Remove snet from Noah LSM #669

climbfuji · 2021-05-28T13:07:10Z

Description

In PR, #627 a commit was made to remove snet (net solar radiation is computed from the incoming SW and albedo), but only from sfc_drv_ruc.F90. sfc_drv.f should also be modified for Noah LSM.

Solution

Remove snet from Noah LSM (check if it can be removed altogether) and compute it there from the radiation output.

Alternatives (optional)

No alternatives.

Related to (optional)

#627

The text was updated successfully, but these errors were encountered:

HelinWei-NOAA · 2021-06-01T13:18:48Z

I modified radiation_surface.f, so the four albedo components over land will be outputted to GFS_radiation_surface.F90 with ialb=1 and then passed to lsm_drv.f. I added the following lines to radiation_surface.f:

        lsmalbdnir(i) = min(0.99,max(0.01,alndnb))
        lsmalbinir(i) = min(0.99,max(0.01,alndnd))
        lsmalbdvis(i) = min(0.99,max(0.01,alndvb))
        lsmalbivis(i) = min(0.99,max(0.01,alndvd))

and changed these four to intent(inout). But the model always crashed on these lines with the error (forrtl: severe (174): SIGSEGV, segmentation fault occurred). Dom, do you have any idea? Thanks.

climbfuji · 2021-06-01T13:31:54Z

These arrays may not be allocated for Noah LSM, can you check GFS_typedefs.F90?

HelinWei-NOAA · 2021-06-01T13:49:41Z

You are absolutely right. Nice catch. Thanks.

HelinWei-NOAA · 2021-06-01T14:06:15Z

So all three land surface models have these four common variables. Should I move them together? This will require more changes. Maybe we can do them later. For simplicity's sake, I can redefine them for Noah LSM. What is your suggestion?

climbfuji · 2021-06-01T14:14:51Z

How about changing the following line in GFS_typedefs.F90? if (Model%lsm == Model%lsm_ruc .or. Model%lsm == Model%lsm_noahmp) then To: if (Model%lsm == Model%lsm_noah .or. Model%lsm == Model%lsm_ruc .or. Model%lsm == Model%lsm_noahmp ) then We may need to check if those additional fields also need to be written to the restart files for Noah LSM.

…

On Jun 1, 2021, at 7:19 AM, HelinWei-NOAA ***@***.***> wrote: I modified radiation_surface.f, so the four albedo components over land will be outputted to GFS_radiation_surface.F90 with ialb=1 and then passed to lsm_drv.f. I added the following lines to radiation_surface.f: lsmalbdnir(i) = min(0.99,max(0.01,alndnb)) lsmalbinir(i) = min(0.99,max(0.01,alndnd)) lsmalbdvis(i) = min(0.99,max(0.01,alndvb)) lsmalbivis(i) = min(0.99,max(0.01,alndvd)) and changed these four to intent(inout). But the model always crashed on these lines with the error (forrtl: severe (174): SIGSEGV, segmentation fault occurred). Dom, do you have any idea? Thanks. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#669 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB5C2RKHDBL5BTFYJ4BS25DTQTM4RANCNFSM45WS5FQQ>.

HelinWei-NOAA · 2021-06-01T14:22:35Z

I don't think that's enough. Maybe we can move these four variables here if (Model%lsm == Model%lsm_noah .or. Model%lsm == Model%lsm_noahmp .or. Model%lsm == Model%lsm_noah_wrfv4) then temp2d(i,j,35) = GFS_Data(nb)%Sfcprop%slc(ix,1) temp2d(i,j,36) = GFS_Data(nb)%Sfcprop%slc(ix,2) temp2d(i,j,37) = GFS_Data(nb)%Sfcprop%slc(ix,3) temp2d(i,j,38) = GFS_Data(nb)%Sfcprop%slc(ix,4) temp2d(i,j,39) = GFS_Data(nb)%Sfcprop%smc(ix,1) temp2d(i,j,40) = GFS_Data(nb)%Sfcprop%smc(ix,2) temp2d(i,j,41) = GFS_Data(nb)%Sfcprop%smc(ix,3) temp2d(i,j,42) = GFS_Data(nb)%Sfcprop%smc(ix,4) temp2d(i,j,43) = GFS_Data(nb)%Sfcprop%stc(ix,1) temp2d(i,j,44) = GFS_Data(nb)%Sfcprop%stc(ix,2) temp2d(i,j,45) = GFS_Data(nb)%Sfcprop%stc(ix,3) temp2d(i,j,46) = GFS_Data(nb)%Sfcprop%stc(ix,4) On Tue, Jun 1, 2021 at 10:15 AM Dom Heinzeller ***@***.***> wrote:

…

How about changing the following line in GFS_typedefs.F90? if (Model%lsm == Model%lsm_ruc .or. Model%lsm == Model%lsm_noahmp) then To: if (Model%lsm == Model%lsm_noah .or. Model%lsm == Model%lsm_ruc .or. Model%lsm == Model%lsm_noahmp ) then We may need to check if those additional fields also need to be written to the restart files for Noah LSM. > On Jun 1, 2021, at 7:19 AM, HelinWei-NOAA ***@***.***> wrote: > > > I modified radiation_surface.f, so the four albedo components over land will be outputted to GFS_radiation_surface.F90 with ialb=1 and then passed to lsm_drv.f. I added the following lines to radiation_surface.f: > > lsmalbdnir(i) = min(0.99,max(0.01,alndnb)) > lsmalbinir(i) = min(0.99,max(0.01,alndnd)) > lsmalbdvis(i) = min(0.99,max(0.01,alndvb)) > lsmalbivis(i) = min(0.99,max(0.01,alndvd)) > and changed these four to intent(inout). But the model always crashed on these lines with the error (forrtl: severe (174): SIGSEGV, segmentation fault occurred). Dom, do you have any idea? Thanks. > > — > You are receiving this because you authored the thread. > Reply to this email directly, view it on GitHub < #669 (comment)>, or unsubscribe < https://github.com/notifications/unsubscribe-auth/AB5C2RKHDBL5BTFYJ4BS25DTQTM4RANCNFSM45WS5FQQ >. > — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub <#669 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ALPHKYGDYWASNTCQLZ6BGW3TQTTOZANCNFSM45WS5FQQ> .

climbfuji · 2021-06-01T14:26:49Z

This code is just for calculating checksums. Can/should be updated, but if you don't it won't affect the runs.

…

On Jun 1, 2021, at 8:22 AM, HelinWei-NOAA ***@***.***> wrote: I don't think that's enough. Maybe we can move these four variables here if (Model%lsm == Model%lsm_noah .or. Model%lsm == Model%lsm_noahmp .or. Model%lsm == Model%lsm_noah_wrfv4) then temp2d(i,j,35) = GFS_Data(nb)%Sfcprop%slc(ix,1) temp2d(i,j,36) = GFS_Data(nb)%Sfcprop%slc(ix,2) temp2d(i,j,37) = GFS_Data(nb)%Sfcprop%slc(ix,3) temp2d(i,j,38) = GFS_Data(nb)%Sfcprop%slc(ix,4) temp2d(i,j,39) = GFS_Data(nb)%Sfcprop%smc(ix,1) temp2d(i,j,40) = GFS_Data(nb)%Sfcprop%smc(ix,2) temp2d(i,j,41) = GFS_Data(nb)%Sfcprop%smc(ix,3) temp2d(i,j,42) = GFS_Data(nb)%Sfcprop%smc(ix,4) temp2d(i,j,43) = GFS_Data(nb)%Sfcprop%stc(ix,1) temp2d(i,j,44) = GFS_Data(nb)%Sfcprop%stc(ix,2) temp2d(i,j,45) = GFS_Data(nb)%Sfcprop%stc(ix,3) temp2d(i,j,46) = GFS_Data(nb)%Sfcprop%stc(ix,4) On Tue, Jun 1, 2021 at 10:15 AM Dom Heinzeller ***@***.***> wrote: > How about changing the following line in GFS_typedefs.F90? > > if (Model%lsm == Model%lsm_ruc .or. Model%lsm == Model%lsm_noahmp) then > > To: > > if (Model%lsm == Model%lsm_noah .or. Model%lsm == Model%lsm_ruc .or. > Model%lsm == Model%lsm_noahmp ) then > > We may need to check if those additional fields also need to be written to > the restart files for Noah LSM. > > > On Jun 1, 2021, at 7:19 AM, HelinWei-NOAA ***@***.***> wrote: > > > > > > I modified radiation_surface.f, so the four albedo components over land > will be outputted to GFS_radiation_surface.F90 with ialb=1 and then passed > to lsm_drv.f. I added the following lines to radiation_surface.f: > > > > lsmalbdnir(i) = min(0.99,max(0.01,alndnb)) > > lsmalbinir(i) = min(0.99,max(0.01,alndnd)) > > lsmalbdvis(i) = min(0.99,max(0.01,alndvb)) > > lsmalbivis(i) = min(0.99,max(0.01,alndvd)) > > and changed these four to intent(inout). But the model always crashed on > these lines with the error (forrtl: severe (174): SIGSEGV, segmentation > fault occurred). Dom, do you have any idea? Thanks. > > > > — > > You are receiving this because you authored the thread. > > Reply to this email directly, view it on GitHub < > #669 (comment)>, > or unsubscribe < > https://github.com/notifications/unsubscribe-auth/AB5C2RKHDBL5BTFYJ4BS25DTQTM4RANCNFSM45WS5FQQ > >. > > > > — > You are receiving this because you were assigned. > Reply to this email directly, view it on GitHub > <#669 (comment)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/ALPHKYGDYWASNTCQLZ6BGW3TQTTOZANCNFSM45WS5FQQ> > . > — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#669 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB5C2RL4UNR3SSNOD2ADRIDTQTULXANCNFSM45WS5FQQ>.

HelinWei-NOAA · 2021-06-01T17:27:14Z

After modifying FV3GFS_io.F90 to include them to restart files, now I got FATAL from PE 153: fms_io(restore_state_all): unable to find mandatory variable albdvis_lnd in restart file sfc_data.tile5.nc This is already an coldstart run, anyway to overcome this problem? On Tue, Jun 1, 2021 at 10:27 AM Dom Heinzeller ***@***.***> wrote:

…

This code is just for calculating checksums. Can/should be updated, but if you don't it won't affect the runs. > On Jun 1, 2021, at 8:22 AM, HelinWei-NOAA ***@***.***> wrote: > > > I don't think that's enough. Maybe we can move these four variables here > > if (Model%lsm == Model%lsm_noah .or. Model%lsm == Model%lsm_noahmp > .or. Model%lsm == Model%lsm_noah_wrfv4) then > temp2d(i,j,35) = GFS_Data(nb)%Sfcprop%slc(ix,1) > temp2d(i,j,36) = GFS_Data(nb)%Sfcprop%slc(ix,2) > temp2d(i,j,37) = GFS_Data(nb)%Sfcprop%slc(ix,3) > temp2d(i,j,38) = GFS_Data(nb)%Sfcprop%slc(ix,4) > temp2d(i,j,39) = GFS_Data(nb)%Sfcprop%smc(ix,1) > temp2d(i,j,40) = GFS_Data(nb)%Sfcprop%smc(ix,2) > temp2d(i,j,41) = GFS_Data(nb)%Sfcprop%smc(ix,3) > temp2d(i,j,42) = GFS_Data(nb)%Sfcprop%smc(ix,4) > temp2d(i,j,43) = GFS_Data(nb)%Sfcprop%stc(ix,1) > temp2d(i,j,44) = GFS_Data(nb)%Sfcprop%stc(ix,2) > temp2d(i,j,45) = GFS_Data(nb)%Sfcprop%stc(ix,3) > temp2d(i,j,46) = GFS_Data(nb)%Sfcprop%stc(ix,4) > > On Tue, Jun 1, 2021 at 10:15 AM Dom Heinzeller ***@***.***> > wrote: > > > How about changing the following line in GFS_typedefs.F90? > > > > if (Model%lsm == Model%lsm_ruc .or. Model%lsm == Model%lsm_noahmp) then > > > > To: > > > > if (Model%lsm == Model%lsm_noah .or. Model%lsm == Model%lsm_ruc .or. > > Model%lsm == Model%lsm_noahmp ) then > > > > We may need to check if those additional fields also need to be written to > > the restart files for Noah LSM. > > > > > On Jun 1, 2021, at 7:19 AM, HelinWei-NOAA ***@***.***> wrote: > > > > > > > > > I modified radiation_surface.f, so the four albedo components over land > > will be outputted to GFS_radiation_surface.F90 with ialb=1 and then passed > > to lsm_drv.f. I added the following lines to radiation_surface.f: > > > > > > lsmalbdnir(i) = min(0.99,max(0.01,alndnb)) > > > lsmalbinir(i) = min(0.99,max(0.01,alndnd)) > > > lsmalbdvis(i) = min(0.99,max(0.01,alndvb)) > > > lsmalbivis(i) = min(0.99,max(0.01,alndvd)) > > > and changed these four to intent(inout). But the model always crashed on > > these lines with the error (forrtl: severe (174): SIGSEGV, segmentation > > fault occurred). Dom, do you have any idea? Thanks. > > > > > > — > > > You are receiving this because you authored the thread. > > > Reply to this email directly, view it on GitHub < > > #669 (comment) >, > > or unsubscribe < > > https://github.com/notifications/unsubscribe-auth/AB5C2RKHDBL5BTFYJ4BS25DTQTM4RANCNFSM45WS5FQQ > > >. > > > > > > > — > > You are receiving this because you were assigned. > > Reply to this email directly, view it on GitHub > > < #669 (comment)>, > > or unsubscribe > > < https://github.com/notifications/unsubscribe-auth/ALPHKYGDYWASNTCQLZ6BGW3TQTTOZANCNFSM45WS5FQQ > > > . > > > — > You are receiving this because you authored the thread. > Reply to this email directly, view it on GitHub < #669 (comment)>, or unsubscribe < https://github.com/notifications/unsubscribe-auth/AB5C2RL4UNR3SSNOD2ADRIDTQTULXANCNFSM45WS5FQQ >. > — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub <#669 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ALPHKYFYTQPPWBLOU6DQKJDTQTU3XANCNFSM45WS5FQQ> .

HelinWei-NOAA · 2021-06-01T17:56:29Z

I am trying to make them optional for restart to see if it is working.

HelinWei-NOAA · 2021-06-01T18:59:52Z

With the change shown below, it is still not working and I got the same error (FATAL from PE 86: fms_io(restore_state_all): unable to find mandatory variable albivis_lnd in restart file sfc_data.tile3.nc)

!--- register the 2D fields
do num = 1,nvar_s2m
var2_p => sfc_var2(:,:,num)
if (trim(sfc_name2(num)) == 'sncovr'.or. trim(sfc_name2(num)) == 'tsfcl' .or. trim(sfc_name2(num)) == 'zorll' &
.or. trim(sfc_name2(num)) == 'zorli' .or. trim(sfc_name2(num)) == 'zorlwav'&
.or.trim(sfc_name2(num)) == 'albdvis_lnd' .or. trim(sfc_name2(num)) == 'albdnir_lnd' &
.or.trim(sfc_name2(num)) == 'albvis_lnd' .or. trim(sfc_name2(num)) == 'albnir_lnd') then
id_restart = register_restart_field(Sfc_restart, fn_srf, sfc_name2(num), var2_p, domain=fv_domain, mandatory=.false.)
else
id_restart = register_restart_field(Sfc_restart, fn_srf, sfc_name2(num), var2_p, domain=fv_domain)
endif
enddo

climbfuji · 2021-06-01T19:17:05Z

Helin, this is more an fv3atm question than a CCPP question. Can you follow what was done for NoahMP and RUC in FV3GFS_io.F90?

…

On Jun 1, 2021, at 1:00 PM, HelinWei-NOAA ***@***.***> wrote: With the change shown below, it is still not working and I got the same error (FATAL from PE 86: fms_io(restore_state_all): unable to find mandatory variable albivis_lnd in restart file sfc_data.tile3.nc) !--- register the 2D fields do num = 1,nvar_s2m var2_p => sfc_var2(:,:,num) if (trim(sfc_name2(num)) == 'sncovr'.or. trim(sfc_name2(num)) == 'tsfcl' .or. trim(sfc_name2(num)) == 'zorll' & .or. trim(sfc_name2(num)) == 'zorli' .or. trim(sfc_name2(num)) == 'zorlwav'& .or.trim(sfc_name2(num)) == 'albdvis_lnd' .or. trim(sfc_name2(num)) == 'albdnir_lnd' & .or.trim(sfc_name2(num)) == 'albvis_lnd' .or. trim(sfc_name2(num)) == 'albnir_lnd') then id_restart = register_restart_field(Sfc_restart, fn_srf, sfc_name2(num), var2_p, domain=fv_domain, mandatory=.false.) else id_restart = register_restart_field(Sfc_restart, fn_srf, sfc_name2(num), var2_p, domain=fv_domain) endif enddo — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#669 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB5C2RMWMCQ53JO647CKTC3TQUU3TANCNFSM45WS5FQQ>.

HelinWei-NOAA · 2021-06-03T13:54:00Z

The modification and tests have been completed. Which branch should I try to merge?

test results: https://www.emc.ncep.noaa.gov/mmb/gcp/gfs/snetfix/www/maps/d2/maps2d_sfc.html
https://github.com/HelinWei-NOAA/ccpp-physics/tree/remove_snet_from_noah_lsm
https://github.com/HelinWei-NOAA/fv3atm/tree/remove_snet_from_noah_lsm

climbfuji · 2021-06-03T14:17:24Z

We need PRs for ccpp-physics, fv3atm (and ufs-weather-model, even if it only is the update of the fv3atm submodule pointer). The submodule pointers and .gitmodules need to temporarily point to your branches so that

git clone -b your_branch_name --recursive https://github.com/HelinWei-NOAA/ufs-weather-model

will give exactly the same code that is in your different branches.

HelinWei-NOAA · 2021-06-03T16:44:43Z

Okay. I created a branch under my ufs-weather-model fork. This modification will also change all baseline runs with Noah LSM. Do I need to recreate those by myself?

climbfuji · 2021-06-03T16:48:21Z

Okay. I created a branch under my ufs-weather-model fork. This modification will also change all baseline runs with Noah LSM. Do I need to recreate those by myself?

It would be good to create baselines on hera with Intel and GNU (using rt.sh -c ...) and then verify against them (using rt.sh -m). This will make sure that all tests run to completion and are run-to-run reproducible. Since other PRs will be merged before your PR that may change the baseline, too, the code managers will assist with creating the final baselines on all platforms when it is time to commit.

HelinWei-NOAA · 2021-06-03T19:22:18Z

I used git clone -b remove_snet_from_noah_lsm --recursive https://github.com/HelinWei-NOAA/ufs-weather-model to check out the code, but I didn't get the right branch for both FV3 and ccpp-physics. Can you please take a look at .gitmodules in /scratch1/NCEPDEV/global/Helin.Wei/save/git/sync/snet/ufs-weather-model? Thanks.

climbfuji · 2021-06-04T13:00:10Z

I checked your directories, I see that both fv3atm and ccpp-physics point to your branches. What is the problem?

climbfuji added the enhancement label May 28, 2021

climbfuji assigned HelinWei-NOAA May 28, 2021

This was referenced Jun 8, 2021

Fix snet bug for Noah LSM #673

Merged

Remove snet from noah lsm NOAA-EMC/fv3atm#325

Closed

climbfuji mentioned this issue Jul 22, 2021

Wrapper PR for: Thompson inner loop, Thompson subcycling bugfix, fix snet bug for Noah LSM #702

Merged

climbfuji closed this as completed in #702 Jul 23, 2021

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Remove snet from Noah LSM #669

Remove snet from Noah LSM #669

climbfuji commented May 28, 2021

HelinWei-NOAA commented Jun 1, 2021

climbfuji commented Jun 1, 2021

HelinWei-NOAA commented Jun 1, 2021

HelinWei-NOAA commented Jun 1, 2021

climbfuji commented Jun 1, 2021 via email

HelinWei-NOAA commented Jun 1, 2021 via email

climbfuji commented Jun 1, 2021 via email

HelinWei-NOAA commented Jun 1, 2021 via email

HelinWei-NOAA commented Jun 1, 2021

HelinWei-NOAA commented Jun 1, 2021

climbfuji commented Jun 1, 2021 via email

HelinWei-NOAA commented Jun 3, 2021

climbfuji commented Jun 3, 2021

HelinWei-NOAA commented Jun 3, 2021

climbfuji commented Jun 3, 2021

HelinWei-NOAA commented Jun 3, 2021

climbfuji commented Jun 4, 2021

Remove snet from Noah LSM #669

Remove snet from Noah LSM #669

Comments

climbfuji commented May 28, 2021

Description

Solution

Alternatives (optional)

Related to (optional)

HelinWei-NOAA commented Jun 1, 2021

climbfuji commented Jun 1, 2021

HelinWei-NOAA commented Jun 1, 2021

HelinWei-NOAA commented Jun 1, 2021

climbfuji commented Jun 1, 2021 via email

HelinWei-NOAA commented Jun 1, 2021 via email

climbfuji commented Jun 1, 2021 via email

HelinWei-NOAA commented Jun 1, 2021 via email

HelinWei-NOAA commented Jun 1, 2021

HelinWei-NOAA commented Jun 1, 2021

climbfuji commented Jun 1, 2021 via email

HelinWei-NOAA commented Jun 3, 2021

climbfuji commented Jun 3, 2021

HelinWei-NOAA commented Jun 3, 2021

climbfuji commented Jun 3, 2021

HelinWei-NOAA commented Jun 3, 2021

climbfuji commented Jun 4, 2021