Hi All,
In my model, the final position of the model top boundary is known (see sketch below). How can I ask ASPECT to move (map) this boundary from its initial position to this pre-defined position?
I would appreciate your help,
Mahdi
Hi @mahdihdr - Welcome to the community and thank you for posting to the forum.
I think you can accomplish this by prescribing a surface velocity boundary condition that produces finite deformation resulting in the observed topography.
However, that may not be exactly what you are hoping to achieve in your model?
If the goal is to start with initial topography (final geometry) on the top boundary, you can achieve this with the Initial topography feature within the Geometry model plugin.
Cheers,
John
Hi John,
Thank you for your reply. I exactly want to do what you said, that is, to prescribe a velocity boundary condition that maps a known initial topography (surface in my 3D model) to a known final topography (surface in my 3D model).
The problem is I do not know how to calculate this velocity function V(x,y,t) from the known initial and final topographies. These topographies do not have an analytical equation (at least the final topography). Do you know how I can calculate the velocity or any tool that can do this?
Thank you,
Mahdi
Hi Mahdi,
Do you know how I can calculate the velocity or any tool that can do this?
I have not done this type of workflow before, but I imagine there must be a number of available tools that can help. I think a key question to consider if whether you can actually uniquely determine the deformation history a priori. In other words, the final deformation state could result from multiple combinations of horizontal and vertical displacements through time.
Assuming heat conduction is not a key process, one thing you could try is starting with the deformed stated and reversing the sign of gravity.
Alternatively, you could run a series of forward models with the conditions that you think lead to the deformed state, and assess the fit. This is a case where you could reduce the total parameter space with a machine learning approach (the work by Boris Kaus’ group immediately comes to mind).
Cheers,
John
@mahdihdr For modest deformations, the tool you are looking for is called “transfinite interpolation” where you only need to know a function that describes the deformation of the top surface, and it gives you a function that describes the deformation in the entire domain. The velocity is then simply the deformation divided by some time scale (e.g., the length of one time step).
That said, if the geometry you want really looks like the one you show, with a large deformation of the top surface, then this will not work for different reasons. You will end up with a mesh that is so deformed that you cannot expect to do useful computations on it. For domains that are as deformed as the one you show, you probably will want to create a mesh from scratch, rather than take the box geometry and deform it into the shape you show.
Best
W.
Thank you John for your helpful comments.
Mahdi
Thank you for your comments. I appreciate it. I will try the tool you suggested and let you know if I have any questions.
Mahdi