Summary
eCLM restart files generated by eCLM–ParFlow coupled simulation contain non-physical H2OSOI_ICE values (1.4e30 – 2.15e34) on all 404 urban columns (ltype=9). A scan of every column-dimensioned restart variable found this corruption nowhere else in the EUR-12 domain. It seems specific to H2OSOI_ICE on urban columns.
Downstream impact
Reusing such a restart as the initial condition for an ICON–eCLM coupled run propagates the bad value through eCLM's ice/energy
budget into the coupled sensible heat flux (ICOSHFLX ≈ 1e35 W/m²), which ICON leads to crashes.
Current state / plan
Root-cause why the eCLM–ParFlow coupled spin-up path leaves H2OSOI_ICE at these out-of-range values on urban columns, and fix it at the source so future spin-up restarts aren't corrupted. As a temporary workaround, the affected restart files could be repaired or the output variables could be modified before writing the restart.
Summary
eCLM restart files generated by eCLM–ParFlow coupled simulation contain non-physical
H2OSOI_ICEvalues (1.4e30 – 2.15e34) on all 404 urban columns (ltype=9). A scan of every column-dimensioned restart variable found this corruption nowhere else in the EUR-12 domain. It seems specific toH2OSOI_ICEon urban columns.Downstream impact
Reusing such a restart as the initial condition for an ICON–eCLM coupled run propagates the bad value through eCLM's ice/energy
budget into the coupled sensible heat flux (
ICOSHFLX≈ 1e35 W/m²), which ICON leads to crashes.Current state / plan
Root-cause why the eCLM–ParFlow coupled spin-up path leaves
H2OSOI_ICEat these out-of-range values on urban columns, and fix it at the source so future spin-up restarts aren't corrupted. As a temporary workaround, the affected restart files could be repaired or the output variables could be modified before writing the restart.