add support for WRFV4 about wrf-suews HOT 34 CLOSED

@hamidrezaomidvar once you go through the coupling guide, let's discuss how we use the latest WRF v4 for the coupling.

from wrf-suews.

Ting, I wrote a script for the automation process, and pushed my branch to the Github. In the branch, there is a folder called SUEWS-automate. I basically, wrote a general and short script that reads the necessary changes from a Jason format file, and makes the changes. There is a very simple Makefile in the folder, which run the scripts and copy the necessary files there. right now, WRF folder needs to be in the same directory as the scripts. We can definitely make the scripts and processes more sophisticated and better. I think it is good start.

from wrf-suews.

source for the open code of WRF: [email protected]:wrf-model/WRF_OPEN.git

from wrf-suews.

@zhenkunl please start to work on this soon; make sure to use the latest SUEWS code.

from wrf-suews.

I think there are some more changes here in phys/module_surface_driver.F that is not mentioned in the wiki page.

CASE (SUEWSSCHEME) ! add for SUEWS by Hamid Omidvar, 11/3/2019

   IF (PRESENT(qv_curr)                            .AND.    &
       PRESENT(capg)        .AND.    &
                                                  .TRUE. ) THEN
       DO j=j_start(ij),j_end(ij)
       DO i=i_start(ij),i_end(ij)
 !          CQS2 ACCOUNTS FOR MAVAIL FOR SFCDIAGS 2M Q
          CQS2(I,J)= CQS2(I,J)*MAVAIL(I,J)
       ENDDO
       ENDDO

       IF ( FRACTIONAL_SEAICE == 1 ) THEN
          CALL wrf_error_fatal('SUEWS scheme cannot be used with fractional seaice')
       ENDIF

  WRITE( message,* ) ' calling suewsdrv',sf_surface_physics
  call wrf_debug(100, message)


      !CALL SLAB(t_phy,qv_curr,p_phy,flhc,flqc,  &
      !   psfc,xland,tmn,hfx,qfx,lh,tsk,qsfc,chklowq,          &
      !   gsw,glw,capg,thc,snowc,emiss,mavail,                 &
      !   dtbl,rcp,xlv,dtmin,ifsnow,                           &
      !   svp1,svp2,svp3,svpt0,ep_2,karman,eomeg,stbolt,       &
      !   tslb,zs,dzs,num_soil_layers,radiation,               &
      !   p1000mb,                                             &
      !   ids,ide, jds,jde, kds,kde,                           &
      !   ims,ime, jms,jme, kms,kme,                           &
      !   i_start(ij),i_end(ij), j_start(ij),j_end(ij), kts,kte)

from wrf-suews.

Great!
Please get them in!

from wrf-suews.

Another step was missing here
I added this:
module_sf_suews.o should be added in the list of Modules in the phys\Makefile

from wrf-suews.

from wrf-suews.

Ting, I was finally able to compile WRF4-SUEWS coupled. There were several steps missing in the WIKI which I have already added. Now I am trying to run a dataset for the coupled version. Do you have any dataset I can use?
is the running of the coupled version similar to the regular version?
The potential files needed are: wrfbdy, wrfinput, namelist.input and namelist.suews(? do we need this in the same folder of the run?) do we need any other file?

from wrf-suews.

Great!
Please see files in /work/scratch/WRF-SUEWS-test/wrf-input:

1. wrfinput_d0x: the wrf input files with domains centred in London.
2. wrfbdy_d01: boundary condition file.
3. namelist.suews: it has the SUEWS specific settings. For anything unclear, let's start another issue and reframe it into a wiki page.
4. add-SUEWS-input.py: the script is used for adding SUEWS-specific variables into ordinary wrfinput files; this should be self-explanatory.
5. namelist.input: the ordinary namelist.input with SUEWS-related additions.

from wrf-suews.

Quick question. The current coupled system is just for serial run right?

from wrf-suews.

No, we can do one-way nested parallel run.

from wrf-suews.

Once you compile a nested version, you use the script /work/scratch/WRF-SUEWS-test/wrf-input to perform parallel runs.

from wrf-suews.

But the script is for real.exe, while we already have wrfbdy and wrfinput files. right? I tried running in parallel, and it seems it is not compatible and gives me the following error although even with np=1:

For domain 3 , the domain size is too small for this many processors, or the decomposition aspect ratio is poor.
Minimum decomposed computational patch size, either x-dir or y-dir, is 10 grid cells.

from wrf-suews.

from wrf-suews.

See this one: /work/scratch/WRF-SUEWS-test/wrf-input/London-3d-201507

from wrf-suews.

Ok thanks. I will run it and let you know how it goes

from wrf-suews.

Once you got it running again, post some plots of results here.
Then we work in the repo to finish the automation scripts.

And I am aware there are several hardwired settings in the current coupling code are NOT appropriate, let's change those as well when get the automation done: I expect this automation can make our next step a lot easier.

from wrf-suews.

I am getting this error when I run the case:
--- ERROR: Unknown sf_surface_physics has no associated number of soil levels -------------- FATAL CALLED --------------- FATAL CALLED FROM FILE: <stdin> LINE: 2115 --- ERROR: sf_surface_physics = 9
Have you encounter with this? I suspect I am doing something wrong in setting on the coupled version

from wrf-suews.

from wrf-suews.

Go to: /work/scratch/WRF-SUEWS-test/hamid/WRFV4.0-SUEWS-coupled/test/em_real

I will do the same with V3 as we already have to see if it is the problem of V4 or on my side!

from wrf-suews.

So it seems there is no problem with V3. either it is a new thing in V4, or I am missing something when doing changes for V4

from wrf-suews.

It's good that you've identified the issue. Have you tried to increase the debug level in v4?

from wrf-suews.

Yes, but no more info!
But I suspect it is because of this update for V4:

Basically, they removed num_soil_layer from namelist.input so the model itself should have it right??

from wrf-suews.

http://www2.mmm.ucar.edu/wrf/users/wrfv4.0/updates-4.0.html

from wrf-suews.

But I think it is just for the template!

from wrf-suews.

See the code here:
https://github.com/wrf-model/WRF/blob/9527846b8782cf1a137913daeb6ed267e8b2f5ca/share/module_check_a_mundo.F#L2240-L2272

Please note that we used SLABSCHEME as the basis for the coupling; so wherever SLABSCHEME appears we should add an SUEWSSCHEME-related option alike.

from wrf-suews.

nice catch! Already in V3, but missing from the steps! I will update it and follow the same way to fix everything!

from wrf-suews.

So it seems we really need some automated way to do this; otherwise we are missing pieces here and there.

A good starting point might be to script something for checking the co-existence of SLABSCHEME and SUEWSSCHEME.

from wrf-suews.

I agree. The V4 coupled model is running now. I am running it for 1 days, so when it finishes, I will look at the results and put it here.

from wrf-suews.

Good.
I'm in my office today and we can have discussion as you need.

from wrf-suews.

Fantastic!
Seems not in yet.
But I'll have a look later on.

BTW, how are the previous results looking like?

from wrf-suews.

I am working on the results today, and we'll let you know!

from wrf-suews.

now I think we can close this one and start a new thread for examination of the results.

from wrf-suews.