Use of MERRA2 data for NoahMp modeling

[Madhuka1990]

I am new to NoahMP modeling and trying to run NoahMP with MERRA2 data in 1km resolution. I regridded the data into geo_em file (means 1km) by using ESMF regrding script.
I'm wondering whether I need to run create_forcing.exe again or just continue to run the model.

[cenlinhe]

If you already have the forcing data from MERRA2 regridded to geo_em file grids, then you can just go and run the model. Also, the current create_forcing.exe cannot handle MERRA2 data.

[Madhuka1990]

Thanks for quick reply.

[Madhuka1990]

If I don't run the create_forcing.exe, how do I create "HRLDAS_SETUP_FILE"?

Could you please give some idea

[Madhuka1990]

If I don't run the create_forcing.exe, how do I create "HRLDAS_SETUP_FILE"?

Could you please give some idea

[cenlinhe]

The most straightforward way to create a HRLDAS_SETUP_FILE is to use ERA5 or GLDAS or NLDAS data (just 1 or 2 time steps data) to run create_forcing.exe to generate the HRLDAS_SETUP_FILE. If your simulation includes spin up, then you do not need to replace the variable values in this HRLDAS_SETUP_FILE. If your spin up is short or no spin up, you can use MERRA2 variable values and regridded to the geo_em grid, and then replace the values in the generated HRLDAS_SETUP_FILE from ERA5 or GLDAS or NLDAS.

[Madhuka1990]

Thanks for the directions. I'll try these

[Madhuka1990]

Hi,
I downloaded GLDAS data to run create_forcing.exe to generate the HRLDAS_SETUP_FILE. I extracted the data from raw files using the given code (extract_gldas.py) and as per the variable names in extracted netcdf files, I adjusted the VTable in namelist.
But when I run the create_forcing.exe, I got error message as "NetCDF data does not match a Vtable entry".
I tried to fix the error, but couldn't.
Could you please help me to solve this issue

[cenlinhe]

See this link for detail: https://github.com/NCAR/hrldas/blob/7d183b9ba6ff82f746a978fb20bf79e1cb4f1466/hrldas/docs/README.GLDAS#L93

Thanks, Zhe!

[cenlinhe]

I did run the extract_gldas.py but not the extract_gldas_init.py (as this is optional). I am confusing with the create_forcing.exe as in the VTable the variable names are as T2D, Q2D etc., but when after extracting data from raw forcing data the variable names of the netcdf files are as Tair_f_inst, Qair_f_inst etc. for temperature, humidity respectively. That's why I changed the Vtable to align with data variable names. Otherwise it gives the error "Can't read variable:.."

You can follow the instructions here: https://github.com/NCAR/hrldas/blob/7d183b9ba6ff82f746a978fb20bf79e1cb4f1466/hrldas/docs/README.GLDAS#L42-L176

[cenlinhe]

The variable names "Tair_f_inst, Qair_f_inst, etc." are the extracted raw variables from GRIB data files and used by create_forcing.exe. You will see these variable names in create_forcing.f. They are not the output variable names in your forcing files.

[Madhuka1990]

Thanks @cenlinhe . I'll follow the instructions and try to sort it out.

[tslin2]

there are two discussions for references
NCAR/hrldas#91
NCAR/hrldas#103

[Madhuka1990]

I followed the steps using global data set as suggested in one of the above discussions. But create_forcing.exe only read the Tair, Qair, Psurf, U, V, LWdown & SWdown files and at the end process was stopped with following error message.
"STOP String is all blanks."

Precipitation and initial condition files were not read during the process as per my understanding.
I checked those files as well which are readable and having data.

Could you please let me know how to solve this issue

[cenlinhe]

This is weird. @CharlesZheZhang or @tslin2 Could you please take a look? Maybe download a few GLDAS example files and run the create_forcing from your end to see if you have the same issue here. Thanks!

[Madhuka1990]

I have a doubt on the create_forcing.F file as it is not reading the Precip files. I used following Makefile to generate that .F file though which name is not similar to name as in discussion in NCAR/hrldas GitHub (create_forcing_netcdf.F)
Could you please whether there is an error in that Makefile?
makefile.zip

[CharlesZheZhang]

Hi Madhuka, we used to have 2 .F files (create_forcing.F for grib format forcing and create_forcing_netcdf.F for netcdf forcing). Then we combined these two .F file into one (create_forcing.F) and add "forcing_type" to specify subroutines to read netcdf or grib data.
As for your question, it seems the code stops after successfully reading SWdown and the next variable is Precf_tavg and it shows "String" is all blank. Did you run combine_precips_netcdf.perl (and the associated compiled .f90 code)? This code will combine the rain and snow data to precipitation and will change the variable names (from Rainf_tavg and Snowf_tavg to Precf_tavg):
https://github.com/NCAR/hrldas/blob/7d183b9ba6ff82f746a978fb20bf79e1cb4f1466/hrldas/HRLDAS_forcing/run/examples/GLDAS/combine_precips_netcdf.f90#L15C1-L17C59
Hope this can help

[Madhuka1990]

@CharlesZheZhang, Thanks for the reply.
I ran the combine_precips_netcdf.perl and combined precipitation files were also generated. The details of one of the generated precipitation file is as follows;
.

Is there any way to resolve this

[tslin2]

This is weird. @CharlesZheZhang or @tslin2 Could you please take a look? Maybe download a few GLDAS example files and run the create_forcing from your end to see if you have the same issue here. Thanks!

I am starting to test this; it looks like there are TabError: inconsistent use of tabs and spaces in indentation for python 3 in the extract_gldas.py file

[Madhuka1990]

I reran the combine_precips and then ran the create_forcing.exe, but still same error is there.
What can I do to solve this

[Madhuka1990]

Hi All,
I was able to resolve the issue. Thanks for your time and helps

[CharlesZheZhang]

Hi @Madhuka1990 , great to hear that you are able to resolve this issue.
Could you also let us know what was the cause and how did you resolve it? So we can make a note and help with future users.
Many thanks! Zhe

[Madhuka1990]

Hi @CharlesZheZhang, it was my mistake. Since I didn't use secondary data for SW and precipitation, I commented out all the command lines which processes are relevant to use secondary data, but I was missed out one location (in line 364 of create_forcing.F). After correcting that program was run continuously through each variable.
But again, I have come across another set of issues and concerns as follows

1. "Source model land/sea data not found; assuming a bitmap exists" : Do I need to include separate landmask file or will it be possible to use landmask from geo_em.d01.nc file?
2. Since I'm using 3 hourly data, where can I specify that, as programm is looking for hourly data because increment in "ihour" is 1.
   

And the process was stopped again by giving above error.

Can suggest some options to resolve these as well?

Thanks
Madhuka

[Madhuka1990]

I have sorted error. Could you guys give some comments on my first 2 concerns

[CharlesZheZhang]

Hi @Madhuka1990, for your two comments:
1 source model land/sea mask is needed (assuming you are using the MERRA2 data, correct?), which you can specify in the namelist: LANDSFILE_TEMPLATE = "///

/LANDSEA.". It is needed because the program will need to interpolate from source data (MERRA2, coarse resolution) to destination domain (geo_em.d01.nc, high resolution), to avoid discrepancy of reading data usually occur near coastal line. You may want to look for this land/sea mask data from MERRA2 source.
2 I don't remember it is necessary to specify ihour in the namelist, it is a variable read from the forcing data timestep
L346: call geth_idts(date(1:13), startdate(1:13), ihour)
https://github.com/NCAR/hrldas/blob/7d183b9ba6ff82f746a978fb20bf79e1cb4f1466/hrldas/HRLDAS_forcing/create_forcing.F#L346C6-L346C56

[Madhuka1990]

Hi @cenlinhe
As you mentioned previously, I generated the HRLDAS_setup file by using GLDAS to use with MERRA2 data. When I ran the hrldas.exe with that setup file, I got following error:

I tried to solve this, but couldn't. Could you please help me with this

[Madhuka1990]

Line 677 is " NoahmpIO%ISBARREN_TABLE = ISBARREN"

Do I need to run "make" command within the "run" directory prior to running hrldas.exe?

[cenlinhe]

no, you do not need to run make under "run" directory. Please check your hrldas_setup file to see if you have ISBARREN value defined in the global attribute of this netcdf file. You can use "ncdump -h hrldas_setup_xxx.nc"

[Madhuka1990]

I checked the hrldas_setup file and there are no ISBARREN value specified in the global attribute.
How could I do that to solve this issue

[cenlinhe]

do you have a geo_em file that was used to create hrldas_setup and forcing files during create_forcing.exe step? If so, you should be able to find all those global attribute in that geo_em file by using ncdump -h geo_em command. And then you can use nco command to copy those global attributes from the geo_em file directly to your hrldas_setup file. (These attributes are supposed to be copied to your hrldas_setup file during the create_forcing.exe step, but I am not sure why they were not copied over). In addition to ISBARREN, please also pay attention to those other global attributes, e.g., ISURBAN, ISICE, etc. those are needed.

[Madhuka1990]

I added the ISBARREN to hrldas_setup file.

But still give same error mentioning that line 677 of file NoahmpReadTableMod.f90 when I ran hrldas.exe.

[Madhuka1990]

Since I'm using MERRA2 data where one data file contains 24 hours of data, I separated the file into 24 individual hourly data file by using ncks -O -d Time,$hour,$hour "$INPUT_FILE" "$OUTPUT_FILE" command. Is this approach correct?
Then I ran the hrldas.exe using those hourly data files, but process was stopped stating that "STOP Error: Energy budget problem in NoahMP LSM" . How I correct this error?

[cenlinhe]

i have not used this ncks command before, but if you just randomly select a few forcing files to double check if the hourly file has the correct values (i.e., same as those in your daily file). If so, I believe your ncks command should be doing the right thing.
Your error message may be caused by two things: (1) unrealistic values in your forcing file; or (2) your model timestep is too large. You can use a smaller NOAH_TIMESTEP (i.e., use 30-min or 15-min). Also, if your model stopped at the first time step, you may also want to check the initial condition file HRLDAS_setup file to make sure there is no unrealistic value.