Opened 5 years ago
Closed 5 years ago
#542 closed defect (fixed)
Initialization problem: temp_sol_new becomes too low and the calculation of lwup becomes negative
Reported by: | jgipsl | Owned by: | jgipsl |
---|---|---|---|
Priority: | major | Milestone: | IPSLCM6.v2 |
Component: | Anthropogenic processes | Version: | |
Keywords: | Cc: |
Description (last modified by jgipsl)
Starting without restart files might create an incoherent temp_sol_new for the first time step. If the difference is to big with temp_sol, lwup can become negative which will make the model crash in enerbil_flux.
Following line in enerbil_flux makes the crash when lwup is negative:
tsol_rad(ji) = (lwup(ji)/ (emis(ji) * c_stefan)) **(1./quatre)
This problem exists in the trunk and CN-CAN. It has been seen for the test case using the CMIP5 PFTmap with 13PFTS, using debug compile options at irene. Same test case is running using less strong options or at other machines (ok with debug options at ada or obelix for example).
It is probably not a problem if the compiler oversees the problem as it only appears at the first time step. For this test case, temp_sol_new=210K and temp_sol=280 when the model crashes. But only for 2 grid cells at the first time step, starting without restart file.
A work around solution has been commited in CN-CAN to start with : [5777]
Change History (4)
comment:1 Changed 5 years ago by jgipsl
- Description modified (diff)
comment:2 Changed 5 years ago by jgipsl
comment:3 Changed 5 years ago by jgipsl
- Milestone set to IPSLCM6.v2
- Owner changed from somebody to jgipsl
- Status changed from new to accepted
comment:4 Changed 5 years ago by jgipsl
- Resolution set to fixed
- Status changed from accepted to closed
Revert commit [5874] containing work around.
The problem seems to be due to error in the initialization of ptn_pftmean done in thermosoil_initialize. Following commit in the trunk correct ptn_pftmean and seems to remove the error above: [5783]. The workaround above should not be needed.