodynamic models are used in a forward manner, in which various theoretical scenarios are simulated as a function of changing parameters such as plate speed, thermal structure of the crust and lithosphere and rock rheology. The best fitting models are the ones that appear to be most consistent with the data. This does teach us something about how lithospheric collision could have occurred. Yet, since the number of model parameters is large and the models remain computationally intensive even in 2D, we cannot check every parameter combination. The result is that we often cannot isolate the key parameters that control the physics of lithospheric and crustal deformation in each case.

Here, I will discuss two examples to illustrate that it is possible to go a step further and to constrain the physics of the Earth on both crustal and lithospheric scale in a more direct manner.  The first example is related to crustal scale deformation in the Zagros Mountains, where space- and surface based geophysical and geological datasets put tight constraints on both the present-day structure and the geological evolution of the crust. Numerical models can fit the data, but only under very specific conditions. By analysing the numerical models in a systematic manner we are able to better understand the physics of crustal scale deformation on one hand and to simultaneously constrain the rheology of the crust on geological timescales. The second example concerns the structure and rheology of the present-day lithosphere in active collision belts. Whereas the large-scale structure of the lithosphere is often reasonably well known from mostly seismological and potential field methods, the rheology of the lithosphere remains poorly understood. Yet, it is the rheology that defines to a large extent how the lithosphere deforms and how it is coupled to the underlying mantle. Changing rheological parameters such as the effective viscosity of for example the mantle lithosphere can have a huge effect on the surface velocity or on mantle anisotropy. We have exploited this effect to develop a joint geodynamical-geophysical inversion technique that combines parallel 3D forward models of gravity and lithospheric-scale deformation with a Monte-Carlo inversion method. For a simple setup we can demonstrate mathematically that this joint approach results in a unique solution (as opposed to inverting for gravity alone which is a well-known non-unique problem). More realistic 3D cases are computationally significantly more challenging but they remain doable with today’s computers and results show that the method is capable of successfully determining the ‘best-fit’ parameters with uncertainty bounds. Combining dynamic forward models with geophysical observational constraints and inverse models is thus a very promising future research direction that will likely teach us more about the physics of the Earth.