NEMS NMMB topic

All other topics about postprocessing model data (GrADS and other software), about other numerical weather prediction software (including WRF-NMM and WRF-ARW discussion unrelated to UEMS/WRF EMS), and general meteorology talk go in this forum.
oib
Posts: 117
Joined: Mon Jan 31, 2011 3:29 pm

Re: NEMS NMMB topic

Post by oib » Thu May 05, 2016 7:50 am

Ivana wrote:Oib, don't know it would help, but you can try to remove ncep_proc_path from namelist.nps. I had the similar problem, don't know if error was like yours but I removed ncep_proc_path from namelist, and it worked.
Ivana, the error I receive from geogrid.exe is at the bottom of that log:

Code: Select all

* . * . * . * . * . * . * . * . * . * . * . * . * . * . * . * . * . * . * . * . 
     PROGRAM GRIDGEN_SFC HAS BEGUN. COMPILED 2005136.00     ORG: NP2
     STARTING DATE-TIME  MAY 05,2016  07:36:25.621  126  THU   2457514


 - READ CONFIGURATION NAMELIST
 &GRID
 DOMAIN_NAME     = b_d01               ,
 DOMAIN_TYPE     = bgrid               ,
 IMDL    =         100,
 JMDL    =         100,
 CENTLAT_MDL     =   42.4000015258789     ,
 CENTLON_MDL     =   12.5000000000000     ,
 DX_MDL  =  8.163499832153320E-002,
 DY_MDL  =  8.153600245714188E-002,
 HEMI_MDL        =  -999.000000000000     ,
 LAT_11_MDL      =  -999.000000000000     ,
 LON_11_MDL      =  -999.000000000000     ,
 ORIENT_LON_MDL  =  -999.000000000000     ,
 TANGENT_LAT_MDL =  -999.000000000000     ,
 IMDL_PARENT     = 10*-999,
 JMDL_PARENT     = 10*-999,
 DX_PARENT_MDL   = 10*-999.000000000000       ,
 DY_PARENT_MDL   = 10*-999.000000000000       ,
 CENTLAT_PARENT_MDL      = 10*-999.000000000000       ,
 CENTLON_PARENT_MDL      = 10*-999.000000000000       ,
 GFS_LPL_FILE    =                                                                                                                   
                                     
 /
 - SET UP MAP PROJECTION INFO FOR: b_d01
 - USER SPECIFIED DOMAIN: b_d01
 
 - CREATING SURFACE FIELDS FOR DOMAIN b_d01
 - DOMAIN TYPE bgrid
 - I/J GRID DIMENSIONS ARE:          100         100
 - MODEL RESOLUTION IN DEGREES IS:   8.158550038933754E-002
 - CENTER LAT/LON ARE:    42.4000015258789        12.5000000000000     
 - OUTPUT DATA IN GRIB2 FORMAT.
 - CALCULATE LAT/LONS ON MODEL GRID
 - CALCULATE LAND/SEA MASK AND OROGRAPHY
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
 - SOURCE GRID BOUNDS IS/IE/JS/JE:        22160       24041        4988
        6451
 - OPEN AND READ SOURCE FILE FOR MASK: 
 /usr/Build_NEMS/geog_complete/gl-latlong-1km-landcover.bsq.fixed
 - BAD OPEN, IERR IS:           16
The error is so simple that I can't understand it :-)
Simply it can't open the file gl-latlong-1km-landcover.bsq.fixed
The file is in the right path and it has the right name and it has the right permission/owner.
And the error happens both if I insert in the namelist.nps the option ncep_proc_path or if I don't insert it at all.

The source code that trigger that error is in the file /NPS/geogrid/src/gridgen_sfc/lsmask_orog.f , subroutine "calc_lsmask_aavg".
The source code is :

Code: Select all

print*,'- OPEN AND READ SOURCE FILE FOR MASK: ',trim(lsmask_file)
 call mpi_file_open(mpi_comm_world, lsmask_file, mpi_mode_rdonly, &
                    mpi_info_null, iunit, ierr)
 if (ierr /= 0) then
   print*,'- BAD OPEN, IERR IS: ', ierr
   call mpi_abort(mpi_comm_world, 1, ierr)
 endif

I hope someone of you guys should have and idea to debug this :)
In case should help, my namelist.nps and testb.nmp_input_igbp (linked from fort.81) looks like this:

namelist.nps

Code: Select all

&share
 wrf_core                   = 'NMB'
 max_dom                    = 1
 start_date                 = '1999-01-01_00:00:00', '1999-01-01_00:00:00'
 end_date                   = '1999-01-06_00:00:00', '1999-01-06_00:00:00'
 interval_seconds           = 10800
 io_form_geogrid            = 2
 opt_output_from_geogrid_path = '/usr/Build_NEMS/NEMSV1.0/DOMAINS/test_NMB4km/geogrid/'
 debug_level                = 1000
/

&geogrid
 parent_id                  = 1, 1
 parent_grid_ratio          = 1, 3
 i_parent_start             = 1, 89
 j_parent_start             = 1, 104
 s_we                       = 1, 1
 e_we                       = 100, 136
 s_sn                       = 1, 1
 e_sn                       = 100, 139,
 geog_data_res              = '5m', '30s'
 dx                         = 0.081635
 dy                         = 0.081536
 map_proj                   = 'rotated_llb'
 ref_lat                    = 42.4
 ref_lon                    = 12.5
 dom_cen_lat                = 42.4, 
 dom_cen_lon                = 12.5, 
 truelat1                   = 43.5
 truelat2                   = 43.5
 stand_lon                  = 12.3
 ref_x                      = 50
 ref_y                      = 50
 geog_data_path             = '/usr/Build_NEMS/geog_complete/'
 opt_geogrid_tbl_path       = '/usr/Build_NEMS/NEMSV1.0/NPS/NPS/geogrid/'
 ncep_processing            = .true.
 ncep_proc_prefix           = 'b'
 ncep_proc_domain_type      = 'bgrid'
 ncep_proc_grib2            = .true.
 do_gwd                     = .false.
 just_last                  = .false.
 use_igbp                   = .true.
/

&ungrib
 out_format                 = 'WPS'
 prefix                     = 'FOO'
 spectral                   = 'FOO'
/

&metgrid
 fg_name                    = FOO
 io_form_metgrid            = 2
 constants_name             = FOO
 opt_output_from_metgrid_path = '/usr/Build_NEMS/NEMSV1.0/DOMAINS/test_NMB4km/metgrid/'
 opt_metgrid_tbl_path       = '/usr/Build_NEMS/NEMSV1.0/NPS/NPS/metgrid/'
/

&nemsinterp
 pt                         = FOO,
 ptsgm                      = FOO,
 nz                         = FOO,
 direct_temp                = FOO,
 global                     = FOO,
 do_clouds                  = FOO,
 grib_src                   = FOO,
 boundary_flux              = FOO,
 lnsh                       = FOO,
 lnsv                       = FOO,
 vcoord                     = FOO,
 coord_levs                 = FOO,
/

testb.nmp_input_igbp (linked from fort.81)

Code: Select all

&grid
 domain_name="testb"
 domain_type="bgrid"
 imdl = 100
 jmdl = 100
 dx_mdl = 0.081635
 dy_mdl = 0.081536
 centlat_mdl = 42.4
 centlon_mdl = 12.5
/

&tiling
 max_total_land_tiles=1
/

&veg_tiling
 max_veg_tiles=1
 veg_tile_threshold=0.05
 default_veg_category=7
 num_veg_groups = 5
!-------------------------------------------------------------
! veg groups
! 1 - forest              categories - 1,2,3,4,5,8,9
! 2 - barren              categories - 15,16,20
! 3 - urban               categories - 13
! 4 - grassland/cropland  categories - 10,12,14
! 5 - shrubland/tundra    categories - 6,7,11,18,19
!
 veg_groups =1,1,1,1,1,5,5,1,1,4,5,4,3,4,2,2,9,5,5,2
!--------------------------------------------------------------
/

&soil_tiling
 max_soil_tiles=1
 soil_tile_threshold=0.05
 default_soil_category=7
 num_soil_groups = 5
!----------------------------------------------
! soil groups
! 1 - sand  - cats 1,2,3
! 2 - silt  - cats 4,5,8
! 3 - loam  - cats 6,7,9
! 4 - clay  - cats 10,11,12
! 5 - other - cats 13,15,16
!----------------------------------------------
 soil_groups = 1,1,1,2,2,3,3,2,3,4,4,4,5,9,5,5
/

!----------------------------------------------------------
! smooth = 1, peak smoothing with 'num_smooth_passes1' passes.
! smooth = 2, smoother/desmoother with 'num_smooth_passes2' passes.
! smooth = 3, both (1, then 2)
! both options smooth the lateral boundaries in
! the same heavy-handed manner.
!----------------------------------------------------------

&lsmask_orog_tiling
 lsmask_aavg=.true.
 lsmask_tiles=.false.
 lsmask_tile_threshold=0.0
 orog_gwd_tiles=.true.
 max_orog_tiles=1
 orog_bin_width=300.0
 orog_tile_threshold=0.05
 smooth=3
 num_smooth_passes1=6
 num_smooth_passes2=4
/

&input_data
 leaf_area_idx_file=""
 gfrac_file="/usr/Build_NEMS/geog_complete/green.0.144.bin"
 mxsnow_alb_file="/usr/Build_NEMS/geog_complete/mxsno.1.0.bin"
 roughness_file=""
 slopetype_file="/usr/Build_NEMS/geog_complete/slope.1.0.bin"
 snowfree_albedo_file="/usr/Build_NEMS/geog_complete/albedo.1.0.bin"
 soiltype_tile_file="/usr/Build_NEMS/geog_complete/topsoil_fao.30s"
 substrate_temp_file="/usr/Build_NEMS/geog_complete/tbot.1.0.bin"
 vegtype_tile_file="/usr/Build_NEMS/geog_complete/veg_igbp1a.bin"
 lsmask_file="/usr/Build_NEMS/geog_complete/gl-latlong-1km-landcover.bsq.fixed"
 orog_file="/usr/Build_NEMS/geog_complete/terrain_usgs.30s"
/

&output_data
 grib2=.true.
/

oib
Posts: 117
Joined: Mon Jan 31, 2011 3:29 pm

Re: NEMS NMMB topic

Post by oib » Thu May 05, 2016 7:57 am

meteoadriatic wrote:Interesting, as I compare my temperatures with ARW run it seems they are also higher than should be, but I didn't noticed that extreme differences. At what conditions your temps were most off? Day, night, during sunny weather, cloudy...?
Hi Ivan, need to do furter testing in order to understand conditions, but on a 4 km grid domain I see T2m DIURNAL values on the Italy plains, during a SUNNY DAY, that are 5 / 10 degrees over the values that came from LISS land surface model!
Let you know better.

meteoadriatic
Posts: 1512
Joined: Wed Aug 19, 2009 10:05 am

Re: NEMS NMMB topic

Post by meteoadriatic » Thu May 05, 2016 10:43 am

It seems you're right about t2m. I will try LISS now.

Ivana
Posts: 14
Joined: Wed May 04, 2016 8:09 am

Re: NEMS NMMB topic

Post by Ivana » Tue May 17, 2016 11:50 am

I am sorry for my late reply.
Oib, did you figure it out where was the problem?
I am looking at your namelist and I can't see where is the problem.

Ivana
Posts: 14
Joined: Wed May 04, 2016 8:09 am

Re: NEMS NMMB topic

Post by Ivana » Thu May 19, 2016 7:58 am

Hi everybody,
Did anyone of you had problem with NMMB output for wind on 10m?
I don't have output in 00, all other parameters in 00 are ok, and wind on 10m in 03, 06... are also ok.
I tried to plot graph in GRADS also with raw data from model, and with GRIB data, but I got same result, no wind in 00...
Is it my mistake (that i can't see) or something in the model...
Thank you.

oib
Posts: 117
Joined: Mon Jan 31, 2011 3:29 pm

Re: NEMS NMMB topic

Post by oib » Thu May 19, 2016 8:10 am

Ivana wrote:I am sorry for my late reply.
Oib, did you figure it out where was the problem?
I am looking at your namelist and I can't see where is the problem.
Yes I did.
The problem wasn't in the NEMS itself but in the library MPI.
Recompiled with the default options and now ncep_processing = .true. works :-)
It seems to me that enabling ncep_processing is useful only for running the model with gravity waves. In that case the static dataset produced by geogrid.exe contains additional fields related to the surface elevation needed by the gravity wave routines.

oib
Posts: 117
Joined: Mon Jan 31, 2011 3:29 pm

Re: NEMS NMMB topic

Post by oib » Thu May 19, 2016 8:13 am

Ivana wrote:Hi everybody,
Did anyone of you had problem with NMMB output for wind on 10m?
I don't have output in 00, all other parameters in 00 are ok, and wind on 10m in 03, 06... are also ok.
I tried to plot graph in GRADS also with raw data from model, and with GRIB data, but I got same result, no wind in 00...
Is it my mistake (that i can't see) or something in the model...
Thank you.
Also RH 2m is empty at 00.
I think this is normal.

oib
Posts: 117
Joined: Mon Jan 31, 2011 3:29 pm

Re: NEMS NMMB topic

Post by oib » Thu May 19, 2016 8:15 am

Did someone managed to run the NEMS with a fractional time step (dt_num not equal to 0) ?
I didn't!
I get the following fatal error message in the NEMS.x log:

Code: Select all

"Timestep of this domain does not divide evenly into the restart interval! ABORTING!"

Ivana
Posts: 14
Joined: Wed May 04, 2016 8:09 am

Re: NEMS NMMB topic

Post by Ivana » Thu May 19, 2016 8:27 am

oib wrote:Did someone managed to run the NEMS with a fractional time step (dt_num not equal to 0) ?
I didn't!
I get the following fatal error message in the NEMS.x log:

Code: Select all

"Timestep of this domain does not divide evenly into the restart interval! ABORTING!"
I glad that you figure it out.
For now, I run it only with dt_num: 0.
Do you have some problem with time step?

oib
Posts: 117
Joined: Mon Jan 31, 2011 3:29 pm

Re: NEMS NMMB topic

Post by oib » Thu May 19, 2016 9:14 am

Ivana wrote: I glad that you figure it out.
For now, I run it only with dt_num: 0.
Do you have some problem with time step?
I'm running with a nested domain (1-way) and I'm playing with time step values.
I realized that I can run successfully only if I use integer time step values for all domains (dt_num: 0)
Furthermore:
+ the result of (minutes_history*60 / dt_int) must be zero for each domain, otherwise the output times will not always happen exactly on every minutes_history.
+ the child domain time step must equal to the result of (parent domain time step) / (parent to child grid size ratio). If not, the output files of only one domain reach the whole simulation length and the model hangs as this happens.

Post Reply

Who is online

Users browsing this forum: No registered users and 2 guests