distance from border is flawed in rgi 5 and 19 since my geometries no not include the ice sheet :) I should include
the ice sheet geometry. An easier alternative would be to only consider in these regions the distance from the cluster
nunataks, and remove the external boundary from the calculation. If the cluster does not contain nunataks
then yes consider the cluster external geometry