Having trouble getting wrf.exe to run with modified wrfinput_d0* files

[stormsteiner23]

I've been trying to run an experiment with WRF using four domains. After I complete the real.exe run and wrfinput_d0* files are generated, I take those files and use xarray in Python to modify the soil moisture fields in each domain. I then save those files and run wrf.exe. The problem is that I get the error attached (in the rsl.error.0000 file) and the run shuts down before it produces anything.

I am not entirely sure what is going on, as I can use ncdump with the wrfinput files and it works just fine. Wrf.exe just can't seem to do anything with the files.

A related question is this: if I want to modify soil moisture initial conditions (i.e. create artificial fields where soil moisture values are set to field capacity or wilting point based on soil type), is the best way to do this to modify the wrfinput files produced by real.exe or is there another, better method?

Attached are my error file and my namelist.
Thanks for the help!

Edit: for reference, here is the error that I keep getting:

FATAL CALLED FROM FILE: <stdin> LINE: 70
program wrf: error opening wrfinput_d01 for reading ierr= -1021

 

[Ming Chen]

Are you able to run wrf.exe using the wrfinput files that are not changed?

 

[stormsteiner23]

Are you able to run wrf.exe using the wrfinput files that are not changed?

Yes, my control run was successful fortunately

 

[Ming Chen]

This indicates that the file you changed somehow is wrong. Please compare the original wrfinput with the one you modified, and pay special attention to those variables you modified. The error message din't provide many information, but I suppose the comparison may tell us more ...

 

[stormsteiner23]

This indicates that the file you changed somehow is wrong. Please compare the original wrfinput with the one you modified, and pay special attention to those variables you modified. The error message din't provide many information, but I suppose the comparison may tell us more ...

I did end up doing the comparison and found that when i was editing the soil moisture variable with xarray that it was not using any coordinate or dimensional information and so I was wondering if that was causing the error. I tried fixing that and was successful but I still get the same error that I was before.

 

[Ming Chen]

Can you attach the original wrfout and the one you modified for me to take a look? thanks.

 

[stormsteiner23]

Hi, sorry about the delay. Here is a link to a One Drive folder with the original wrfinput and the modified wrfinput files for the first domain.

https://buckeyemailosu-my.sharepoint.com/:f:/g/personal/steiner_273_buckeyemail_osu_edu/EpNNbRoG7DdFqamWJB9b1fIBj1LRa_8MIYGuTBiGk_Y7Lg?e=JhucsX

 

[Ming Chen]

If we compare the two files, we can find obvious differences. For example,

"float XLAT(Time, south_north, west_east)" versus "float XLAT(south_north, west_east)"

Also, no missing value is a allowed in all data files related to WRF.

There are many other differences between the two files. Please follow the exact format of the original wrfinout when you create the new datafile.

 

[shaider2]

Hi I am having similar troubles. When I am decreasing the soil moisture by 15 percent I am getting negative values. So when I am trying to modifying it usinf xarray in python. Unfortunately, the wrf.exe is not working. Is there any other way to remove the negative values using NCL

 

[Ming Chen]

Note that soil moisture has its physical meaning and it cannot be negative. The negative value indicates that something went wrong in your python script. If you would like to try NCL, please take a look at the same here:

NC: File Handling, File IO: Opening and Editing an existing netCDF file

NCL file IO example page. Demonstrates how to change a netCDF file from within NCL using addfile.

www.ncl.ucar.edu

 

[Ming Chen]

Note that soil moisture has its physical meaning and it cannot be negative. The negative value indicates that something went wrong in your python script. If you would like to try NCL, please take a look at the same here:

NC: File Handling, File IO: Opening and Editing an existing netCDF file

NCL file IO example page. Demonstrates how to change a netCDF file from within NCL using addfile.

www.ncl.ucar.edu

 

[shaider2]

Hi,
Thanks for replying. I am using NARR 3 hr reanalysis data for boundary condition. However, as I am trying to do a sensitivity analysis with soil moisture I was trying to modify the soil moisture value in the wrf_input files. When I was trying to decrease my soil mositure by 15 percent I started getting negatie values. Therefore, I intend to change the negative values to 0 to get a correct result. For replacing the negative values with 0 in NCL i used the following expression
SM( :,:,i,j) = where(SM( :,:,i,j).lt.0,SM( :,:,i,j)*0,SM ( :,:,i,j)). After the modification wrf.exe is not working. I am also attaching my NCL scripts for your reference.

 

[Ming Chen]

I don't think it is physically reasonable to specify zero soil moisture. WRF fails due to unreasonable input data.

 

[shaider2]

How do I know the difference between the input data?
Is 0 soil moisture is the problem? How should I get rid of the negative values?

 

[stormsteiner23]

If we compare the two files, we can find obvious differences. For example,

"float XLAT(Time, south_north, west_east)" versus "float XLAT(south_north, west_east)"

Also, no missing value is a allowed in all data files related to WRF.

There are many other differences between the two files. Please follow the exact format of the original wrfinout when you create the new datafile.

Just to update on this, I was able to figure it out. The issue arose because my Python script was making a copy of the wrfinput file and then modifying the copy rather than modifying the original and I wasn't able to construct the new netcdf file in such a way that would work for the model