ERROR M50: component_mod:check_fields NaN found in ATM instance

[xueyanzhang520@gmail_com]

Hi, I only run CLM50 and build it successfully. However, when I ./case.submit, there is an error and the cesm.log showed:84  (t_initf) Read in prof_inparm namelist from: drv_in 85  (t_initf) Using profile_disable=          F 86  (t_initf)       profile_timer=                      4 87  (t_initf)       profile_depth_limit=                4 88  (t_initf)       profile_detail_limit=               2 89  (t_initf)       profile_barrier=          F 90  (t_initf)       profile_outpe_num=                  1 91  (t_initf)       profile_outpe_stride=               0 92  (t_initf)       profile_single_file=      F 93  (t_initf)       profile_global_stats=     T 94  (t_initf)       profile_ovhd_measurement= F 95  (t_initf)       profile_add_detail=       F 96  (t_initf)       profile_papi_enable=      F 97  ERROR: 98  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo 99  bal index:     4578100  ERROR:101  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo102  bal index:     6782103  ERROR:104  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo105  bal index:     9155106  ERROR:107  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo108  bal index:    11344109  ERROR:110  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo111  bal index:    12538112  ERROR:113  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo114  bal index:    14727115  ERROR:116  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo117  bal index:     3392118  ERROR:119  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo120  bal index:     5772121  ERROR:122  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo123  bal index:    15921124  ERROR:125  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo126  bal index:    17115127  ERROR:128  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo129  bal index:    18110130  ERROR:131  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo132  bal index:    19304133  ERROR:134  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo135  bal index:    21493136  ERROR:137  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo138  bal index:    22687139  ERROR:140  component_mod:check_fields NaN found in ATM instance:    1 field Sa_dens 1d glo141  bal index:    24876 I dont know this is because my forcing data or clm. 

 

[oleson]

This is an atm forcing data problem.  Sa_dens is air density.

 

[xueyanzhang520@gmail_com]

Hi, Thanks for reply! But forcing data does not include air density, this variable. It only has radiation, wind, humidity...

 

[oleson]

What atmospheric forcing dataset are you using?

 

[xueyanzhang520@gmail_com]

NLDAS-2 forcing product

 

[oleson]

If you are using 'CLMNCEP' mode for your forcing, then air density is calculated in cime/src/components/data_comps/datm/datm_comp_mod.F90 as:           !--- density ---          vp = (a2x%rAttr(kshum,n)*pbot) / (0.622_R8 + 0.378_R8 * a2x%rAttr(kshum,n))          a2x%rAttr(kdens,n) = (pbot - 0.378_R8 * vp) / (tbot*rdair) So maybe some combination of humidity, pressure, temperature is giving you a bad value of density?

 

[xueyanzhang520@gmail_com]

I used CLMNCEP! I saw air temperature used in this formula! So Can I have missing values in my forcing data to drive CLM model because I think this error maybe due to missing values of air temeprature from NLDAS-2

 

[oleson]

Yes, that's possible.

 

[xueyanzhang520@gmail_com]

I checked the missing values in NLDAS data and set it as _Fillvalue (-1.e+36). I run the model again, but I found another error: 2548  NetCDF: Variable not found2549  ENDRUN:2550  ENDRUN:2551  lnd_import                       ERROR: Longwave down sent from the atmosphere2552  model is negative or zero2553  ERROR: Unknown error submitted to shr_abort_abort.2554  ENDRUN:2555  lnd_import                       ERROR: Longwave down sent from the atmosphere2556  model is negative or zero2557  ERROR: Unknown error submitted to shr_abort_abort.2558  lnd_import                       ERROR: Longwave down sent from the atmosphere2559  ENDRUN:2560  lnd_import                       ERROR: Longwave down sent from the atmosphere2561  model is negative or zero2562  ERROR: Unknown error submitted to shr_abort_abort.2563  model is negative or zero2564  ERROR: Unknown error submitted to shr_abort_abort.2565  ENDRUN:  

 

[oleson]

Maybe your domain file is not consistent with the NLDAS2 data.  The "mask" field on that file should have ones where there is valid forcing data and zeros where there is not.The model is probably interpreting -1.e36 as a valid downward longwave value (longwave down sent from the atmosphere is negative or zero).

 

[xueyanzhang520@gmail_com]

Hi, I asked my fellow and they told me to use this command cdo fillmiss2 input.file output.file to replace all missing values. It works now! Thank you very much!

 

[jack]

Maybe your domain file is not consistent with the NLDAS2 data. The "mask" field on that file should have ones where there is valid forcing data and zeros where there is not.The model is probably interpreting -1.e36 as a valid downward longwave value (longwave down sent from the atmosphere is negative or zero).

Hi,oleson, I also met the same problem, when I ./case.submit, there is an error in cesm.log : ERROR: component_mod:check_fields NaN found in ATM instance: 1 field Sa_u 1d global index: 398. I want to ask what does Sa_u means? The atmospheric forcing dataset I'm using is CLDAS, and using ‘CLMGSWP3v1’ mode. Strangely, I can successfully run in another study area (i.e. with another domain file and surface data). It's the matter of atmospheric forcing or domain file?

 

[zqzq]

This is an atm forcing data problem. Sa_dens is air density.

component_mod:check_fields NaN found in ATM instance: 1 field Sa_shum 1d glo，what is Sa_shum?thanks

 

[Gabriel.Hes]

Dear all,
I am running a historical simulation and a future simulation with CLM. Somehow I am facing a similar issue to what is mentioned in this thread.
-My historical simulation (forced by GSWP3v1) starts in 1990 and runs until 2012-05-18 where the run fails and I get the following error in the cesm.log:

ERROR:
component_mod:check_fields NaN found in ATM instance: 1 field Faxa_rainc 1d
global index: 214534

-My future simulation starts at 2015 and runs until 2026-05-18 where the run fails with the same error as above. This simulation is forced by RCP8.5 atmospheric anomaly forcing from CMIP6.

I have tried to change the forcing data with the abovementioned command "cdo fillmiss2" but it still gets stuck.
What surprises me is that the model run fails on May 18th in both simulations, although the case is different. Does anyone know what the problem could arise from? Is the precipitation file the only forcing file driving the Faxa_rainc variable ? I'm not sure I understand which of the domain file or atmospheric forcing files should be filled for missing values.
Thanks for your help!

 

[oleson]

Sorry I never answered this post @Gabriel.Hes , did you find out what the problem was with Faxa_rainc? There is another user with what I think is the same problem and I don't have any solutions:

ERROR: (shr_stream_getCalendar) ERROR: nf90_open file /media/dell/data/cesm/inputdata /lmwg/atm_forcing.datm7.GSWP3.0.5d.v1.c170516/Precip/clmforc

hi dear all, I am trying to run CLM5 and found the below error: I guess it was form reading the NC file and thus, I checked the files and make sure that they are not empty. Can anybody help me with this ? Thanks！...

bb.cgd.ucar.edu

 

[kaeritai]

Sorry I never answered this post @Gabriel.Hes , did you find out what the problem was with Faxa_rainc? There is another user with what I think is the same problem and I don't have any solutions:

ERROR: (shr_stream_getCalendar) ERROR: nf90_open file /media/dell/data/cesm/inputdata /lmwg/atm_forcing.datm7.GSWP3.0.5d.v1.c170516/Precip/clmforc

hi dear all, I am trying to run CLM5 and found the below error: I guess it was form reading the NC file and thus, I checked the files and make sure that they are not empty. Can anybody help me with this ? Thanks！...

bb.cgd.ucar.edu

Hi Oleson,
thank you for your kind reply and help.
I will try myself too in the following days.

 

[Gabriel.Hes]

Hi @oleson and @kaeritai,
Yes, I think the problem came from corrupt atmospheric Precipitation files. I noticed that some files were really small compared to the average monthly GSWP precipitation files that were around 248M. @kaeritai can you check that your files are all good?
In my case I had to redownload the specific corrupt files with the command: check_input_data –download and it solved the error we were mentioning:
component_mod:check_fields NaN found in ATM instance: 1 field Faxa_rainc 1d
I hope this can help!
Gabriel

 

[kaeritai]

Hi @oleson and @kaeritai,
Yes, I think the problem came from corrupt atmospheric Precipitation files. I noticed that some files were really small compared to the average monthly GSWP precipitation files that were around 248M. @kaeritai can you check that your files are all good?
In my case I had to redownload the specific corrupt files with the command: check_input_data –download and it solved the error we were mentioning:
component_mod:check_fields NaN found in ATM instance: 1 field Faxa_rainc 1d
I hope this can help!
Gabriel

Hi dear Gabriel,
thank you very much for your kind reply. Following your suggestions, I checked the data(all the default GSWP data in 2010) and found
nothing wrong. Therefore, I downloaded the data in 2011 and resumit the case it seems that the model can read the data but with another error:

ERROR: (shr_stream_verifyTCoord) ERROR: calendar dates must be increasing

I found a similar thread on this website (ERROR: calendar dates must be increasing, user added atmospheric forcing) and thus tried to figure out this way; but found nothing wrong in my namelist files(I upload them and changed the extensions).

One question is that ALL the data are the default data , why the others found no such problems ?
I guess this should not be caused by the data ; Is it caused by the setup of the case by myself?



Can you or @oleson help me out?

 

[oleson]

Can you attach all of your log files, and your datm_in file?
You might also try ncdump -v time on your forcing files and make sure they have valid time values.

 

[kaeritai]

Can you attach all of your log files, and your datm_in file?
You might also try ncdump -v time on your forcing files and make sure they have valid time values.

hi dear Oleson,thank you for your suggestion. I thus checked the other input data and found it was INDEED caused by the 'time' and therefore,
I download the update ones.
My another question is that Since I want to explore the land use change effects on the runoff variations over the globe in recent 40 years, I am not sure whether the choice of compset "I2000Clm50Sp" is suitable for my objective? otherwise, do I need to choose the HIST compsets? Sorry that I am very new to CLM.