Uneven temperature distribution on a flat domain

wadewei

Hi everybody,

I'd like to get your thoughts on a simple setup I have that produced puzzling results.

It's a flat-bed and flat surface land ice, initialized with uniform temperature and geothermal heat flux. Everything but thermal is disabled for transient run. I expected the temperature profile to be perfectly symmetric. Yet the section-view shows a lot of irregularities. Any idea what's gone wrong?

I've attached the said plots. The model file (3MB) is too large to upload and can be retrieved from https://pkueducn-my.sharepoint.com/:u:/g/personal/wadew_pkueducn_onmicrosoft_com/Eca55JiG6j9AkEh__KMO9b4BPORId2eP3zsCL-6x0kpQlw?e=XvpTeW

Thanks a lot!

Wade

mathieumorlighem

Hello Wade

very simple answer: the problem is not ISSM, it is because you chose a (faster) iterative solver, which will make the solution slowly drift:

     toolkit: 'petsc'
    mat_type: 'mpiaij'
    ksp_type: 'fgmres'
     pc_type: 'bjacobi'

Change your solver to MUMPS (just type md.toolkits = toolkits(); since it is ISSM's default solver) and you will see that the temperature is what you would expect. You can try to add PETSc options (convergence criteria) to make gmres more accurate if you don't want to use MUMPS.

All the best
Mathieu