ERROR: too much NH4 uptake predicted by FUN

[Xueli Huo]

Hi all,

I got the error from cesm.log file

"
2073: WARNING: Wet_Bulb algorithm failed to converge. Setting to T: WB, P, T, RH, Q,
3094: problem with limitations on nh4 uptake 0.000000000000000E+000
3094: -Infinity
3094: ERROR: too much NH4 uptake predicted by FUN
3094:MPT ERROR: Rank 3094(g:3094) is aborting with error code 1001.
"

Here is the configuration of the case

resolution f09_f09_mg17
compset 2000_DATM%GSWP3v1_CLM50%BGC-CROP_SICE_SOCN_MOSART_SGLC_SWAV

I use the clm5.0.06.

Is there any suggestion to fix the error ?

Sincerely,

 

[oleson]

During CLM5 development, we ran across this error occasionally. We didn't track down the cause of it, but it generally seemed to be triggered by certain parameter sets (i.e., the clm paramfile). Are you running with a different parameter set or any code mods?
The errors we encountered during development were caused by slightly exceeding the threshold on sminn_to_plant_fun_nh4_vr(c,j)-smin_nh4_to_plant_vr(c,j) in SoilBiogeochemCompetitionMod.F90 (which is 1.e-7). However, your output indicates that smin_nh4_to_plant_vr(c,j) = -infinity, which makes me think there could be something more fundamentally wrong with your case. The fact that the wet bulb algorithm is not converging sometimes indicates a problem with atmospheric forcing (e.g., air temperature, humidity, solar/longwave radiation, etc.).
So you might check that.
Is your case on cheyenne?

 

[Xueli Huo]

Hi Oleson,

Thanks so much for your reply and indication.

I am running clm 5.0.06 with data assimilation. I use the default parameter set and code mods.

There are 60 ensemble members in the case. Here is the case directory
/cases/ctsm/clm5.0.06_f09_assim_inf_rundecade_LAI_e60
and the corresponding run directory
/glade/scratch/huoxl/ctsm/clm5.0.06_f09_assim_inf_rundecade_LAI_e60/run

 

[oleson]

You are using code mods here:
/glade/work/huoxl/cases/ctsm/clm5.0.06_f09_assim_inf_rundecade_LAI_e60/SourceMods/src.clm

But I don't see that they would necessarily have anything to do with your error.

I've poked around a bit but haven't found any problems with your case or your atmospheric forcing, although...
one thing I'm confused about is that is seems you are running with gregorian calendar, yet your atmospheric forcing files have a "no_leap" attribute and no data for leap days. Normally, I think this would cause an error in year 2000 because the datm wouldn't know what to do, however, it seems like you've run successfully past this point.
I can't tell if just one of your 60 members is throwing the error, or if they all are.
What are you assimilating? Obviously, I would check to see if reasonable values are being assimilated.

The only thing I can think of at this point is to track back in the code why smin_nh4_to_plant_vr(c,j) = -infinity, perhaps by isolating the offending case and putting some print statements in. Alternatively, try to perturb the offending case (e.g., with different initial conditions) and see if you get past the error.
Sorry I can't be of more help here.

 

[oleson]

I was noticing some "h1" files in your run directory that have TLAI in them. However, there seem to be missing values (9.96921e36) of TLAI in southern south america. Is that expected?

 

[Xueli Huo]

Hi Keith,

When I checked the h1 files yesterday, I noticed this abnormal phenomena. It is not expected. Since the first day of 2001, the TLAI in southern south america is missing which is not expected.