Error when running the real.exe

[kwerner]

Artur,

So I think I figured out the problem. When I ran real.exe with your met_em* files, I also got the error, so I then created my own met_em* files with your intermediate files. When I did this, using your exact namelist.wps file, met_em files were created, but real.exe failed again. When I removed the 'process_only_bdy' parameter from the &metgrid namelist, the met_em* files that were created were able to be processed by real.exe. This namelist variable is something that should only be used if no interpolation of data is necessary, and therefore I don't think this option will work with your case. Try turning that off and seeing if that makes a difference.

Kelly

 

[sekluzia]

Dear Kelly,

I have good news The real.exe and wrf.exe work!
I would like to thank you and your team for your great efforts in supporting me to solve this problem.
Hope that it was also useful for you to identify such kind of issue.

Best regards,
Artur

 

[kwerner]

Artur,
That is great news, and a relief! I was out of ideas after those suggestions. I'm very glad to hear that you were able to get past this error. Thank you for letting me know.

Kelly

 

[sekluzia]

Dear Kelly,

I would like to get back to the the issue of failing to run the real.exe with the 'process_only_bdy' parameter option from the &metgrid namelist, if the WRF compiled with the dmpar option. In my 1-3 km simulations forced by ECMWF ~9 km operational model data (using 137 model level data) I found the better performance with 'process_only_bdy' parameter turned on. As you know, I am able to run the real.exe with smpar, but not with dmpar, when the 'process_only_bdy' parameter option is turned on. Have you been able to understand why this occurs? Please note that I have to compile the WRF with the dmpar on the cluster I am using now.

Best regards,
Artur

 

[kwerner]

Hi Artur,
I am trying to get to the bottom of this. The reason it happens for a dmpar run, and not for an smpar fun is because of the way the code is written, but I'm not sure WHY the code is written that way. Our software engineer is out until the week after next, so I'm trying to track down the answer myself. I can tell you that it does seem to be specific to your input data. When I use your namelist with my own input data, I don't get this error. I will keep you posted, and I apologize for the delay.

 

[sekluzia]

Hi Kelly,

Thanks for your reply! You are right, the problem occurs when I use the ECMWF model-level data (at 137 levels), while everything works with the ECMWF pressure level input data (25 pressure levels). Looking forward any updates on this issue! Happy New Year!

Warm regards,
Artur

 

[kwerner]

Artur,
We think this may be a bug that needs to be fixed in the code, as we likely have never tested this with ECMWF model-level data. This will likely take some time, and unfortunately the only thing we can recommend for now is to not use the "process_only_bdy" condition. I apologize for the inconvenience.

 

[sekluzia]

Thanks for your reply! Looking forward for the bug fix!
As far as I understand this can be done not earlier than next WRF release. Am I right?

Best regards,
Artur

 

[kwerner]

Artur,
It just depends on how much time we get to investigate it and figure it out. Some issues are quick to solve, while others take weeks. Unfortunately we are pretty short on resources in our group, and have quite a "to-do" list at the moment. I have marked this as something that needs to be looked into, though. One thing that would help us is if you would like to post to our "Issues" page on our GitHub WRF Repository (https://github.com/wrf-model/WRF/issues). You can mention that you already discussed this via the forum, and then describe the problem specifics. Mention that the problem occurs with the metgrid setting, and then when running real. Once we look into it further, we can assess whether it's actually a WPS or a WRF issue.