The use of computational methods within the geological community has increased
considerably over the last decade, partly thanks to advances in hardware and software. At the
same time, there has been a tendency to increase the complexity of the models by adding ever
more processes that are thought to be relevant (fluid migration, melting, metamorphic phase
transitions, etc.). Whereas it can be argued that this makes the models more ‘realistic’, such
complexity results in a situation where understanding the outcome of simulations
becomes a difficult task. Yet, if the models are used in a systematic manner and are
combined with dimensional analysis and (semi)-analytical solutions, one can obtain a
thorough insight into the underlying physics of the processes. Scaling laws can be
derived that allow predicting the outcome of numerical simulations before performing
them. Once this is achieved, one can compare models with natural observations,
and –in some cases– extract material parameters from natural observations that
are otherwise difficult or impossible to obtain. Here, I will show two examples
to illustrate how numerical models can be used to obtain insights into geological
processes:
(1) Understanding the rheology of crystal-bearing magma is important for modelling
volcano dynamics. Whereas laboratory experiments on pure melts show that the rheology of
melt is Newtonian, more recent experiments show that melt with embedded crystals is
strongly non-Newtonian and becomes weaker with faster deformation rates. Understanding
the underlying causes for this has been tricky, partly because we typically only see the
end-results of the laboratory experiments. We therefore performed numerical simulations in
which digitized images of laboratory samples were used as model input, and the relative
importance of mechanisms such as shear heating and finite strain was tested. Results show
that brittle fracturing of crystals during the experiments is the most likely cause for the
non-Newtonian rheology. These results can thus be used to design new laboratory
experiments.
(2) The Zagros Mountains are a prime example of a folding-dominated fold-and-thrust
belt, in which the folds have a spacing of ~15 km. Thanks to its location in a desert and the
importance of the region for hydrocarbon exploration, the surface and subsurface geology are
very well constrained. It should thus be straightforward to apply models to the
data and constrain the rheology of the crust in this manner. Yet, it turns out that
this is not the case: if the basal salt layer is the only detachment layer, numerical
simulations are dominated by faulting rather than by folding, irrespective of model
parameters. Yet, in the Zagros there are up to 3 additional weak detachment layers
within the crust, and if these are taken into account, the simulations are instead
folding-dominated. We developed a technique to predict the spacing of folds in the presence
of a brittle overburden and show that the effective friction angle of the crust and
the viscosity of the weak layers are the two key parameters controlling this. Since
we have independent constraints on salt viscosity, a friction angle no more than
10 degree is required to fit constraints. This suggests that the crust in this region
was quite weak on geological timescales, likely the result of large fluid pressures. |