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

WRF regional climate simulation

This post was from a previous version of the WRF&MPAS-A Support Forum. New replies have been disabled and if you have follow up questions related to this post, then please start a new thread from the forum home page.


New member
I am trying to perform a 2-year regional climate simulation using the WRF model, version 4.0.1. The model has been run with two two-way nested domains, with a horizontal resolution of 50 and 10 km, respectively, and 40 vertical levels. Initial and boundary conditions are supplied by the 6-hourly ERA-Interim reanalysis data (using, while the SST field is also updated every 6 hours. The thing is that the model crashes after approximately 6 hours of simulation and that's why I thought there might be a problem with the input data. I downloaded and tested two different SST datasets:
- -> weekly dataset (using Vtable.SST)
-!/dataset/reanalysis-era5-single-levels-preliminary-back-extension -> 6-hourly ERA5 dataset (using Vtable.ECMWF)
but still the model crashes. I also tested several physics options (i.e., radiation, land-surface, boundary-layer, cumulus scheme) but I didn't manage to fix the problem. I hereby attach the namelist.input and namelist.wps, as well as a .tar file including the .rsl output and error files for your review.

Any ideas or tips on how to resolve this issue would be greatly appreciated!

Thank you in advance.


  • file.tar.gz
    88.8 MB · Views: 29
  • namelist.input
    5.1 KB · Views: 49
  • namelist.wps
    1.5 KB · Views: 43
I believe this is an issue related to inconsistent landsea mask between your SST data and ERA data. Is there any special reason you want to use different SST ? If not, I would suggest you use SST from the same ERA data you use to drive WRF.
Please let me know if you have more questions.
Hello and many thanks for the quick response!
So in this case I could use the SST data included in the ERA-I dataset (updated every 6 hours) and keep SST_UPDATE off? Because my simulation is long (2 years + 1 week spin-up time) I thought I should update the SST field in both domains.
Thank you for your time,
You still need to stay with sst_update =1, but the SST is from ERA-I you used to drive WRF.
When you ungrib ERA-I, you will have SS in your intermediate file, and metgrid will process SST the same as it processes other variables. After running RREAL with pst_update, you will have the wrflowinp file that includes SST. Then you can run WRF with pst_update.
Please try nd let me know if you have any problems.
I tried to rerun the model using only the ERA-Interim pressure-level and surface data, but unfortunately it crashed once again after about 6 hours of simulation with error message: "WOULD GO OFF TOP: KF_ETA_PARA I,J,DPTHMX,DPMIN". I found the following post discussing this kind of errors:

The model top was set to 50 hPa which is sufficiently high, while I wasn't able to find any CFL errors in the rsl output files. I also checked the wrfinput data, but both TSK and TSLB look reasonable. I am sharing with you a .tar file including the new .rsl output files, and I uploaded the input files (wrfinput_d0* and wrflowinp_d0*) to Google drive because Nextcloud couldn't provide a link:

Maybe there are some inconsistencies in the input files that I cannot detect... or would you suggest another reason why WRF crashes?

Thank you so much,


  • outputs.tar.gz
    87 MB · Views: 25
Your inputs.tar.gz is too large, which makes my computer die and I have to go through lots of troubles to restart this machine.
Now that your case crashed after 6 hours, would you please create a smaller dataset that includes only 12 hours of data ?
I need the following info:
(1) namelist.wps
(2) namelist.input
(3) wrfinput, wrfbdy and wrflowinp (for the first 12 hours)
(4) Let me know which version of WRF you are running.
Hello! I just set up a 12-hour simulation using the version 4.0.1 of WRF. Hopefully this dataset will be easier to handle.
Please find attached a .tar file including all the required inputs. The model crashed again after 6 h with the same error message ("WOULD GO OFF TOP: KF_ETA_PARA I,J,DPTHMX,DPMIN").
Thank you for your help!


  • file2.tar.gz
    189.8 MB · Views: 22
Hi again! I would just like to let you know that I used the Betts-Miller-Janjic cumulus scheme (no2) instead of the Kain-Fritsch (no1) and now the model crashed a few minutes later than with the previous set-up. It failed with a segmentation fault again, but this time the error message is not about the parcel path going off model top. I am attaching the new rsl.* files.


  • rsl.tar.gz
    103 MB · Views: 25
Just an update:
WRF does not crash as long as the grid nesting is one-way (feedback = 0). In my case, the model seems to run with two-way nesting only when the SST field is fixed. I am not sure if this is reasonable though..
I would really appreciate any help on this post!
I tried to repeat your case using ERA-I as the forcing data. I found that the SST field is not consistent with landuse type along the coastal region. Please see the attached plot of SST.

View attachment

This will potentially lead to problem in WRF simulation, especially for long term climate simulation with sst_update.

To solve this issue, please use skintemp to replace SST following the steps below:
(1) in Vtable.ERA-I, delete the line below:
34 | 1 | 0 | | SST | K | Sea-Surface Temperature |
(2) rerun ungrib.exe and metgrid.exe
(3) run real.exe and wrf.exe

I have run a test case and it seems to work fine. Please try and let me know if you still have problem.