3.2 Hydrostatic regional model nested into global model

      The example for this configuration can be found in $DISK/[machine]/g2r97. There are three files under this directory, named DCLSYS, LOCRSM and rung2n. A detailed description of these three files can be found in the third section called 'Elements of an experimental design'. We will describe the major definition to have this kind of the configuration.

      In file DCLSYS, the following definition is required to have this type of experiment:

     .....
%DCL MODEL OPTIONS;
 % ##NON   = 'C-NON'; 
 % ##HYD   = '     '; 
 % ##G2R   = '     '; 
 % ##C2R   = 'C-C2R'; 
 % ##N2R   = 'C-N2R'; 
    .......

This indicates that non-hydrostatic (##NON) is off (commented out by C-NON), hydrostatic (##HYD) is on, using global file to regional domain (##G2R) is on, coarse regional file to regional mesh (##C2R) is off (C-C2R) and coarse non-hydrostatic file to regional mesh (##N2R) is off (C-N2R). Note that you have to give 5 spaces inside the '' in order to have that option turn on.

      In file LOCRSM, the following example should be given

&NAMLOC
   RPROJ    = 0.0       ,
   RTRUTH   = 20.0      ,
   RORIENT  = -157.     ,
   RDELX    = 10000.    ,
   RDELY    = 10000.    ,
   RCENLAT  = 20.0      ,
   RCENLON  = -157.     ,
   RLFTGRD  = 53.       ,
   RBTMGRD  = 35.       ,
   CPROJ    = 0.        ,
   CTRUTH   = 0.        ,
   CORIENT  = 0.        ,
   CDELX    = 0.        ,
   CDELY    = 0.        ,
   CCENLAT  = 0.        ,
   CCENLON  = 0.        ,
   CLFTGRD  = 0.        ,
   CBTMGRD  = 0.        ,
 &END

where all the variables with prefix C are used for global in this case, and the global grid definition is known, so all C* here can be all zero (in fact, not used). All others with prefix R can be specified by the way described in the third section called 'Elements of an experimental design'.

      In file rung2r, we have to make the directory definition, compile option and others are right. The WAVINP, LEVINP are related to initial grab resolution, WAVGAS and LEVGAS have to be the dimension related to global model data used in RSM.


3.3 Non-hydrostatic regional nested into global

      An example of this configuration can be found in $DISK/[machine]/g2n97. There are three files under this directory, DCLSYS, LOCRSM and rung2n. A detailed description of these three files can be found in the third section called 'Elements of an experimental design'. We will describe the major definition to have this kind of the configuration.

      In file DCLSYS, the following definition is required to have this type of experiment:

     .....
%DCL MODEL OPTIONS;
 % ##NON   = '     '; 
 % ##HYD   = 'C-HYD'; 
 % ##G2R   = '     '; 
 % ##C2R   = 'C-C2R'; 
 % ##N2R   = 'C-N2R'; 
    .......

This indicates that non-hydrostatic (##NON) is on, hydrostatic (##HYD) is off, using global file to regional domain (##G2R) is on, coarse regional file to regional mesh (##C2R) is off (C-C2R) and coarse non-hydrostatic file to regional mesh (##N2R) is off (C-N2R). Note that you have to give 5 spaces inside ' ' in order to have that option turn on.

      In file LOCRSM, the following example should be given

&NAMLOC
   RPROJ    = 0.0       ,
   RTRUTH   = 20.0      ,
   RORIENT  = -157.     ,
   RDELX    = 10000.    ,
   RDELY    = 10000.    ,
   RCENLAT  = 20.0      ,
   RCENLON  = -157.     ,
   RLFTGRD  = 53.       ,
   RBTMGRD  = 35.       ,
   CPROJ    = 0.        ,
   CTRUTH   = 0.        ,
   CORIENT  = 0.        ,
   CDELX    = 0.        ,
   CDELY    = 0.        ,
   CCENLAT  = 0.        ,
   CCENLON  = 0.        ,
   CLFTGRD  = 0.        ,
   CBTMGRD  = 0.        ,
 &END

where all the variables with prefix C are used for global in this case, and the global grid definition is known, so all C* here can be all zero (in fact, not used). All others with prefix R can be specified by the way described in he third sections called 'Elements of an experimental design'.

      In file rung2n, we have to make the directory definition, compile option and others are right. The WAVINP, LEVINP are related to initial grab resolution, WAVGAS and LEVGAS have to be the dimension related to global model data used in RSM.


3.4 Hydrostatic regional nested into hydrostatic regional

      An example of this configuration can be found in $DISK/[machine]/c2r97. There are three files under this directory, named DCLSYS, LOCRSM and rung2n. The detailed description of these three files can be found in the third sections called 'Elements of an experimental design'. We will describe the major definition to have this kind of the configuration.

      In file DCLSYS, the following definition is need in order to have this type of experiment:

     .....
%DCL MODEL OPTIONS;
 % ##NON   = 'C-NON'; 
 % ##HYD   = '     '; 
 % ##G2R   = 'C-G2R'; 
 % ##C2R   = '     '; 
 % ##N2R   = 'C-N2R'; 
    .......

The above indicates that non-hydrostatic (##NON) is off (comment out by C-NON), hydrostatic (##HYD) is on, using global file to regional domain (##G2R) is off (C-G2R), coarse regional file to regional mesh (##C2R) is on and coarse non-hydrostatic file to regional mesh (##N2R) is off (C-N2R). Again, note that you have to give 5 spaces inside ' ' in order to have that option turn on.

      In file LOCRSM, an example is

&NAMLOC
   RPROJ    = 0.0       ,
   RTRUTH   = 40.0      ,
   RORIENT  = -157.     ,
   RDELX    = 10000.    ,
   RDELY    = 10000.    ,
   RCENLAT  = 20.0      ,
   RCENLON  = -157.     ,
   RLFTGRD  = 53.       ,
   RBTMGRD  = 35.       ,
   CPROJ    = 1.        ,
   CTRUTH   = 60.       ,
   CORIENT  = -157.     ,
   CDELX    = 40000.    ,
   CDELY    = 40000.    ,
   CCENLAT  = 90.       ,
   CCENLON  = 0.        ,
   CLFTGRD  = 72.       ,
   CBTMGRD  = 220.      ,
 &END

where all the variables with prefix C are used for coarse grid mesh in this case and R* are for fine mesh. They can be specified by the way described in the third section called 'Elements of an experimental design'.

      In file runc2r, you have to make sure the directory definition, compile option and others are correct.


3.5 Non-hydrostatic regional nested into hydrostatic regional

      An example of this configuration can be found in $DISK/[machine]/c2n97. There are three files under this directory, DCLSYS, LOCRSM and run2n. A detailed description of these three files can be found in the third section called 'Elements of an experimental design'. We will describe the major definitions to needed for this kind of configuration.

      In file DCLSYS, the following definitions are required for this type of experiment:

     .....
%DCL MODEL OPTIONS;
 % ##NON   = '     '; 
 % ##HYD   = 'C-HYD'; 
 % ##G2R   = 'C-G2R'; 
 % ##C2R   = '     '; 
 % ##N2R   = 'C-N2R'; 
    .......

The above indicate that the non-hydrostatic (##NON) is on, hydrostatic (##HYD) is off, using global file to force the regional domain (##G2R) is off, the coarse regional file to regional mesh (##C2R) is on and the coarse non-hydrostatic file to regional mesh (##N2R) is off (C-N2R). Again, note that you have to give 5 spaces inside ' ' in order to have that option turn on.

       In file LOCRSM, the following example is given.

&NAMLOC
   RPROJ    = 0.0       ,
   RTRUTH   = 40.0      ,
   RORIENT  = -157.     ,
   RDELX    = 10000.    ,
   RDELY    = 10000.    ,
   RCENLAT  = 20.0      ,
   RCENLON  = -157.     ,
   RLFTGRD  = 53.       ,
   RBTMGRD  = 35.       ,
   CPROJ    = 1.        ,
   CTRUTH   = 60.       ,
   CORIENT  = -157.     ,
   CDELX    = 40000.    ,
   CDELY    = 40000.    ,
   CCENLAT  = 90.       ,
   CCENLON  = 0.        ,
   CLFTGRD  = 72.       ,
   CBTMGRD  = 220.      ,
 &END

All the variables with prefix C are used for coarse grid mesh in this case and R* are for fine mesh. They can be specified by the way described in he third sections called 'Elements of an experimental design'.

      In file runc2r, we have to make sure the directory definition, compile option and others are correct.  


3.6 Non-hydrostatic regional nested into non-hydrostatic regional

      An example of this configuration can be found in $DISK/[machine]/n2n97. There are three files under this directory, named DCLSYS, LOCRSM and runn2n. A detailed description of these three files can be found in the third sections called 'Elements of an experimental design'. We will describe the major definitions needed for this configuration.

      In file DCLSYS, the following definitions are required.

     .....
%DCL MODEL OPTIONS;
 % ##NON   = '     '; 
 % ##HYD   = 'C-HYD'; 
 % ##G2R   = 'C-G2R'; 
 % ##C2R   = '     '; 
 % ##N2R   = '     '; 
    .......

The above indicates that non-hydrostatic (##NON) is on, hydrostatic (##HYD) is ff, using global file to regional domain (##G2R) is off, coarse regional file to regional mesh (##C2R) is on, and coarse non-hydrostatic file to regional mesh (##N2R) is on. Again, note that you have to give 5 spaces inside ' ' in order to have that option turn on.

      In file LOCRSM, an example is

&NAMLOC
   RPROJ    = 0.0       ,
   RTRUTH   = 20.0      ,
   RORIENT  = -157.     ,
   RDELX    = 10000.    ,
   RDELY    = 10000.    ,
   RCENLAT  = 20.0      ,
   RCENLON  = -157.     ,
   RLFTGRD  = 53.       ,
   RBTMGRD  = 35.       ,
   CPROJ    = 0.        ,
   CTRUTH   = 20.       ,
   CORIENT  = -155.     ,
   CDELX    = 30000.    ,
   CDELY    = 30000.    ,
   CCENLAT  = 20.       ,
   CCENLON  = -157.     ,
   CLFTGRD  = 72.       ,
   CBTMGRD  = 50.       ,
 &END

where all the variables with prefix C are used for coarse grid mesh in this case and R* are for fine mesh. They can be specified by the way described in he third sections called 'Elements of an experimental design'.

      In file runc2r, make sure the directory definition, compile option and other options are correct.  


3.7 Global model experiment

      This example shows how to run the global spectral model by using the initial condition from the NCEP. It may be able to be found in $DISK/[machine]/gsm97. There is no LOCRSM for this directory, and DCLSYS should be like the following.

%DCL DEFINE WHICH MACHINE TO RUN;
 % ##CRA   = 'C-CRA';  %DCL CRAY CODE ;
 % ##AFA   = '     ';  %DCL WORKSTATION CODE ;
%DCL DEFINE HYDROSTATIC OR NONHYDROSTATIC AND CLOUD NUMBERS;
 % ##HYD   = '     ';
 % ##NON   = 'C-NON';
 % #NCLDG  = '1';      %DCL NUMBERS OF CLOUD SPECIES FOR GSM;
 % #NCLD   = '1';      %DCL NUMBERS OF CLOUD SPECIES FOR RSM;
%DCL DEFINE MODEL PHYSIC DIMENSION FOR LOCAL;
 % #ILOT   = #LONB*2 ;
 % #KLOT   = #LEVS ;
 % #ILOR   = #LONR*2 ;
%DCL FOR POST;
 % #IO     = #LONB ;
 % #JO     = #LATB ;
 % #RECL   = 'NBYTES' ;

We could also try to run from a reduced resolution instead of from the NCEP high resolution, in which case,

the comiple portion in run script should be

# ------------------- CMPL -----------------
if [ $1 = cmpl ]
then
    cd $DIRCMPL || exit
    export GFCST;       GFCST=yes
    export GCHGR;       GCHGR=yes
    export GPOST;       GPOST=yes
    $DIRPROC/PCMPL || exit
fi

and the change resolution portion before forecast in run script should be

# -------------------- CHGR -----------------
if [ $1 = chgr -o $1 = all ]
then
    cd $DIRTEMP || exit
    if [ $JCAPI = $JCAP -a $LEVSI = $LEVS ]
    then
        echo " &NAMCHG                     "  >gchgrparm
        echo "  NIN=11,NOU=51,LTRN=.TRUE., " >>gchgrparm
        echo "  LNEW=.TRUE.,LREP=.FALSE.,  " >>gchgrparm
        echo " &END                        " >>gchgrparm
        $DIRPROC/PGCHGR || exit
    else
        cp siganl sigi
        cp sfcanl sfci
    fi
    mkdir -p $DIROUTP
    cp sigi  $DIROUTP/sigf00
    cp sfci  $DIROUTP/sfcf00
    cp sigi sigim
fi

However, the version you obtained may have already had this directory.


3.8 The difference between daily forecast and climate run

      There is only one major difference between a daily forecast and climate run. For a climate run, in either forecast mode or simulation mode, some of the surface conditions have to be changed during the integration, for example sea surface temperature, to make sure there are monthly and seasonal changes of the external conditions, which affect to the atmosphere. By forecast mode, we mean that except for the surface condition the observation or analysis field for the atmosphere is used only for the initial condition. By simulation mode, especially for regional model, we mean that not only the initial condition but also the base fields use the observation or analysis fields in addition to the changing surface condition. There is no FULLY coupled forecast model with the current version. All the surface conditions, like sea surface temperature is either obtained from observations or from a climatology. (However, we do have a coupled atmosphere and ocean model at NCEP, which uses the atmospheric model provided here.)

      There are some modifications needed to run the script and adding one more script in $DISK/common/script is necessary. Some machine types (and software versions) may have them but some may be not. They are provided from the following steps:

For modification to run script from

            # -------------------RSM RESTART ---------------
            mv r_sigfm r_sigim
            mv r_sigf  r_sigi
            mv r_sfcf  r_sfci
            mv basef basei

to

            # -------------------RSM RESTART ----------------
            mv r_sigfm r_sigim
            mv r_sigf  r_sigi
            mv r_sfcf  r_sfci
            mv basef basei
            CHECK=` expr $FH % 12 `
            if [ $CHECK = 0 ]
            then
                DAY=`expr $FH \/ 24 `
                CDATE=`expr $INIDATE + $DAY `
                CYCLE=`expr $FH % 24 `
                cp $DIRINPT/sfcanl$JCAP.$CDATE$CYCLE.fmt.Z . && \
                 uncompress sfcanl$JCAP.$CDATE$CYCLE.fmt.Z && \
                  $DIRETC/unfmtsfc$JCAP.s \
                   sfcanl$JCAP.$CDATE$CYCLE.fmt basesfc && \
                    /bin/rm sfcanl$JCAP.$CDATE$CYCLE.fmt
                mv r_sfci r_sfctmp
                $DIRPROC/PRMSFC $FH
                mv r_sfcn r_sfci
            fi

And add script called PRMSFC to $DISK/common/script:

#! /bin/sh
set -x
NEWMTN=FALSE
## Prepare regional surface ##
mtnvar=$DIRCONS/mtnvar.$JCAP
echo " &NAMRIN                                           "  >rinpparm
echo "    SIG2RG=.FALSE.,SFC2RG=.TRUE.,PERCMTN=0.2,      " >>rinpparm
echo "    NEWSIG=.FALSE.,NEWMTN=.$NEWMTN.,NEWHOR=.FALSE.," >>rinpparm
echo " &END                                              " >>rinpparm
/bin/rm -rf fort.*
ln -s $mtnvar    fort.10
ln -s basesfc    fort.12
ln -s basesfcr   fort.52
rinpexec=$DIREXEC/rinp.x
cat rinpparm rsmlocation >stdinp.rinp
$rinpexec <stdinp.rinp >stdout.rinp  || exit $?

/bin/rm -rf fort.*
ln -s basesfcr     fort.10
ln -s r_sfctmp     fort.11
ln -s r_sfcn       fort.51
rsfcexec=$DIREXEC/rsfc.x
$rsfcexec >stdout.rsfc  || exit $?

And we have to have INCRSFCM in $DISK/common/include and MAINRSFCM in $DISK/common/source97. The concept of merge surface file are (1) the use of TSEA from basesfcr over ocean according to r_sfctmp's SLMSK (sea land mask), (2) use of TG3 and ALBEDO from basesfcr, and (3) use of ZORL and PLANTR from basesfcr over land.


3.9 Other experiments and display experimental results

      The examples we mentioned in the previous sections may not be the limitation for the current version of this package. In fact, we have other capabilities in case of any possible experiments, thus this section is reserved for any future experimental design. Let's show you one experiment with 2-dimensional option of using this 3-dimensional model. Then let's discuss how to display your results by GrADS or NCAR graphics.


2 D experiment

      To run any new experiment, you can make a new directory under $DISK/[machine], for example, say 2d97 is the new directory used for two dimensional experiment. Go inside the 2d97 directory and copy DCLSYS, LOCRSM and runc2n from c2n97 directory as

unix % cd  $DISK/[machine]
unix % mkdir  2d97
unix % cd  2d97
unix % cp  ../c2n97/DCLSYS  .
unix % cp  ../c2n97/LOCRSM  .
unix % cp  ../c2n97/runc2n  run2d

 where [machine] can be sgi, hp, or dec etc, then DCLSYS has to be something like:

%DCL THEORETICAL AND OTHER OPTIONS;
 % ##2D    = '    ';  %DCL TURN ON 2D VERSION;
 % ##VD    = '    ';  %DCL 2D VERTICAL DIFFUSION VERSION;
 % #VDIFU  =  0.2;  %DCL 2D VERTICAL DIFFUSION VERSION;
 % ##COR   = 'C-COR';  %DCL CORIOLIS OPTION;
............
%DCL REGIONAL MODEL GRID DIMENSIONS;
 % #IGRD    = '27'; %DCL SHOULD BE STILL FACTOR OF 2 AND 3 ONLY;
 % #JGRD    = '3';  %DCL SHOULD BE LARGER THAN 3;
 % #LEVR    = '100'; 
............

After ##2D is turn on, Coriolis parameter, ##COR, used to be turn off. Then vertical diffusion ##VD and its diffusion coefficient ##VDIFU can be turn on or off, depend on whether you need it or not. You can check how ##2D works by 'grep ##2D $DISK/common/source97/*'. You should see the lateral boundary condition remain only in x-direction. Thus it sets up 2D in x-direction, all j row should have to same results, thus you can have the small number of j row for #JGRD. Then you have to write your own initial condition for 2D. There are some examples in source97, say BUBBLINP or MTWAVINP. It may require several fine tune many thing. We expect any experienced user should be asked for help this kind of configuration.

      The LOCRSM should be given at any point, however, I suggest the best is to use Mercater projection and the center at tropic. But it is assumed to be any projection and any location because the input data have record to write latitude and longitude, the map factors etc.

      The run2d can be the same as other experiemnts, if you would like to use RSM, copy from c2r97, or c2n97 to use MSM. The reason why use c2n or c2r because the initial condition and base field which we will prepared should regional files, not global file. Refer to the file record in chapter 4.


Display GRIB files

      Let's try to use GrADS to see the output, the detail and more comprehensive guide should refer to GrADS documentation. Any experiment, say g2n, g2r, c2n, or c2r, the output directory, $DISK/run/[exp]/output, after the experiment finished should have something like:

........
r_flxf00
r_flxf03
........
r_pgbf00
r_pgbf00.ctlprs
r_pgbf03
r_pgbf03.ctlprs
...............
r_sfcf00
r_sfcf03
.........
r_sgbf00
r_sgbf00.ctlsig
r_sgbf03
r_sgbf03.ctlsig
............
r_sigf00
r_sigf03

as what we may mentioned, or check chapter 4, all r_* is related to regional files, r_flxf?? is regional binary flux file, r_pgbf?? is pressure+flux GRIB file, r_sfcf?? are regional binary surface file, r_sgbf?? is sigma GRIB file, r_sigf?? is regional binary sigma file. Here, we can see each GRIB file has its associated file as r_pgbf??.ctlprs for r_pgbf?? and r_sgbf??.ctlsig for r_sgbf??.ctlsig. The *ctl* is called GrAD control file which used to display the GRIB results by GrADS utilities.

      Before you can display the result, the initial configuration or some of the steps described in chapter 2 related to GrADS configuration have to be done, especially for 'setenv DISPLAY ....' and you may need to put 'xhost +' to allow your monitor to display from any machine etc. It is easy to see whether your environment is OK for GrADS, you do

unix % echo $DISPLAY
unix % echo $GADDIR
unix % echo $PATH

If the above response you something as following, respectively:

:0     or       140.90.112.91:0
/disk2/f97/dec/gradsdec
/usr/bin:/usr/bin/X11:.:/disk2/f97/dec/gradsdec

then you have everything ready. Let's back to the control file. The control file should be something like following, for example r_pgbf03.ctlprs,:

dset /atmos/f97/dec/g2n97/../../run/g2n97/output/r_pgbf03
dtype grib
options template
index /atmos/f97/dec/g2n97/../../run/g2n97/output/r_pgbf03.map
undef -9.99E+33
title EXP1
xdef   97 linear -161.976    0.096
ydef   76 levels
  16.914  17.006  17.097  17.188  17.280  17.371  17.463  17.554  17.645  17.736
  17.827  17.918  18.009  18.100  18.191  18.282  18.373  18.464  18.555  18.645
  18.736  18.827  18.917  19.008  19.098  19.189  19.279  19.369  19.460  19.550
  19.640  19.730  19.820  19.910  20.000  20.090  20.180  20.270  20.359  20.449
  20.539  20.628  20.718  20.807  20.897  20.986  21.075  21.165  21.254  21.343
  21.432  21.521  21.610  21.699  21.788  21.877  21.966  22.054  22.143  22.232
  22.320  22.409  22.497  22.586  22.674  22.762  22.850  22.939  23.027  23.115
  23.203  23.291  23.379  23.466  23.554  23.642
zdef   20 levels
    1000     950     925     900     850     800     750     700     650     600
     550     500     450     400     350     300     250     200     150     100
tdef 99 linear 00Z15NOV96         03hr
vars 77
hgtprs        20   7,100,0 GEOPOTENTIAL HEIGHT (M)
ugrdprs       20  33,100,0 ZONAL WIND (M/S)
..........
tminhag        0  16,105,0 MINIMUM TEMPERATURE (K)
hpblsfc        0 221,  1,0 PLANETARY BOUNDARY LAYER HEIGHT (M)
rhhag          0  52,105,0 RELATIVE HUMIDITY (PERCENT)
endvars

There is no difference among all the r_pgbf??.ctlprs except the place with forecast hour with bold face, say 03. In another word, for r_pgbf06.ctlprs, there should have 06 in those three places. To see the result in each forecast, for example r_pgbf09.ctlprs here, you do

unix % gribmap -i r_pgbf09.ctlprs

gribmap is GrADS utility to read GRIB file, the control file tells gribmap the GRIB filename is the one after 'dset' in the control file. After gribmap, it will generate a record index file which filename is given in the control file after 'index'. After you run gribmap as above, you should get

gribmap v1.2 for GrADS v1.5.1.14
mf 950913 -e punchout at end of file
Scanning binary GRIB file(s):
 /atmos/f97/dec/g2n97/../../run/g2n97/output/r_pgbf09
Reached EOF

There is no error, so you have index file which has the index of each field in the GRIB file as a direct access type. Then you can run grads, the same convention that italic characters are what you have to type, as:

unix % grads
.................
Landscape mode? (no for portrait):   "type no or just Enter"
............                         "GrADS window shown up"
ga> open r_pgbf09.ctlprs             "enter after the prompt"
Scanning description file:  r_pgbf09.ctlprs
Data file .........../r_pgbf09 is open as file 1
LON set to -161.976 -152.76
LAT set to 16.914 23.642
LEV set to 1000 1000
Time values set: 1996:11:15:0 1996:11:15:0
ga> set t 2                           "set to the correct time"
Time values set: 1996:11:15:9 1996:11:15:9

It shows you the correct time as above. You can not have map display without the correct time related to your data. Since your data is 1996:11:15:9 for r_pgbf09 and written in r_pgbf09.ctlprs in the line of 'tdef .....', then

ga> set mpdset hires               "use high resolution mapdata"
ga> d tmphag                       "display 2m temperature"
ga> c                              "clear the plot"
ga> set lev 850                    "set level to 850mb"
ga> d tmpprs                       "display 850mb temperature"
ga> d ugrdprs;vgrdprs              "display 850mb wind"

If there are too many vector, use skip GrADS function as

ga> d skip(ugrdprs,3);skip(vgrdprs,3)

If you want to make 500mb height and wind with title, and others, then print to your printer call 'phaserPS', you do

ga> c
ga> set lev 500
ga> d hgtprs
ga> d skip(ugrdprs,2);skip(vgrdprs,2)
ga> draw title 500mb Z and WIND at 1996:11:15:9
ga> enable print p.meta
ga> print
ga> c
ga> d tmphag
ga> d skip(ugrdhag,2);vgrdhag
ga> draw title 2mT and 10mWIND at 1996:11:15:9
ga> print
ga> disable print
ga> !gxps -c -i p.meta -o p.ps
ga> !lpr -P phaserPS p.ps

If you like to have GrADS to save some record in ieee type so you can use FORTRAN code to read later. You do:

ga> c
ga> set gxout fwrite
ga> d apcpsfc
Wrote 7448 elements to grads.fwrite

then you can use the file called grads.fwrite. The record length of the file should be 4 time the number after xdef plus 1 time the the number after ydef in the control file. For current control file, it should be 4*(97+1)*76. Each element has 4 word, but why there is one more element is x direction is not clear. Then the it can be read as:

      program readfr
      parameter(im=97,jm=76,len=(im+1)*jm)
      dimension f(im,jm)
      ....
      open(unit=10,file='grads.fwrite',form='unformatted',
     1     access='direct',recl=len)
      .....
      call getone(f,1)
      call getone(f,2)
      ....
      stop
      end
      subroutine getone(f,nc)
      parameter(im=97,jm=76,len=(im+1)*jm)
      dimension f(im,jm),tmp(im+1,jm)
      read(10,rec=nc) tmp
      do j=1,jm
        do i=1,im
          f(i,j)=tmp(i,j)
        enddo
       enddo
       return
      end

This may not be a good way to read, you may be able to find a better one than this, however, it just tries to show a way to get it. If you want to quit the grads, just do

ga> quit

Then you should be back to unix prompt, and all the *.meta, *.ps and grads.fwrite should be still in the directory for you to use later.

     There is a way to use the entire experiemntal results by follwoing:

unix % cp r_pgbf03.ctlprs r_pgb.ctl
unix % vi r_pgb.ctl

change all the first 03 to be %f2 and remove the second 03, change the number after tdef from 99 to 5 if you had 2 hr forecast with output in 3hr interval, so r_pgb.ctl should be like:

dset /atmos/f97/dec/g2n97/../../run/g2n97/output/r_pgbf%f2
dtype grib
options template
index /atmos/f97/dec/g2n97/../../run/g2n97/output/r_pgbf.map
undef -9.99E+33
..........
tdef 5 linear 00Z15NOV96        03hr
………
endvar

then you do next

unix % gribmap -i r_pgb.ctl
gribmap v1.2 for GrADS v1.5.1.14
mf 950913 -e punchout at end of file
Scanning binary GRIB file(s):
 /atmos/f97/dec/g2n97/../../run/g2n97/output/r_pgbf00
 /atmos/f97/dec/g2n97/../../run/g2n97/output/r_pgbf03
 /atmos/f97/dec/g2n97/../../run/g2n97/output/r_pgbf06
 /atmos/f97/dec/g2n97/../../run/g2n97/output/r_pgbf09
 /atmos/f97/dec/g2n97/../../run/g2n97/output/r_pgbf12
Reached EOF
unix % grads
....
ga> open r_pgb.ctl
ga> set mpdset hires
ga> set t 1 5
Time values set: 1996:11:15:0 1996:11:15:12
ga> d tmphag          "animation from t=1 to t=5"
Contouring: 286 to 300 interval 2
Contouring: 284 to 300 interval 2
Contouring: 282 to 300 interval 2
Contouring: 282 to 300 interval 2
Contouring: 282 to 300 interval 2
Press enter to continue:::::
ga> set t 4
Time values set: 1996:11:15:9 1996:11:15:9
ga> d tmphag          "only display time at 1996:11:15:9"

The others should be the same.


Display binary files

      To display the binary files, r_sig*, r_sfc* and r_flx*, the convient utility is NCAR graphics. Since NCAR graphics has copyright and it is not a freeware. We have planed to use GrADS for all the output display. In other word, any field you would like to have, we should create it in r_pgb* or r_sgb* files for display. However, there are some source-files you can follow to create NCAR graphics output by reading binary files. They are

$disk/common/source97/MAINPPRS   "plot pressure surface"
                      MAINNSIG   "plot sigma surface"
                      MAINPSFC   "plot surface field"
                      MAINPCRS   "plot cross section"

And the script, as an example, is

$disk/common/script/PRPLOT

Further command related to these two graphics package should refer to their document.


webmaster: Hann-Ming Henry Juang
henry.juang@noaa.gov