Help Needed – WRF Setup Using ECMWF IFS Data (Ungrib Failure)

[Irsyad]

Hello everyone,
I’m new to WRF. I’m currently working on a comparative study using both GFS and ECMWF IFS datasets to simulate weather using the WRF model.

I’ve successfully run the model using GFS data, but I’m now facing difficulties when trying to use the ECMWF IFS data downloaded from rda.ucar.edu. I understand that ECMWF datasets contain many parameters, and I’m unsure which variables are actually required for WRF to run properly.

I have already tried using Vtable.ECMWF in the ungrib step, but the process always fails. I suspect it might be due to missing or unnecessary variables in the downloaded dataset.

My question is:

What specific ECMWF IFS parameters (fields) are required to be downloaded from rda.ucar.edu in order to successfully run WRF simulations?

Any advice, guidance, or example configurations would be greatly appreciated. Thank you very much in advance

 

[Ming Chen]

Hi,

We have attempted to address the issues related to IFS data, which is relatively new to us. Please see various solutions proposed at

New ERA5 Vtable by JoshLovesFun · Pull Request #265 · wrf-model/WPS

I have made a new ERA5 Vtable using the Vtable.ECMWF as a template. I have not made any changes that should make this Vtable.ERA5 function differently from the current Vtable.ECMWF. I have just mad...

github.com

Unfortunately IFS49r1 data are archived by various agencies and these archived data could be different, making it a complicated issue for us to address its usage in WPS/WRF.

We don't have an official document how to use it and the available solutions proposed in the above website don't work yet.

Under this situation, I wonder whether you would like to switch to ERA5?

 

[Irsyad]

Thank you for your message. I understand the concern regarding the complexity and inconsistency of IFS49r1 data from various archives.

For my study, I have been using IFS-HRES data downloaded from the NCAR RDA website, specifically the isobaric analysis and surface analysis fields. In addition, I have also contacted the official meteorological agency in Indonesia (BMKG) to request access to the same data for my case study on 27–28 October 2024.

However, I have been encountering issues when trying to process this data with WPS/WRF, especially during the ungrib.exe stage. Unfortunately, I haven’t been able to find clear documentation or a working tutorial on how to correctly preprocess IFS data for WRF.

That being said, if you have any references, examples, or tutorials (even unofficial) regarding how to use IFS data with WRF, I would greatly appreciate it if you could share them. It would be a great help to move the work forward.

Thank you in advance for your support

 

[Ming Chen]

In November 2024, ECMWF released version 49r1 of the operational IFS. This new released data broke the ECMWF Vtable that had been released in the 4.6.0 of WPS. We have tried to fix the issue but didn't make it yet. Please see

update era5/ifs soil field processing by jimbresch · Pull Request #266 · wrf-model/WPS

Modifications to rd_grib2.F to allow decoding of the era5/ifs soil fields. (referenced in #264 and #265). . Note that EC uses a local table for the grib codes and that g2print is not able to proces...

github.com

Please switch to other EC data at present ....

 

[Irsyad]

Thank you very much for your kind help and the valuable information.

I would like to kindly confirm, does it mean that running a weather simulation using ECMWF data from before November 2024 is still supposed to work properly with WPS?
I have been using data from October 2024, prior to the release of IFS 49r1, but I am still encountering issues during the ungrib process.

Additionally, I would like to ask:
Does the IFS 49r1 update also affect ERA5 reanalysis data? Or is it still possible to run WRF simulations smoothly using ERA5 data instead?

I would greatly appreciate any clarification you could offer on this matter.

 

[scottlynn73]

hi, our group is using the IFS (and AI variant thereof) in similar comparisons so it definitely works- as I recall we rebuilt WPS with the source code changes needed for ungrib, and used a new Vtable shared elsewhere in this forum. I'm not on the forums very much but my email is in my signature if you have any questions

 

[Ming Chen]

@scottlynn73
Hi Scott,
Thank you very much for the kind information. Would you please clarify:
(1) where did you download IFS 49r1 data?
(2) which version of WPS do you use to process the data? Can you share with us the WPS codes you have changed?
(3) I suppose you also modified Vtable. If so, can you share it with us?
Thanks again !

 

[Ming Chen]

@ Irsyad

Would you please run WPSV4.7/WRFV4.7 using ECMWF IFS data as input, and modify line 20 of ungrib/Variable_Tables/Vtable.ECMWF.IFS.grib2 as:

235 | 1 | 0 | | SKINTEMP | K | Skin Temperature | 0 | 0 | 17 | 1 |

Then use this vtable to ungrib IFS data.

Please let me know whether it works. Thanks !

 

[Irsyad]

Alright, I will try updating to WRF and WPS version 4.7. However, please note that the IFS data I downloaded from NCAR RDA Dataset d113000 as well as the data I received from the official institution are all in GRIB1 format.

Is there any special treatment required for the data to be converted to GRIB2, or can it still be used as is, even though it's in GRIB1 format?
I would really appreciate your guidance on this

 

[Ming Chen]

HI, WPS can process both GRIB1 and GRIB2 data. Therefore, it is ok to stay with IFS data in GRIB1 format.

 

[Irsyad]

Alright, I will try updating to WRF and WPS version 4.7. However, please note that the IFS data I downloaded from NCAR RDA Dataset d113000 as well as the data I received from the official institution are all in GRIB1 format.

Is there any special treatment required for the data to be converted to GRIB2, or can it still be used as is, even though it's in GRIB1 format?
I would really appreciate your guidance on this

Also, I’ve already updated WRF to v4.7.1, but I can’t find WPS v4.7 on GitHub. Are there alternative sources to download it?

 

[haiqingsong]

Also, I’ve already updated WRF to v4.7.1, but I can’t find WPS v4.7 on GitHub. Are there alternative sources to download it?

WPS4.7 has not been released yet.

 

[Ming Chen]

WPSv4.6 should work. Please try.

 

[Irsyad]

I have installed WRF 4.7.1 with WP 4.6, then modified the Vtable as you instructed, but I am still having trouble with the ungrib process and am getting the attached error.
Please help.
*For additional information, I downloaded data from the website NCAR RDA Dataset d113001 using surface analysis and isobaric analysis.
thanks

 

[Irsyad]

HI, WPS can process both GRIB1 and GRIB2 data. Therefore, it is ok to stay with IFS data in GRIB1 format.

 

[Ming Chen]

I just didi a quick test and got the same error as you did. Apparently we have some issues with the IFS data and no solution is available yet...