How to set up and run WRF model Wei Wang NCAR/MMM.

How to set up and run WRF model

Wei Wang

NCAR/MMM

Outline

How to download and compile the WRF model code?

Namelist Input and output files

How to download and compile WRF?

Download WRF source code from

http://www.wrf-model.org/users/download.html What you get is

WRFV1.3.TAR.gz After gunzip and untar, you should see a directory

WRFV1/ cd to WRFV1/ directory, and you should see ..

WRFV1 directoryCHANGESMakefileREADMEREADME_test_casesRegistry/ - add/remove input/output, new variablesarch/ - architecture-dependent compile optionscleancompileconfiguredyn_eh/dyn_em/external/frame/inc/main/ - main programs: real, ideal, wrf, executablesphys/ - physics routinesrun/share/test/ - where you run the modeltools/

How to compile?

Type ‘configure’ to create configure.wrf file for your particular computer:

> ./configurechecking for perl5... nochecking for perl... found /usr/local/bin/perl (perl)Will use NETCDF in dir: /usr/local/netcdf-------------------------------------------------------Please select from among the following supported platforms.

1. Compaq OSF1 alpha (single-threaded) 2. Compaq OSF1 alpha SM (OpenMP) 3. Compaq OSF1 alpha DM/SM (RSL, MPICH, RSL IO, OpenMP) 4. Compaq OSF1 alpha DM/SM (RSL, DECMPI, RSL IO, OpenMP)

Enter selection [1-4] : 1

How to compile?You’ll see, for example:

You have chosen: Compaq OSF1 alpha (single-threaded)These are the default options for this platform:--------------------------------------------------------------#OMP = OMPCPP = FC = f90CC = ccCFLAGS =FCOPTIM = -fast -O4 -inline all……--------------------------------------------------------------These will be written to the file configure.wrf here in the top-leveldirectory. If you wish to change settings, please edit that file.If you wish to change the default options, edit the file: arch/configure.defaults

Configuration successful. To build the model type compile .

How to compile?

You may want to edit configure.wrf to change default compile options.

In WRF V1.3, compile options are provided for

SGI (single, OpenMP, MPI)

Sun (single and OpenMP – higher version required)

IBM AIX (single, OpenMP, MPI)

Alpha OSF (single, OpenMP, MPI)

Alpha Linux (single, MPI)

Linix (single, OpenMP, MPI)

How to compile?

After typing compile, you will be given a list of options:

compile em_b_wave

compile em_hill2d_x

compile em_quarter_ss

compile em_real

compile em_squall2d_x

compile em_squall2d_y

Typing one of the above will produce both initialization and model executables

How to compile?

Type one of the above to compile e.g. compile em_real If compile is successful, you should have two executables

built in directory main/: real.exe wrf.exe And these executables are linked to ./test/em_real/ if you

typed compile em_real If you type compile em_quarter_ss you’d have ideal.exe wrf.exe linked to ./test/em_quarter_ss/

How to compile? - Hints

If the netCDF library is built in a ‘unusual’ location on your computer, you may use the environment variable NETCDF to provide the correct path to configure. e.g.

setenv NETCDF /usr/local/lib32/r4i4

If you compile on a Linux computer, make sure that you have netCDF library installed with PGI compiler.

WRFV1/test directory

eh_b_wave/eh_hill2d_x/eh_quarter_ss/eh_real/eh_squall2d_x/eh_squall2d_y/em_b_wave/em_hill2d_x/em_quarter_ss/em_real/em_squall2d_x/em_squall2d_y/

There are six subdirectories for each dynamical core (we recommend that you use the em core):

The following are in the test/em_real/ directory:

LANDUSE.TBL

RRTM_DATA

eta_micro_loopup.dat

tr49t67

tr49t85

tr67t85

namelist.input - require editing

real.exe -> ../../main/real.exe

wrf.exe -> ../../main/wrf.exe

test/em_real directory

Physics data

Before you run real.exe or ideal.exe, and wrf.exe, edit the namelist.input for runtime options

More information on namelist can be found in

WRFV1/run/README.namelist

and on the Web:

http://www.mmm.ucar.edu/wrf/users/wrf-namelist.html

namelist.input

&namelist_01 time_step_max = 1, number of time steps to run max_dom = 1, number of domains (leave as is) dyn_opt = 2, dynamics option: 1 = Eulerian height coordinate 2 = Eulerian mass coordinate 3 = Semi-Lagrangian (not yet implemented)

rk_ord = 3, time-integration scheme option

2 = Runge-Kutta 2nd order

3 = Runge-Kutta 3rd order

Description of namelist_01

diff_opt = 1, diffusion option:

0 = no turbulence or explicit spatial numerical filters

1 = old diffusion scheme, evaluates diffusion terms on coordinate surfaces(uses khdiv,kvdif)

2 = new diffusion scheme, evaluates mixing terms in physical space (stress form turbulence parameterization by specifying km_opt)

km_opt = 1, eddy coefficient option 1 = constant (use khdiv kvdif) 2 = 1.5 order TKE closure 3 = Smagorinsky first order closure (3d)

4 = Smagorinsky first order closure (2d)

damp_opt = 1, upper level damping flag (for diff_opt=2 only) 0 = without damping 1 = with damping


ISFFLX = 1, heat and moisture fluxes from the surface 1 = with fluxes from the surface 0 = no flux from the surface

IFSNOW = 0, snow-cover effects (for bl_surface_physics=1 only)

1 = with snow-cover effect 0 = without snow-cover effect

ICLOUD = 1, cloud effect to the optical depth in radiation

(for ra_lw_physics = 1 and ra_sw_physics = 1 only) 1 = with cloud effect

0 = without cloud effect


num_soil_layers = 4, number of soil layers in land surface model

spec_bdy_width= 5, number of rows for specified boundary value nudging

spec_zone = 1, number of points in specified zone (spec b.c. option)

relax_zone = 4, number of points in relaxation zone (spec b.c. option)

tile_sz_x = 0, number of points in tile x direction

tile_sz_y = 0, number of points in tile y direction

numtiles = 1, number of tiles per patch (alternative to above two)

debug_level = 0/ 50,100,200,300 values give increasing prints


&namelist_02

grid_id = 1, domain identifier

level = 1, domain nest level

s_we = 1, start index in x (west-east) direction

e_we = 32, end index in x (west-east) direction

s_sn = 1, start index in y (south-north) direction

e_sn = 32, end index in y (south-north) direction

s_vert = 1, start index in z (vertical) direction

e_vert = 31, end index in z (vertical) direction



time_step_count_start = 0, start time-step (leave as is)

time_step_count_end = 10, end time-step (this is inactive)

time_step_count_output = 10, time-steps between history outputs

frames_per_outfile = 10, output times per history file

time_step_count_restart = 10, time-steps between restart outputs

time_step_begin_restart = 0, time-step of restart beginning

0=not a restart

time_step_sound = 4, number of sound steps per timestep

&namelist_03

dx = 2000, grid length in x direction

dy = 2000, grid length in y direction

dt = 12., time-step for advection

ztop = 20000., the height of the model top

zdamp = 5000., damping depth from model top

dampcoef = 0.2, damping coefficient (dampcoef <= 0.25)

smdiv = 0, divergence damping (0.1 is typical)

emdiv = 0.01, external-mode filter coef for mass coord (0.01 is typical for real data)

epssm = .1, time off-centering for vertical sound waves

khdif = 0, horizontal diffusion constant (m^2/s)

kvdif = 0, vertical diffusion constant (m^2/s)

mix_cr_len = 200, the critical value of the mixing length for

isotropic and anisotropic diffusion


radt = 10, minutes between radiation physics calls

bldt = 0, minutes between boundary-layer physics calls

cudt = 5, minutes between cumulus physics calls

julyr = 0, Julian Year for model start

julday = 1, Julian Day for model start

gmt = 0./ GMT time for model start


&namelist_04

periodic_x = .false., periodic boundary conditions in x direction

symmetric_xs = .false., symmetric boundary conditions at x start (west)

symmetric_xe = .false., symmetric boundary conditions at x end (east)

open_xs = .false., open boundary conditions at x start (west)

open_xe = .false., open boundary conditions at x end (east)

periodic_y = .false., periodic boundary conditions in y direction

symmetric_ys = .false., symmetric boundary conditions at y start (south)

symmetric_ye = .false., symmetric boundary conditions at y end (north)

open_ys = .false., open boundary conditions at y start (south)

open_ye = .false., open boundary conditions at y end (north)

nested = .false., nested boundary conditions (inactive)

specified = .true., specified boundary conditions (inactive)

top_radiation= .false., upper radiative boundary conditions (inactive)


chem_opt = 0, chemistry option

mp_physics microphysics option

= 0, no microphysics

= 1, Kessler scheme

= 2, Lin et al. scheme

= 3, NCEP 3-class simple ice scheme

= 4, NCEP 5-class scheme

= 5, Ferrier (new Eta) microphysics

= 99, Zhao-Carr (old Eta) microphysics


ra_lw_physics longwave radiation option

= 0, no longwave radiation

= 1, rrtm scheme

= 99, GFDL (Eta) longwave (semi-supported)

*remove –DTRIEDNTRUE from compile flags

ra_sw_physics shortwave radiation option

= 0, no shortwave radiation

= 1, Dudhia scheme

= 2, Goddard short wave

= 99, GFDL (Eta) shortwave (semi-supported)

*remove –DTRIEDNTRUE from compile flags


bl_sfclay_physics surface-layer option

= 0, no surface-layer

= 1, Monin-Obukhov scheme

= 2, Monin-Obukhov (Janjic Eta) scheme

bl_surface_physics land-surface option

= 0, no land-surface

= 1, thermal diffusion scheme

= 2, OSU land-surface model

bl_pbl_physics boundary-layer option

= 0, no boundary-layer

= 1, mrf scheme

= 2, Mellor-Yamada-Janjic (Eta) TKE scheme


cu_physics cumulus option

= 0, no cumulus

= 1, Kain-Fritsch (new) scheme

= 2, Betts-Miller-Janjic (Eta) scheme

= 99, previous Kain-Fritsch scheme


h_mom_adv_order = 5, horizontal momentum advection order (5=5th,etc.)

v_mom_adv_order= 3, vertical momentum advection order

h_sca_adv_order= 5, horizontal scalar advection order

v_sca_adv_order= 3, vertical scalar advection order

io_form_history = 2,

io_form_restart = 2,

io_form_initial = 2,

io_form_boundary = 2,

= 1, (not active)

= 2, netCDF format


&namelist_05 (only used for real.exe pre-processed files, ignored otherwise)

start_year = 2000, four digit year of starting time

start_month= 11, two digit (01-12) month of starting time

start_day = 20, two digit (01-31) day of starting time

start_hour = 12, two digit (00-23) hour of starting time

start_minute= 00, two digit (00-59) minute of starting time

start_second= 00, two digit (00-59) second of starting time

end_year = 2000, four digit year of ending time

end_month = 11, two digit (01-12) month of ending time

end_day = 21, two digit (01-31) day of ending time

end_hour = 00, two digit (00-23) hour of ending time

end_minute = 00, two digit (00-59) minute of ending time

end_second = 00, two digit (00-59) second of ending time

interval_seconds= 43200, time interval in seconds between analysis (and boundary) times


real_data_init_type which type of real input:

= 1, WRF SI

= 2, MM5 data (height model only at this time)

= 3, user written routine read_generic in module_si_io.F


&namelist_quilt

nio_tasks_per_group = 0, default value is 0, no quilting

nio_groups = 1, default 1, don’t change

Description of namelist_quilt

The following namelist controls asynchronized I/O for MPI:

WRF input files

Running real.exe turns

real_input_em.d01.yyyy-mm-dd_hh:mm:ss

to

wrfinput_d01 and wrfbdy_d01

Running ideal.exe creates

wrfinput_d01 (and wrfbdy_d01, but wrfbdy is not used)

How to run WRF?

To run WRF on a single processor, type

> wrf.exe >& wrf.out & To run WRF using OpenMP, one may need to choose how

many processors to use. To do this, type

> setenv OMP_NUM_THREADS 4

> wrf.exe >& wrf.out &

to select to run on 4-processor shared-memory computer. (Note, on some computer, if one doesn’t set the number of processors to use, the wrf.exe may use all processors available.)

How to run WRF?

To run WRF on a MPI machine with MPICH, type > mpirun –np 4 wrf.exe The standard-out and error will go into the following files: show_domain_0000: domain-deposition info rsl.out.0000 rsl.error.0000 rsl.out.0001 rsl.error.0001 rsl.out.0002 rsl.error.0002 rsl.out.0003 rsl.error.0003 There will be one pair of files for each running processor

For other MPI-implementation, the run command may be different. For example, on IBM, the MPI run-command is

> poe wrf.exe

WRF output files

After one runs wrf.exe, WRF produces these output files:

wrfout_d01_000000: history file

wrfrst_d01_nnnnnn: restart file where nnnnnn is model time step when the restart file is written.

If one splits output files, one may obtain:

wrfout_d01_xxxxxx

where xxxxxx is the time step at which the output file is written. For example,

wrfout_d01_000000

wrfout_d01_003600

WRF output files

WRF input and output files

wrfinput_d01

wrfbdy_d01

wrfout_d01_xxxxxx

wrfrst_d01_xxxxxx

are netCDF files. So one can use netCDF utilities to look at file header and values of any fields. e.g.

ncdump –h wrfout_d01_000000 – gives a list of fields

ncdump –v U wrfout_d01_000000 – prints out values of

U field in the file.

How to set up and run WRF model Wei Wang NCAR/MMM.

Documents

Transcript of How to set up and run WRF model Wei Wang NCAR/MMM.