One-Way Nested WRF/Chem 3.7 - Using ndown

Any issues with running WRF Chemistry

One-Way Nested WRF/Chem 3.7 - Using ndown

Postby pccampb2 » Fri Oct 16, 2015 5:34 pm

Dear WRF User's,

I am having an issue with performing a one-way nested simulation using the released version of WRF/Chem 3.7.
Some background on my simulation is as follows:

We have successfully run our configuration of WRF/Chem 3.7 for a 36-km North American domain (ICs/BCs downscaled from CESM simulations), and I am using these 36 km output files with the basic ndown methodology for generating the necessary inputs for a 12-km CONUS domain. It seems my real and ndown procedures complete successfully (i.e., generated full wrfinput_d01 and wrfbdy_d01 files), but I am getting a segmentation fault at the first hour of running wrf. A similar error also occurs when I run WRF 3.7 met-only using ndown from our 36 km output. The fault occurs near the radiation code, and seemingly near the following:

Image PC Routine Line Source
wrf.exe 00000000004EF581 module_integrate_ 313 module_integrate.f90

The compiled version of WRF/Chem 3.7 runs OK when using regridded inputs from 36 km, that are not from ndown.

Does anyone have much experience with running one-way nested WRF/Chem 3.7 using ndown, and would you be able to provide any valuable insight into our errors?

Thank you,

Patrick
pccampb2
 
Posts: 10
Joined: Wed Aug 06, 2014 5:09 pm

Re: One-Way Nested WRF/Chem 3.7 - Using ndown

Postby kwthomas » Fri Oct 16, 2015 6:15 pm

Hi...

With regular WRF, seg faults mean the timestep is too large. You can try your it and see if that helps.
Kevin W. Thomas
Center for Analysis and Prediction of Storms
University of Oklahoma
kwthomas
 
Posts: 220
Joined: Thu Aug 07, 2008 6:53 pm

Re: One-Way Nested WRF/Chem 3.7 - Using ndown

Postby pccampb2 » Mon Oct 19, 2015 11:12 am

Thank you for the reply! However, I have tested it with numerous lower time steps, and I still get the segmentation fault. The WRF/Chem 3.7 model crashes after using ndown (runs OK) to provide its inputs, while wrf seemingly runs OK when not using ndown.

Thus, this seems like an ndown issue here, particularly with some likely bad inputs when calling the photolysis driver, as WRF crashes with the segmentation fault:

photolysis_driver: called for domain 1
d01 2005-01-01_00:00:00 calling ftuv_driver
d01 2005-01-01_00:00:00 SORGAM aerosols initialization
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
wrf.exe 0000000003BE7A4F module_ftuv_subs_ 2032 module_ftuv_subs.f90
wrf.exe 0000000003BE53DF module_ftuv_subs_ 1768 module_ftuv_subs.f90
wrf.exe 0000000003BDB2F8 module_ftuv_subs_ 444 module_ftuv_subs.f90
wrf.exe 0000000002422158 module_ftuv_drive 1238 module_ftuv_driver.f90
wrf.exe 000000000240F8A5 module_ftuv_drive 508 module_ftuv_driver.f90
wrf.exe 0000000001B61F05 photolysis_driver 176 photolysis_driver.f90
wrf.exe 0000000001B3FF78 chem_driver_ 961 chem_driver.f90
wrf.exe 00000000019E6A6D solve_interface_ 172 solve_interface.f90
wrf.exe 00000000004EF581 module_integrate_ 313 module_integrate.f90

Any further help is much appreciated!

Thanks,

Patrick
pccampb2
 
Posts: 10
Joined: Wed Aug 06, 2014 5:09 pm

Re: One-Way Nested WRF/Chem 3.7 - Using ndown

Postby pccampb2 » Mon Oct 19, 2015 12:06 pm

Update to my previous post on this WRF/Chem 3.7 ndown issue:

After looking more closely at the wrfbdy_d01 file generated by ndown, although ndown seemingly runs OK (successfully complete - no error messages), there are numerous NaN and Infinityf values contained in the wrfbdy_d01 file (coming from the 36 km output).

So, it is clear that this is causing my WRF/Chem 12 km runs to crash at the first time step after using ndown. I am not yet sure if this is either a fundamental problem with ndown here (or my methods), or if there is a problem when trying to ndown our 36 km output.

Thanks!

Patrick
pccampb2
 
Posts: 10
Joined: Wed Aug 06, 2014 5:09 pm

Re: One-Way Nested WRF/Chem 3.7 - Using ndown

Postby pccampb2 » Tue Oct 20, 2015 2:57 pm

*Another update as of 151020 to my WRF/Chem 3.7 ndown issue:

I have had the privilege to be able to work with John Cazes via RT, and after running parallel computing DDT debugger on ndown.exe with my inputs, this is what has been discovered (courtesy of John Cazes):

"I recompiled this morning with only the "-ftrapuv -traceback" option to find out where the nans and infinities are coming from. These options cause the code to stop when an overflow is detected and report the routine that it came from.

This allows me to set a breakpoint in the proper place using ddt.

The problem is coming from this line in module_big_step_utilities_em.f90:

572: rfield(i,k,j)=field(i,k,j)*muu(i,j)/msf(i,j)


The whole msf array is set to zero. This is a dummy array that comes from the msfuy array. Which according to google is a "map scale factor on the u-grid in the y-direction" .

This appears to be the factor by which U should be scaled in the y-direction. But, it's not being set by ndown.exe."

We are still looking further into this ndown issue.

I post this in case anyone has the time to check/monitor this thread, and hopefully there may be further input from anyone who has encountered this issue. Or, maybe it can help someone in the future.

Thanks,

Patrick
pccampb2
 
Posts: 10
Joined: Wed Aug 06, 2014 5:09 pm

Re: One-Way Nested WRF/Chem 3.7 - Using ndown

Postby kwthomas » Tue Oct 20, 2015 4:32 pm

Hi Patrick...

You might want to try WRF/Chem 3.7.1. There is always the possibility the problem was fixed.
Kevin W. Thomas
Center for Analysis and Prediction of Storms
University of Oklahoma
kwthomas
 
Posts: 220
Joined: Thu Aug 07, 2008 6:53 pm

Re: One-Way Nested WRF/Chem 3.7 - Using ndown

Postby pccampb2 » Wed Oct 21, 2015 5:51 pm

Hello, and thanks again for your reply!

I just tried to use my same inputs with WRF/Chem 3.7.1 as you suggest, but run into the exact same error here.

Oh well, I am still trying to find out why this whole array is zero from ndown.

Thanks again,

Patrick
pccampb2
 
Posts: 10
Joined: Wed Aug 06, 2014 5:09 pm

Re: One-Way Nested WRF/Chem 3.7 - Using ndown

Postby pccampb2 » Thu Nov 05, 2015 12:31 am

Hello, another update on my ndown problem in WRF/Chem.

While working with the WRF help desk, it appears there may be a bug in their ndown code, possibly specific to the WRF/Chem definitions in the compilation.

I will keep this post updated with whatever I find out.

Thanks,

Patrick
pccampb2
 
Posts: 10
Joined: Wed Aug 06, 2014 5:09 pm

Re: One-Way Nested WRF/Chem 3.7 - Using ndown

Postby easy » Wed Jun 08, 2016 2:08 am

Have you solved ndown problem in WRF-Chem 3.7. Could you post the way how you make it. Thanks. :roll: :roll: :roll:
easy
 
Posts: 10
Joined: Sun Jun 07, 2015 10:26 pm

Re: One-Way Nested WRF/Chem 3.7 - Using ndown

Postby sru20 » Wed Mar 22, 2017 11:16 pm

Are you using add_remove file to remove or add some variables in your run folder? I was. So after ndown, the wrf run was crashing as it couldnt find some variables. very frustrating.
sru20
 
Posts: 6
Joined: Sat May 01, 2010 5:49 am


Return to Runtime Problems

Who is online

Users browsing this forum: No registered users and 3 guests

cron