Opened 23 months ago
Closed 20 months ago
#124 closed defect (wontfix)
Crash of GCM runs because of dust over the NPC
| Reported by: | jnaar | Owned by: | jnaar |
|---|---|---|---|
| Priority: | major | Milestone: | MARS: MCDv6.0alpha |
| Component: | MARS GCM | Version: | |
| Keywords: | Cc: |
Description
GCM run crashing because NaN appears in dust tracers at coordinates (lon,lat)=(129.375,86.25). It appears that this grid cell is located on the NPC (ie watercaptag=.true.).
Furthermore there seems to be a strong correlation between dust tracers going ludicrous and watercaptag=.true., which is obviously a bug. Strong suspicion on the subtimestep in vdifc_mod.F for water which may cause this problem somehow.
Attachments (2)
Change History (3)
Changed 23 months ago by
| Attachment: | ncview2.PNG added |
|---|
comment:1 Changed 20 months ago by
| Resolution: | → wontfix |
|---|---|
| Status: | assigned → closed |
This type of crash was actually caused by excessive latent heat release within the subtimestep of cloud microphysics on high obliquity runs where sublimation is extremely intense. It can be adressed by increasing imicro (default is 30 for present-day climate, can be increased to ~200-500 to solve this).
ncview startfi crashing