Given the higher mesh densities at the boundaries (and therefore finer mesh resolutions), the simple approach for filtering the fictitious ice thickness added by Elmer ( $h_{\rm min}$ ) no longer works correctly:
If the vertical mesh resolution is fine enough, the ice thickness in the second (or even third) vertical nodes from the bottom will still be less then $h_{\rm min}$ and therefore forced to zero.
Proposed Solution:
Can we use the calculate_length function to aid in the filtering? Since it only used the ice-thickness along the free surface it shouldn't be subject to this same bug.
Given the higher mesh densities at the boundaries (and therefore finer mesh resolutions), the simple approach for filtering the fictitious ice thickness added by Elmer ( $h_{\rm min}$ ) no longer works correctly:
https://github.com/andrewdnolan/thermal-structure/blob/fe33d8bbeb79c1a4dcf72c4200790cf4f5cac98b/src/thermal/open.py#L108
If the vertical mesh resolution is fine enough, the ice thickness in the second (or even third) vertical nodes from the bottom will still be less then $h_{\rm min}$ and therefore forced to zero.
Proposed Solution:
calculate_lengthfunction to aid in the filtering? Since it only used the ice-thickness along the free surface it shouldn't be subject to this same bug.