Restart and compositional initialization issue

ASPECT
1

Hi Everyone,

I’m working on a global mantle convection model (spherical) where I want to include continents, but I’m intentionally not introducing them at the beginning of the simulation. The reason is that the initial condition does not produce a realistic convecting flow, and during this early transient phase, any pre-existing continents tend to deform significantly.

In my workflow, I typically:

  • Run the initial part of the model with constant viscosity to allow thermal boundary layers (TBLs) to develop,

  • Then switch to a viscoplastic rheology to obtain plate-like behavior at the surface.

I would like the continents (which have different viscosity and density) to be introduced only after this initial constant-viscosity stage, so that they remain relatively intact once plate-like motion begins.

What I have tried so far:

  • I first ran the model with constant viscosity and without actively using continents. However, a continent compositional field was present but effectively inactive (initialized to zero).

  • I then restarted from a checkpoint and attempted to introduce continents using an initial compositional field.

Although the model restarts without errors, the new compositional field is not being assigned—the model continues as if no continents were added. I also tried initializing temperature field from the first step while starting with viscoplastic rheology, but this did not produce the expected plate-like behavior.

My questions:

  1. Is it possible to introduce a new compositional field after restarting from a checkpoint in ASPECT?

  2. Is there a recommended workflow to introduce features like continents only after an initial transient phase?

  3. Is it possible to define an initial composition that depends on time, so that continents appear at a designated point during the run? If such a time-dependent approach is possible, could someone point me toward any appropriate method or plugin to implement this?

Any suggestions or guidance would be greatly appreciated!

Thanks in advance,
Arnab.

2

Hey Arnab,

Thanks for posting this on the forum and what can/cannot be done after restarting a checkpoint is a great question. Things like the Initial composition model, Initial temperature model etc., will not be called after restarting from a checkpoint. Therefore, you need to make sure that you include the continent composition (i.e. nonzero values within the domain) in the initial conditions and then the question would be how to make sure that this composition does not influence the model evolution until you want it to. You mention that you want the continents to remain intact, and so I’m guessing that having this composition just be a passive field is undesirable because this field would get deformed and brought down into the mantle while the thermal boundary layer develops.

What you can change is the compositional field method. You could initialize a continent compositional field and then make it a static field until the thermal boundary layers form, and then change the field method to be field after the checkpoint. Assigning the field to be static will prevent the field from moving during the formation of the thermal boundary layers.

Hopefully this helps, don’t hesitate to let us know if this does/does not do what you are looking to achieve!

Cheers,

Daniel

3

@arnie94 - A quick note to say that I agree with the suggestion from @daniel.douglas to initially set the continents (and other fields representing lithologies) to static, and then switching them to type field after the initial transient phase. Please let us know if you need any help implementing this approach.