WRF & WPS seasonal forecasting

[nyundo]

Yes please. i still need help. I have been keeping myself busy with simple real time simulations and currently stuck at data processing. i am failing to understand how RAINC and RAINNC work. soo playing through ncl and python codes and its challenging

For forecasters/observers to understand in a simple language RAINC is rainfall from convective clouds while RAINNC is from non-convective clouds. To be able to get total rainfall as per your WRF model run just add those two i.e, rainfall=RAINC+RAINNC

 

[kcmonaka]

Thank you for you response. i will give feedback soon.. Many thanks

 

[kcmonaka]

Below is the content folder, i have an issue with running metgrid.exe
Also the vtables that you mentioned i do not have them, i only have the generic vtable (vtable.CFSR)

 

[nyundo]

Below is the content folder, i have an issue with running metgrid.exe
Also the vtables that you mentioned i do not have them, i only have the generic vtable (vtable.CFSR)

Without correct Vtables don't run metgrid, it will fails. Run metgrid after you have run ungrib for both data successfully.

Correct, in some of the new versions of WPS (I don't remember exactly from which version of WPS) those mentioned Vtables are not included, but you can still use the ones from the early versions of WPS. Kindly find the attached two Vtables I am using.

 

[cyantan]

Alright!

You need to do the following:

1. Download CFSV2 data for your season of interest. I normally do a 3 months and 6 months model runs for seasonal outlook applications.

These data are stored separately for surface data (flx) and variables at pressure levels (pgb).

It is recommended to download them into two separate directories like PRESS for variables at pressure levels data and FLX for variables at the surface.

Here is the link to those data Index of /pub/data/nccf/com/cfs/prod

I normally use 6 hourly grib1 files


2. You have different data so you should ungrib those data twice.

Here are the lines in the namelist.wps which you should have

&ungrib
out_format = 'WPS',
!prefix = 'FLX',
!prefix = 'PRES',

Here both have been commented by !

&metgrid
io_form_metgrid = 2,
fg_name = 'FLX','PRES',



Step 1: Ungrib any data say we start with FLX. This is normaly very fast because we have less variables

First, un-comment by removing the ! in the &ungrib for FLX, section and also link the correct Vtable

ln -sf Variable_Tables/Vtable.CFSR_sfc_flxf06 Vtable

Remember to link the correct data with ./link_grib.csh PATH TO YOUR FLX DATA

Run ungrib

Step 2: Run ungrib for PRES data, normally it takes some minutes depending on the length of the season and your domain size

Comment the line for FLX and un-comment the one for PRES in the section of &ungrib

link the correct Vtable for data at pressure levels

ln -sf Variable_Tables/Vtable.CFSR_press_pgbh06 Vtable

Remember to link the correct data with ./link_grib.csh PATH TO YOUR PRES DATA

Run ungrib

After step 1 and 2 have been successfully completed now proceed with metgrid and wrf part as you normally do.

I suggest you to first do it step by step and later you can automate the process after all the steps have been clear to you.

Drop a message if these steps work for you or incase of any challenges and you need more clarification

Thank you very much, I want to use the operational CFSv2 data to force WRF V4.4.1 to do subseasonal-to-seasonal dynamic downscale. When i was searching for how to do this, I saw your post. These instructions are very detailed and helpful. I just got my first dynamic downscale product with your idea!

 

[kcmonaka]

Thank you very much, I want to use the operational CFSv2 data to force WRF V4.4.1 to do subseasonal-to-seasonal dynamic downscale. When i was searching for how to do this, I saw your post. These instructions are very detailed and helpful. I just got my first dynamic downscale product with your idea!

how did you manage?? i tried following the steps but it failed. where did you get the Vtables? please share the namelists for your domain. Thank you in advance.

 

[nyundo]

how did you manage?? i tried following the steps but it failed. where did you get the Vtables? please share the namelists for your domain. Thank you in advance.

I have shared the Vtables in the earlier message. Kindly download those and try to run the model for say 10 days in advance. If you find any difficulties drop a message.

Alternatively share your email address and I can send them to you directly.

 

[nyundo]

Thank you very much, I want to use the operational CFSv2 data to force WRF V4.4.1 to do subseasonal-to-seasonal dynamic downscale. When i was searching for how to do this, I saw your post. These instructions are very detailed and helpful. I just got my first dynamic downscale product with your idea!

Happy to hear that these instructions were useful to you

 

[JOUMA]

Hi Nyundo,
Kindly guide on how to set up the namelist.input for this seasonal forecast using CFSV2. I tried turning on sst_update =1, it did not work.

 

[dkaustubh]

@nyundo I am facing similar challenges with running WRF with CFSv2 data. Your reply gave me some idea, however I have a few doubts. Appreciate if you could clarify.
1. CFSv2 has different grib data called 6hrly_grib_01/02/03/04 for the same date and initialization. Are these data different or the same.
2. The Vtables that you have mentioned here: Vtable.CFSR_press_pgbh06 and Vtable.CFSR_sfc_flxf06 is not available by default in my WRF installation. Have you created it on your own by using Vtable.CFSR? Can you share these Vtables here if possible?

 

[MUWRFDA]

Hello, I hope i can reach someone who can help on this forum. I am trying to run WRF for seasonal simulation. I have started with a 1-month trial run. following the guide above, I downloaded flux and pressure level data, run ungrib.exe for both of them and then metgrid.exe. Both ungrib and megrid inidcated successful runs. Coming to real.exe, I get the error "d01 2024-11-01_00:00:00 No average surface temperature for use with inland lakes" note that I had to remove " sst_update = 1".
Please help

 

[cyantan]

@nyundo I am facing similar challenges with running WRF with CFSv2 data. Your reply gave me some idea, however I have a few doubts. Appreciate if you could clarify.
1. CFSv2 has different grib data called 6hrly_grib_01/02/03/04 for the same date and initialization. Are these data different or the same.
2. The Vtables that you have mentioned here: Vtable.CFSR_press_pgbh06 and Vtable.CFSR_sfc_flxf06 is not available by default in my WRF installation. Have you created it on your own by using Vtable.CFSR? Can you share these Vtables here if possible?

Hi dkaustubh,

Based on my knowledge, for your first question, I believe there are 4 ensemble member at each initial time step (e.g., 2025-05-13_00Z), the "6hrly_grib_01" is the control run, and the "6hrly_grib_02/03/04" are three pertubed runs. You may notice that they have different forecast valid days.

For your second question, @nyundo have already shared the two Vtables. I just attach them here for your convenience.