Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

Help with Obs Nudging problem.


New member
Hi all,
I try to use obs-nudging with WRF run. I use both upper air and surface data in little-r format from website ( It successful generated OBS_DOMAIN101 and OBS_DOMAIN102. Run ./real.exe worked well and generated wrfinput_d01, wrfinput_d02 and wrfbdy_d01. But as i run the wrf.exe, the error file showed:

*** PROBLEM: sounding, p and ht undefined 40.40000 116.3000
ob too early
n= 1 unknown ob of type FM-88 SATOB
ob too early
n= 1 unknown ob of type FM-88 SATOB
Tile Strategy is not specified. Assuming 1D-Y
WRF TILE 1 IS 1 IE 48 JS 1 JE 30
OBS NUDGING is requested on a total of 2 domain(s).
++++++CALL ERROB AT KTAU = 0 AND INEST = 1: NSTA = 777 ++++++

and the model stop running.

If i change the namelist.input parameter from obs_nudge_opt = 1,1, to obs_nudge_opt = 0,1, Same error file information but wrf.exe can work.

So, what is the problem? Is the d01 domain too large? Any help for this problem would be greatly appreciated.
I've also attached the namelist.input and rsl.error.0000.

Thanks a lot.

Dong Liu


  • rsl.error.0000
    118.5 KB · Views: 3
  • namelist.input
    7 KB · Views: 4
Hi Dong,
Do you mind packaging together all of the rsl* files for the failed case into a single *.tar file, and if you happen to still have the rsl* files for the successful simulation (with obs_nudge_opt = 0,1), can you package those up separately into their own *.tar file and attach those, as well? Thanks!
I have a problem similar to the case above and need your crucial help.
I have used MADIS data from NOAA that includes surface, soundings, aircraft, ship and profilers observations. After several troubleshooting with the preprocessor codes, I was able to convert the MADIS data to Little_r format using the MADIS2LITTLER converter code. And then, I used OBSGRID code to process for a quality control and produce (OBS_DOMAIN?01, and, where ? = 1,2,3). One-day range of WRF-Obs Nudging TEST run seems to complete normally, although it reported several warning issues, for example " n= 123 unknown ob of type FM-13". When I run WRF with Obs Nudge for several days, however, it failed at the 35-forecast hour with "Program received signal SIGSEGV: Segmentation fault - invalid memory reference" screen report, as you will also see in detail in the tarred rsl-all attachment. I have also included the 'namelist.input'.
So, I have two problems and need your needed help:
1) Why WRF-Obs Nudging coding system can not identify the 'Type FM-??' obs that are from NOAA's MADIS database; is there any remedy I can apply to resolve this issue in the preprocessors? I think the segmentation fault issue could be from this problem. Please NOTE that I have modified MADIS2LITTLER to mark observations as soundings (in module_output.F, subroutine write_littler_onelvl by setting is_sound = .TRUE.) before compiling the code.
2) Can the "Segmentation fault - invalid memory reference" be a different independent problem and can anyone suggest some solution?
Thank you so much!


  • namelist.input
    11.9 KB · Views: 4
  • rsl-all_MADIS4ObsNudg_3dmns.tar.gz
    1.6 MB · Views: 3
Quick update: WRF with Obs Nudging completed successfully after the time step dropped from 6*delta_x to 3*delta_x (So this was a CFL issue). My former control WRF runs used 6*delta_x time step.
BUT, I still have the:
"n= 46 unknown ob of type FM-13
n= 47 unknown ob of type FM-15
n= 48 unknown ob of type FM-132 ..."
Problem (hundreds of them). Any suggestion in resolving this issue would be much appreciated. Again, as you can see in my writeup above, I used NOAA's MADIS NetCDF formatted all available observations for the observational nudging.
Thank you!
I did some searching and found this additional OBSGRID Users' Guide that was shared with us several years ago. On page 19 of that guide, I found this paragraph that essentially states that these error messages are not a concern and the observations will still be used, so I believe you should be okay to proceed.

If obs_ipf_in4dob is activated certain details of the nudging configuration are listed at thebeginning of the simulation (e.g., nudging strength, spatial and temporal weighting functionparameters) as well as details about reading in observations for data assimilation. The timewindow of observations being read in is listed (in model-relative time) as well as the number ofobservations currently available for observation nudging. If an observation is read in with aplatform name not recognized by the observation nudging code, it will print an error that includesthe message “unknown ob of type XXX” where XXX is the platform. However, the observationwill still be used. The code indicates that the platform type may be used in the future to createplatform-specific nudging weight adjustments, but this is not done at this time. The platform typeis currently only used to do a small number of adjustments for specific platforms (e.g., moistureobservations with a type of “SHIP” are marked as missing if the relative humidity is less than70%).