About the radiation schemes

Discuss advection schemes, timestepping, pressure gradients, coordinates, etc.

About the radiation schemes

Postby Remi » Tue Apr 14, 2009 5:28 am

Hi,

I am running the NMM core of version 3.0.1.1 (just about to move to version 3.1).

I have had several crashes in the GFDL longwave & shortwave schemes : the model otherwise runs smoothly but if at a timestep of 40s, I set the radiation calls at 270 timesteps (ie 6h) I get a systematic crash.

typically, it is during the module_ra_gfdleta.f90 (gfdllw) call

SUN-EARTH DISTANCE CALCULATION FINISHED IN SOLARD
YEAR= 2009 MONTH= 4 DAY= 10 HOUR= 3 R1= 1.0018
forrtl: severe (174): SIGSEGV, segmentation fault occurred

Before I go into deeper debugging, I was wondering if anyone had encountered such crashes and could point me in the right direction?

Thanks a lot,

Remi
Remi
 
Posts: 8
Joined: Wed Apr 08, 2009 10:30 am

Re: About the radiation schemes

Postby KeiDev » Tue Jun 23, 2009 10:09 pm

Hi,

I got the same problem when I run NMM ver.3.1 by dmpar.
---
SUN-EARTH DISTANCE CALCULATION FINISHED IN SOLARD
YEAR= 2009 MONTH= 6 DAY= 22 HOUR= 1 R1= 1.0163
[localhost:10209] *** Process received signal ***
[localhost:10209] Signal: Segmentation fault (11)
[localhost:10209] Signal code: Address not mapped (1)
[localhost:10209] Failing at address: 0xfffffffe04b5674c
[localhost:10209] *** End of error message ***
---

Such crach also occured when the other radiation schemes (ex. RRTM) were set.
It seems that there are nothing wrong with module_ra_gfdleta.f.

I tried to run the model with no radiation and then wrf.exe successful completed, but it's not good idea.

I have not been solved this problem yet.

Please let me know if you have any progress.


Thanks,
KeiDev
 
Posts: 2
Joined: Mon Jun 15, 2009 11:03 pm

Re: About the radiation schemes

Postby leeperd » Fri Oct 16, 2009 3:48 pm

Hi,

I am currently having the same issue with WRF NMM 3.1. When I set the radiation scheme to zero the model runs fine. I think it has to do with MPI because I ran with both radiation schemes (99 and 1) and they both worked properly when I ran the model on a single CPU. I was wondering how you guys solved this issue? Did you stick with WRF NMM 3.0?? Any help from anyone on this matter would be greatly appreciated, because simulating nested domains at high resolution on a single processor with a machine outfitted with 8 CPUs or a cluster doesn't make much sense either. I usually only use the WRF ARW core and this is not an issue for its current 3.1 version.

Thanks,
Ronnie Leeper
Western Kentucky University
Bowling Green, KY
Ronnie Leeper
USCRN Quality Assurance Specialist, NCDC
Research Associate, CICS-NC
Global Climate Applications Division, NCDC
http://www.cicsnc.org/people/ronald-leeper/
leeperd
 
Posts: 21
Joined: Thu Nov 20, 2008 5:26 pm

Re: About the radiation schemes

Postby scaine » Sat Feb 02, 2013 12:18 am

Has anyone solved this problem? I am also getting this in version 3.3.1. (NMM core)
THe model crashes on radiation call (usually in the first few minutes of the simulation) and only runs to completion if the radiation physics are turned off.

also, probably unrelated (because without radiation physics on the simulation competes) I also get the following at the start of the simulation (repeated many times for different points)

PSEUDO HYDROSTATIC IMBALANCE AT THE WESTERN BOUNDARY AT 1 66 GRID # 1
..
..
and then a short while later

very nonhydro...: 69 8 28 0.76766163
..
..

Does anyone know whether or not it needs to be addressed, or does the model fix it up by itself.
This seems to be a NMM issue, I compiled WRF using the ARW core, and can complete the simulation using the same intput data and similar domain size/extent.
scaine
 
Posts: 17
Joined: Tue Dec 23, 2008 9:53 am

Re: About the radiation schemes

Postby scaine » Fri Mar 15, 2013 6:08 am

responding to my own issues since no-one else on these damn almost useless forums seems to answer anything.
If you decrease the time-step you can solve this issue
scaine
 
Posts: 17
Joined: Tue Dec 23, 2008 9:53 am

Re: About the radiation schemes

Postby jimmyc » Sun Mar 17, 2013 11:35 pm

Sorry that few can help. My impression is that very few people use the NMM core, at least from the questions we get in the forum. The NMM core was from NCEP and therefore receives little in the way of support. They have moved to the NMMB core now so there is probably less now.

This forum is largely made up of users. If you have questions you can also email wrfhelp@ucar.edu
They too are largely unfunded but do offer some support, especially if you believe there is a bug in the model code.

Again apologies.
The views expressed in this message do not necessarily reflect those of NOAA or the National Weather Service or the University of Oklahoma.
James Correia, Jr
jimmyc
 
Posts: 519
Joined: Tue Apr 15, 2008 1:10 am

Re: About the radiation schemes

Postby scaine » Mon Mar 18, 2013 3:36 am

No worries, was a little frustrated the day I made my last post, frustrated like only computers can make you be.
I am as guilty as everyone else in not posting, which is why I have made a vow to check the forums as frequently as possible and answer any questions that I can.
scaine
 
Posts: 17
Joined: Tue Dec 23, 2008 9:53 am


Return to Dynamics and Numerics

Who is online

Users browsing this forum: No registered users and 2 guests