I would like to ask whether there is a way to set zero conductivity for the lithosphere in the Steinberger model. With the evolution of our model, the lithosphere becomes colder and thicker. We would like to suppress this effect by setting the lithospheric thermal conductivity to zero. However, it seems that in Steinberger model, we can either use a constant thermal conductivity for the whole domain or a p-T-dependent conductivity. Any ideas about this will be highly appreciated. At the moment, our lithosphere is a low- temperature, thus high-viscosity layer with different thickness at different locations in our 3D model. It’s not a compositional field yet.
Thanks for posting on the forum! If you want a variable thermal conductivity (i.e. zero in the lithosphere and non-zero else where) then I think the best way to achieve something of this effect in the steinberger material model would be to use the p-T-dependent conductivity. In the p-T-dependent model you can choose the depth of transition between reference conductivity values, and you can set the variables which control the p-T dependence on the conductivity such that there is no p-T dependence. All the parameters can be found in this part of the code.
The only issue that I see with this approach is that this won’t account for the variable thickness of your lithosphere. If this ends up being an issue then I think defining a compositional field that represent the variable thickness lithosphere will be the best path forward. One caveat though is that it doesn’t look like the constant thermal conductivity model currently supports different values for different compositional fields, so this feature would have to be added to the code.
As @daniel.douglas noted, you could hard code in a p-T transition, but it may be challenging to do this in a way that allows the lithospheric thickness to evolve dynamically.
I imagine the most practical long-term approach to achieve your goals would be to multiple compositional fields with distinct properties. If you plan on moving in this direction, it may be worth looking at the entropy model material model, which I believe now supports handling multiple compositions. A recent paper using this material model (but not with multiple compositions) can be found here.
I’d second the recommendation to go with multiple compositional fields. Especially in lithospheric modeling it easily becomes a hefty task to construct highly heterogenous layers just within a single field. Also I do have a need to use Steinberger material model, so I could read e.g. density from lookup tables but my models also require highly distinctive rheological parameters, conductivity etc. for different compositional fields. Hence the approach I’ve taken uses compositing material model, which is able to combine several material models. For example, in my case:
subsection Material model
set Model name = compositing
set Material averaging = harmonic average only viscosity
subsection Compositing
set Compressibility = visco plastic
set Density = Steinberger
set Entropy derivative pressure = Steinberger
set Entropy derivative temperature = Steinberger
set Reaction terms = visco plastic
set Specific heat = visco plastic
set Thermal conductivity = visco plastic
set Thermal expansion coefficient = Steinberger
set Viscosity = visco plastic
end
subsection Steinberger model
...
end
subsection Visco Plastic
...
# Thermal properties
set Define thermal conductivities = false
set Heat capacities = ...
set Thermal diffusivities = ...
...
end
end
It seems that in order to take into account the variable thickness of the lithosphere, we would need to either modify the thermal conductivity source code or switch to a different material model (entropy model or using compositing material model). We will look into these options!