User's Guide for the NMM Core of the
Weather Research and Forecast (WRF)
Modeling System Version 2.2
Chapter 4: WRF-NMM Initialization
Table of Contents
· Initialization for Real Data Cases
The real_nmm.exe
portion of the code generates initial and boundary conditions for the WRF-NMM
model (wrf.exe) that are derived from output files provided by
the WPS. Inputs required for the WRF-NMM model are not restricted
to WPS alone. Several variables are defined/re-defined
using the real_nmm part of the routines. For instance, the
WRF-NMM core uses the definition of the Coriolis parameter in real_nmm,
rather than that in WPS.
The real_nmm program performs
the following tasks:
· Reads data from the namelist
· Allocates space
· Initializes remaining variables
· Reads input data from the WRF Preprocessing System (WPS)
· Prepares soil fields for use in the model (usually vertical interpolation to the requested levels)
· Checks to verify soil categories, land use, land mask, soil temperature, and sea surface temperature are all consistent with each other
·
Vertically interpolates to the models
computational surfaces.
·
Generates initial condition file
· Generates lateral condition file
The real_nmm.exe program may be run as a distributed memory job, but there will be no computational speed up since this program relies heavily on I/O and does few computations.
The
real code includes:
(1) ./main/real_nmm.F
(2) ./dyn_nmm/module_initialize_real.F
(3) ./dyn_nmm/start_domain_nmm.F
(4) ./share/module_soil_pre.F
(5) ./main/real_nmm.exe
Initialization for Real Data Cases
The real_nmm.exe code uses data files provided by the WRF Preprocessing System (WPS) as input. The data processed by the WPS typically come from a previously run, large-scale forecast model. The original data are generally in “GriB” format and are ingested into the WPS by first using “ftp” to retrieve the raw GriB data from one of the national weather agencies anonymous ftp sites.
For example, a forecast from 2005 January 23 0000 UTC to 2005 January 24 0000 UTC which has original GriB data available at 3h increments will have the following files previously generated by the WPS:
met_nmm.d01.2005-01-23_00:
met_nmm.d01.2005-01-23_03:
met_nmm.d01.2005-01-23_06:
met_nmm.d01.2005-01-23_09:
met_nmm.d01.2005-01-23_12:
met_nmm.d01.2005-01-23_15:
met_nmm.d01.2005-01-23_18:
met_nmm.d01.2005-01-23_21:
met_nmm.d01.2005-01-24_00:
The convention is to use “met_nmm” to signify data that are output from the WPS and used as input into the real_nmm.exe program. The “d01” part of the name is used to identify to which domain this data refers. The trailing characters are the date, where each WPS output file has only a single time-slice of processed data. The WPS package delivers data that are ready to be used in the WRF-NMM system.
The
following statements apply to these data:
· The data adheres to the WRF IO API.
· The data has been horizontally interpolated to the correct grid-point staggering for each variable.
· The 3-D meteorological data from the WRF-NMM SI include: u, v, temperature, specific humidity.
· The 3-D surface data from the WRF-NMM SI include: soil temperature, soil moisture, soil liquid.
· The
2-D static data from the WRF-NMM SI include: terrain, land categories,
soil info, etc.
· There are 1D arrays describing the vertical coordinate.
· There are constants which include: domain size, date, lists of available optional fields, etc.
The procedure outlined below is used for
single or multiple (nested) grid runs.
1. Change to the working directory of choice (cd test/nmm_real or cd run).
2. Make sure the files listed below reside in or are linked to the working-directory chosen to run the model:
ETAMPNEW_DATA (WRFV2/run)
GENPARM.TBL (WRFV2/run)
gribmap.txt (WRFV2/run)
LANDUSE.TBL (WRFV2/run)
namelist.input (WRFV2/test/nmm_real)
real_nmm.exe (WRFV2/run)
RRTM_DATA (WRFV2/run)
SOILPARM.TBL (WRFV2/run)
tr49t67 (WRFV2/run)
tr49t85 (WRFV2/run)
tr67t85 (WRFV2/run)
VEGPARM.TBL (WRFV2/run)
wrf.exe (WRFV2/run)
3. Make sure the met_nmm.d01* files from the WPS either reside in or are linked to the working directory chosen to run the model.
4. Edit the namelist.input file in the working directory for dates, domain size, time step, output options, and physics options (see Chapter 5, Description of Namelist Variables section for details).
5. The command issued to run real_nmm.exe in the working directory will depend on the operating system.
On LINUX-MPI systems, the command
is:
mpirun
-np n real_nmm.exe
where “n” defines the number of processors to use. For single processor PCs, use 1.
For batch jobs on some IBM systems
(such as NCAR’s IBM bluevista),
the command is:
mpirun.lsf
real_nmm.exe
and for interactive runs, the
command is:
mpirun.lsf real_nmm.exe -rmpool 1
-procs n
where “n” stands for the number of processors (CPUs) to be used.
When real_nmm.exe is successful, the following files that are used by wrf.exe should be found in the working-directory:
wrfinput_d01 (Initial conditions, single time level data.)
wrfbdy_d01 (Boundary
conditions data for multiple time steps.)
To check whether the run is successful, look for “SUCCESS COMPLETE REAL_NMM INIT” at the end of rsl.out.0000.
The real_nmm.exe portion of the code does not input or output any file relevant to nested domains. Initial and boundary conditions for WRF-NMM nests are interpolated down from the parent grids during the WRF model run.
More details regarding the real data test case for 2005 January
23/00 through 24/00 is given in Chapter
5, Real Data Test Case.