Hi Judy,
The melt global model does not implement melt extraction through the lithosphere. What we implemented in ASPECT is the flow of magma through a porous medium (so appropriate for the part of the mantle that deforms viscously, with melt fractions below 30%). If you want to extract your melt through the lithosphere, that would require some sort of additional physics.
The problem with some of the simpler melt extraction routines implemented in other codes, where you take out melt from one cell and then place it at the surface, is ASPECT’s adaptive mesh. Taking our melt and putting it somewhere else requires searching through all cells to find the right one, which is very slow with an unstructured mesh (which you need for adaptive mesh refinement). This is especially problematic in models that run in parallel on many processors. Therefore we have not implemented models like that in ASPECT, but if you have a small model and are fine with the computational cost, you could definitely implement something like that yourself.
But this is why I suggested reducing the permeability instead. This is something that would be possible within ASPECT’s finite element framework (we would still solve the same equation). You would have to change the code in the melt global model and assign a different reference permeability for your different compositional fields.
Best wishes,
Juliane