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

Run real.exe again for restart run?

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.

userwrfch

Member
Hi,

I have a long-term WRF run from 2000-2020. I have the problem that I have too many files that I cannot have in one folder at the same time on the cluster that I am using (they told me that their parallel file system cannot handle that), and metgrid.exe creates separate files for each input time. I have 6-hourly input data for 20 years (around 30'000 metgrid files in total for one domain, meaning around 90'000 in total for all three domains). It is not possible for me to have all of those 90'000 metgrid files in one folder (or links to those files) in order to run real.exe from 2000-2020. Can you please help me to find a good way to move forward?

I was thinking of, e.g.,
1. run real.exe from 2000-2005,
2. run wrf.exe from 2000-2005, and
3. run real.exe again from 2005-2010 and
4. restart wrf.exe from 2005
5. repeat for 2010-2015, etc.
but I think that this will mess up with the initial conditions. I still want ONE WRF run in total from 2000-2020.

Any help would be greatly appreciated.

Thanks!
 
Hi, userwrfch,
In this case you definitely need to take advantage of the restart capability of WRF. You may run WRF year-by-year, and save the wrfrst file at the end of each year. Then restart from that time to continue for the simulation of the next year.

Detailed description how to implement the restart capability of WRF can be found at
https://www2.mmm.ucar.edu/wrf/users/docs/user_guide_v4/v4.3/users_guide_chap5.html#restart

Please try na let me know if you have any problem.
 
Thank you very much Ming Chen! However, my problem is not a wall time limit for wrf.exe, or other restrictions here, but my problem is that I cannot run real.exe for the whole period (2000-2020) because there are too many metgrid files that I cannot link to one folder. As far as I understood from the documentation, the restart run works such that real.exe is run ONCE for the whole time and produces wrfinput* and wrfbdy* files. Then ONLY the wrf.exe is run again with restart options enabled from the time of the last wrfrst* files, e.g. after reaching the wall time limit. However, as mentioned, this procedure is not possible for me.

My question is how I can overcome the issue of not being able to run real.exe for the whole time period. I thought about running real.exe first for a shorter time period (e.g. 2000-2005) and then running wrf.exe and get wrfrst* files at the end of 2005, and then running real.exe again for 2005-2010 before restarting wrf.exe from 2005, but I actually don't think that's possible because wrf.exe with restart option enabled takes only the wrfrst* files into account and not the wrfinput* and wrfbdy* (please correct me if I am wrong). Is there another way to overcome this problem?

Thank you very much!
 
Hi, userwrfch,
You don't need to run real.exe just once to cover the entire period of simulation. Instead, you can run real.exe, say over 1-year, then run it for another year, and anther year. Of course you need to save wrfbdy_d01 for different years, e.g., wrfbdy_d01.year1, wrfbdy_d01.year2, etc.
When you run wrf over one specific year, you can copy wrfbdy_d01 of that year to your working directly, rename it as wrfbdy_d01.
Hope this is helpful. Let me know if you still feel confused.
 
Thank you very much! This is really helpful. I did a small test run and it worked!

I need to save both wrfbdy_d01 AND wrfinput* for each year (not only wrfbdy_d01), right?

I will do a physics ensemble, where I will run wrf.exe several times with several different physics schemes. Can I use the same wrfbdy_01 and wrfinput* files here (i.e., only wrf. exe starts to use information on different physics schemes), or do I need to run them for each physics scheme separately (i.e., real.exe already prepares the data for each physics parameterization separately)?

Thanks again!
 
Hi,
Please see my answers below:

(1) I need to save both wrfbdy_d01 AND wrfinput* for each year (not only wrfbdy_d01), right?

No. You only need to save wrfbdy files for each year. wrfinput at the initial time is needed. For restart run, this file is no longer needed.

(2) I will do a physics ensemble, where I will run wrf.exe several times with several different physics schemes. Can I use the same wrfbdy_01 and wrfinput* files here (i.e., only wrf. exe starts to use information on different physics schemes), or do I need to run them for each physics scheme separately (i.e., real.exe already prepares the data for each physics parameterization separately)?

If you change surface scheme, you need to rerun REAL to produce wrfinput. But you can use the same wrfbdy files no matter what physics schemes you will use.
 
Hi Ming Chen,

I tried the procedure that you described above and it works well. However, I need SST update (6-hourly), and once I include that, it does not work anymore. As described above, I did a wrf.exe run for one year with SST update, then reran real.exe for the next year and included the new wrfbdy_d01 file. I just changed the dates in namelist.input and set restart=.true. The tail of the error is:

d01 2001-01-01_00:00:00 Input data is acceptable to use: wrfrst_d02_2001-01-01_00:00:00
Timing for processing restart file for domain 2: 5.40389 elapsed seconds
SOIL TEXTURE CLASSIFICATION = STAS FOUND 19 CATEGORIES
d01 2001-01-01_00:00:00 Error trying to read metadata
d01 2001-01-01_00:00:00 File name that is causing troubles = wrfrst_d02_2001-01-01_00:00:00
d01 2001-01-01_00:00:00 You can try 1) ensure that the input file was created with WRF v4 pre-processors, or
d01 2001-01-01_00:00:00 2) use force_use_old_data=T in the time_control record of the namelist.input file
-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 332
---- ERROR: The input file appears to be from a pre-v4 version of WRF initialization routines
-------------------------------------------


---------
You mentioned that I only need to save wrfbdy files for each year, and wrfinput at the initial time is needed. For restart run, this file is no longer needed. However, do I also need to save the wrflowinp files for each year if I use SST update (apparently it does not seem to be included in the wrfrst files)? How do I solve that error?

Thanks!
 
Hi,
When you run real.exe, if set_update is activated, you will produce wrfbdy and also wrflowinput files. Once set_update = 1, you need to save both wrfbdy and wrflowinput for the restart run.
 
Top