[GEMPAK #SKK-738821]: dcgrib2 problem with GEMPAK5.11.4

James,

Attached is a patched decode_grib2.c for GEMPAK 5.11.4 to be placed in 
$NAWIPS/unidata/ldmbridge/dcgrib2/

to build after copying the file:

cd $NAWIPS/unidata/ldmbridge/dcgrib2/
make all
make install

This update fixes a grib2->gempak grid decoding conflict assigning ensemble 
extensions to parameter names.

Michael James
Unidata Support Staff 


> Hi Michael,
> 
> I haven't received any response in nearly a week. So, I'm resending an
> email I sent out last Wednesday(in the event you've been so busy that my
> original email got lost in the shuffle).
> --------------------------------------------------------------------------
> 
> 
> Since I already missed the 12Z ensemble run, I ran dcgrib2 manually with
> a random ensemble file(ftp-ed from NCEP server). As you suggested, I
> included ENSEXT into the command:
> 
> dcgrib2 -c 090819/1200 -v 2 -e ENSEXT=0 -e
> gec00.t12z.pgrb2af30
> 
> When I run gdinfo, I see the following(just looking at 500 mb level but
> similar for other levels):
> 
> NUM       TIME1              TIME2           LEVL1 LEVL2  VCORD PARM
> 27     090819/1200F030                        500         PRES HGHTC002
> 28     090819/1200F030                        500         PRES TMPKC002
> 29     090819/1200F030                        500         PRES RELHC002
> 30     090819/1200F030                        500         PRES URELC002
> 31     090819/1200F030                        500         PRES VRELC002
> 
> If I run ENSEXT=1, I see this:
> 
> NUM       TIME1              TIME2           LEVL1 LEVL2  VCORD PARM
> 27     090819/1200F030                        500         PRES
> HGHTC002C002
> 28     090819/1200F030                        500         PRES
> TMPKC002C002
> 29     090819/1200F030                        500         PRES
> RELHC002C002
> 30     090819/1200F030                        500         PRES
> URELC002C002
> 31     090819/1200F030                        500         PRES
> VRELC002C002
> 
> I also ran it pointing to the 5.11.1 tables, but no change occurred. I
> don't think the 5.11.4 tables are at fault. When I ran dcgrib2 from
> 5.11.1, the parameter names didn't include the ensemble member name even
> when pointing to the 5.11.4 tables(ENSEXT=1 will append ensemble name,
> however). So, am I using the wrong syntax? Do you get the proper result
> with your dcgrib2(5.11.4)? It's also my interpretation that grib2 data
> will default with no extension name if ENSEXT is not set(which it isn't
> in my pqact for this data...using template found in
> $NAWIPS/ldm/etc/templates/pqact.gempak_decoders.in).
> 
> On a happier note, the problem with gempak not recording the GVCORD name
> "HYBL" was my fault for using an old Sun workstation(version 5.10.2 when
> "HYBL" wasn't a vertical coordinate option).
> 
> James
> 
> ----------------------------------------------
> James Murakami
> Staff Meteorologist/Student Affairs
> Department of Atmospheric and Oceanic Sciences
> University of California, Los Angeles
> 405 Hilgard Ave.
> Los Angeles, CA  90095-1565
> 
> 
> e-mail: address@hidden
> telephone: 310-825-2418
> Fax: 310-206-5219
> ----------------------------------------------
> 
> 
> 
> Unidata GEMPAK Support wrote:
> > Hi James,
> >
> > You're going to want to check that ENSEXT is not set to 1 in your 
> > pqact.gempak entries for ensemble grids.
> >
> > >From the dcgrib2 man page,
> >
> > http://www.unidata.ucar.edu/cgi-bin/gempak/manual/decoders_index?dcgrib2
> >
> > ENSEXT : Environmental variable for adding ensemble extensions to parameter 
> > names
> >             If ENSEXT is not set, use of PDSEXT parameter name extensions 
> > defaults
> >             to yes for grib1 and no for grib2. If ENSEXT is set, and equal 
> > to 1,
> >             then the extension will be added to parameter names. If ENSEXT 
> > is
> >             set, and not equal to 1, the PDS extension will not be added to 
> > parameter
> >             names.
> >
> >
> > That should do it, let me know if you have further questions.
> >
> > Michael James
> > Unidata User Support
> >
> >
> >
> >
> >> Hi,
> >>
> >> I installed the latest GEMPAK(5.11.4) last week, and I noticed some
> >> problems. For this email, I'm addressing a problem with how dcgrib2
> >> decodes GFS ensemble and RUC data(via the CONDUIT feed). What I see is
> >> dcgrib2 labeling ensemble parameters with the run variation name.
> >> That is, if one looks at temperature data, dcgrib2 produces labels of
> >> TMPKP001, TMPKP002, ...TMPKP20(TMPKC002 for the control run) for any
> >> given level. The prior version(5.11.1) just used the label, TMPK
> >> regardless of ensemble member(this is the preferred labeling). Also, for
> >> RUC data, there is no VCORD name given for ones that should be labeled
> >> "HYBL"(using gdinfo program to view all this). Again, the previous
> >> version did not omit this.
> >>
> >> Since I didn't see any mention of the above in the support email
> >> archive, I'm assuming that my GEMPAK didn't compile quite
> >> right(functional  but with flaws). It was installed on a computer with
> >> OS Mandriva 2008(a derivative of Redhat, I'm told). I've attached the
> >> make.out using g77(got a similar result compiling with gfortran). I
> >> compressed the file as it's rather large(2 megabytes). There were no
> >> "fatal" errors, but many warnings show up. However, I should tell you
> >> that for the previous version, a lot of similar warning messages
> >> occurred too. 5.11.1 continues to be our operational version for
> >> now(I've not encountered any relevant problems with usage).
> >>
> >> I'd appreciate your input on what I can do to correct some of these
> >> warning statements(if they are relevant for a correctly functional
> >> GEMPAK package). It's possible that another problem related to how NMAP2
> >> works might then be corrected(then, I won't need to send a separate
> >> email for this problem).
> >>
> >> James
> >>
> >> --
> >> ----------------------------------------------
> >> James Murakami
> >> Staff Meteorologist/Student Affairs
> >> Department of Atmospheric and Oceanic Sciences
> >> University of California, Los Angeles
> >> 405 Hilgard Ave.
> >> Los Angeles, CA  90095-1565
> >>
> >>
> >> e-mail: address@hidden
> >> telephone: 310-825-2418
> >> Fax: 310-206-5219
> >> ----------------------------------------------
> >>
> >>
> >>
> >>
> >>
> >
> >
> > Ticket Details
> > ===================
> > Ticket ID: SKK-738821
> > Department: Support GEMPAK
> > Priority: Normal
> > Status: Open
> >
> >
> >
> 
> 


Ticket Details
===================
Ticket ID: SKK-738821
Department: Support GEMPAK
Priority: Normal
Status: Open

#include "gb2def.h"
#include "dccmn.h"              /* for ivrblv global variable */
#include "gbcmn.h"              /* for gds global variable */

#include "dcgrib.h"

extern int ensext;
extern int pkmeth;  /* pkmeth = MDGRB2 defaults to GRIB2 storage, MDGDEC uses 
old method */

void
decode_grib2 (unsigned char *cgrib, int lenbul,
              char *gemfil, char *filnam, int maxgrd)
/************************************************************************
* void decode_grib2 ( cgrib, lenbul, gemfil, filnam, naxgrd)            *
*       unsigned char   *cgrib  Input grib product                      *
*       int             lenbul  length of input product                 *
*       char            *gemfil User specified output file name         *
*       char            *filnam final output filename                   *
*       int             maxgrd  Maximum number of grids in new file     *
*                                                                       *
* Error Codes                                                           *
* Log:                                                                  *
* Chiz/Unidata                  12/03   Created                         *
* James/Unidata                 9/09    Accounted for prmext            *       
*************************************************************************/
{
  int numgrds = maxgrd;
  int i, n, ilen1, ilen2, prmext, ier, iret;
  int iflno;
  g2int unpack = 0, expand = 0;
  g2int listsec0[3], listsec1[13], numlocal;
  Gribmsg curr_g2;
  Geminfo curr_gem;
  char vcoord[20], errstr[133], gdattm[2][DTTMSZ], extstr[20];
  int ighdr[LLGDHD];
  int gdflag = 255, iprec;
  static int tblinit = 0;
  static char errgrp[] = "DECODE_GRIB2";
  static int replace = 1; 

  static char g2tables[5][LLMXLN] = { 0 }, *tbllist[5];

  curr_g2.cgrib2 = cgrib;
  curr_g2.mlength = lenbul;
  curr_g2.gfld = NULL;
  curr_g2.field_tot = 0;

  if (!tblinit)
    {
      for (i = 0; i < 5; i++)
        tbllist[i] = g2tables[i];
      tblinit = !0;
    }

  if ((ier =
       g2_info (curr_g2.cgrib2, listsec0, listsec1, &(curr_g2.field_tot),
                &numlocal)) != 0)
    {
      dc_wclg (0, errgrp, ier, "g2_info failed", &iret);
      return;
    }

  for (n = 0; n < curr_g2.field_tot; n++)
    {
      ier = g2_getfld (curr_g2.cgrib2, n + 1, unpack, expand, &curr_g2.gfld);
  
      /* initialize strings in geminfo structure */
      memset ( curr_gem.cproj, 0, sizeof(curr_gem.cproj));
      memset ( curr_gem.parm, 0, sizeof(curr_gem.parm));
      memset ( curr_gem.gdattm1, 0, sizeof(curr_gem.gdattm1));
      memset ( curr_gem.gdattm2, 0, sizeof(curr_gem.gdattm2));

      gb2_2gem (&curr_g2, &curr_gem, tbllist, ensext, &ier);
      /* make sure the times are null terminated, since the routine above
         strncpy's 1 more character than the fortran routine provides */
      curr_gem.gdattm1[DTTMSZ-1] = '\0';
      curr_gem.gdattm2[DTTMSZ-1] = '\0';

      if (ivrblv >= 4)
        gb2_diag (curr_g2.gfld, gdflag);

      /* TODO: Convert these to GEMPAK weather codes and store as grid data */
      if (ivrblv >= 3)
        {
        if ( ( numlocal > 0 ) && ( curr_g2.gfld->locallen > 0 ) && ( 
curr_g2.gfld->local != NULL ) )
           {
           int i2vers = (int)curr_g2.gfld->local[0];
           int i2grp =  gb_btoi ( curr_g2.gfld->local,1,2,0);
           int i2val =  gb_btoi ( curr_g2.gfld->local,3,4,0);
           int i2ref = gb_btoi ( curr_g2.gfld->local,7,4,0);
           int i2scale = gb_btoi ( curr_g2.gfld->local,11,2,0);
           int i2bits = (int)curr_g2.gfld->local[13];
           int i2type = (int)curr_g2.gfld->local[14];
           g2int *i2out;

           printf("Local Section2 used [length= %d]\n",curr_g2.gfld->locallen);
           printf("check curr_g2.gfld->idsect[3] %d\n",curr_g2.gfld->idsect[3]);
           printf("check type %d\n",i2type);
           printf("section2 local %d %d %d\n",i2vers,i2grp,i2val);

           if(i2type == 1) 
              { /* simple packing using integers */
              i2out = (g2int *)malloc(i2val*(sizeof(int)));
              gbits ( curr_g2.gfld->local, i2out, (g2int)(15*8), (g2int)i2bits, 
(g2int)0, (g2int)i2val);
                 
              for(i=0;i<i2val;i++)
                 {
                 if ( i2out[i] == 0 ) 
                    printf("\n");
                 else
                    printf("%c",i2out[i]);
                 }
              printf("\n");
              free(i2out);
              }
           }
        }

      if (ier != 0)
        {
          sprintf (errstr, "Could not determine parameter name %d %d %d %d 
[%d]\0",
                   curr_g2.gfld->discipline, curr_g2.gfld->ipdtmpl[0],
                   curr_g2.gfld->ipdtmpl[1],curr_g2.gfld->ipdtnum, 
curr_gem.vcord);
          dc_wclg (0, errgrp, ier, errstr, &iret);
          g2_free (curr_g2.gfld);
          curr_g2.gfld = NULL;
          continue;
        }

      if (curr_g2.gfld->griddef == 0)   /* note... the center defined grid 
number is not provided */
        pds.grid_id = decode_g2gnum (curr_g2.gfld);
      else
        pds.grid_id = curr_g2.gfld->griddef;

      pds.center = curr_g2.gfld->idsect[0];
      pds.izero = curr_g2.gfld->idsect[1];
      pds.process = curr_g2.gfld->ipdtmpl[4];
      gds.kx = curr_g2.kx;
      gds.ky = curr_g2.ky;

      cst_rmbl ( curr_gem.gdattm1, gdattm[0], &i, &iret );
      cst_rmbl ( curr_gem.gdattm2, gdattm[1], &i, &iret );

      dcflnam2 (gemfil, filnam, &numgrds, curr_g2.gfld, gdattm[0], gdattm[1], 
&ier);

      if (ier == 0)
        {
          memset (ighdr, 0, sizeof (ighdr));

          dcogfil2 (filnam, &iflno, numgrds, curr_gem);

          if ( pkmeth == MDGRB2 )
            {
            gb2_wgem(curr_g2, curr_gem, n + 1, iflno, replace, 
                ighdr, &ier);
            /* method for storing necessary portion of GRIB2 product (Chiz)
             * sections 3,4,5,6,7. Write template arrays for sect 3,4.
             * Determine location to start of section 5, and length through 
section 7.
             * Write length of data to GEMPAK grid file.
             * Can be read through g2_unpack#.c routines.
             */
            }
          else
            {
            if (FGSIZ < (curr_g2.kx * curr_g2.ky))
              {
              FGRID =
                (float *) realloc (FGRID,
                                   (curr_g2.kx * curr_g2.ky) *
                                   sizeof (float));
              if ( FGRID == NULL ) {
                sprintf (errstr, "Reallocating grid for %d points [nx %d ny %d] 
failed\0", 
                        curr_g2.kx * curr_g2.ky, curr_g2.kx, curr_g2.ky);
                dc_wclg (3, errgrp, 0, errstr, &iret);
                g2_free (curr_g2.gfld);
                curr_g2.gfld = NULL;
                FGSIZ = 0;
                return;
              }
              FGSIZ = curr_g2.kx * curr_g2.ky;
              }

            /*
             *   Unpack GRIB2 grid
             */
            gb2_grid (&curr_g2, curr_gem.iuscal, curr_gem.rmsval,
                    &iprec, FGRID, &iret);

            /* Check JPEG & PNG missing values -- not needed currently -- 
            if ((curr_g2.gfld->idrtnum == 40000)
              || (curr_g2.gfld->idrtnum == 40)
              || (curr_g2.gfld->idrtnum == 41)
              || (curr_g2.gfld->idrtnum == 40010))
              dcmisjpg (curr_g2, FGRID);*/

            /*
             *   write grid out to gempak file.
             */
            /*gd_wpgd (&iflno, FGRID, &curr_g2.kx, &curr_g2.ky, ighdr,
                   (char *)gdattm, curr_gem.level,
                   &curr_gem.vcord, curr_gem.parm, &replace,
                   &pkmeth, &iprec, &ier, DTTMSZ, 12);*/
            cgd_wpgd (&iflno, FGRID, &curr_g2.kx, &curr_g2.ky, ighdr,
                   curr_gem.gdattm1, curr_gem.gdattm2, 
                   &curr_gem.level[0], &curr_gem.level[1],
                   &curr_gem.vcord, curr_gem.parm, &replace,
                   &pkmeth, &iprec, &ier );
            }


          lv_ccrd (&curr_gem.vcord, vcoord, &iret, sizeof (vcoord) - 1);
          vcoord[sizeof (vcoord) - 1] = '\0';
          cst_rmbl ( vcoord, vcoord, &i, &iret );

          curr_gem.parm[sizeof(curr_gem.parm)-1] = '\0';
          cst_rmbl( curr_gem.parm, curr_gem.parm, &i, &iret);

          sprintf (errstr, "%s [%s] %d:%d %s %d %d\0", curr_gem.parm,
             gdattm[0], curr_gem.level[0], curr_gem.level[1], 
             vcoord, curr_g2.kx, curr_g2.ky);
          if (ier != 0)
             dc_wclg (0, errgrp, ier, errstr, &iret);
          else
             dc_wclg (1, errgrp, ier, errstr, &iret);

        }
      else
        {
          sprintf (errstr, "decode_g2gds failed\0");
          dc_wclg (0, errgrp, ier, errstr, &iret);
        }

      g2_free (curr_g2.gfld);
      curr_g2.gfld = NULL;
    }

  if (ivrblv >= 2)
    {
      sprintf (errstr, "numfields %d numlocal %d\0", curr_g2.field_tot,
               numlocal);
      dc_wclg (2, errgrp, ier, errstr, &iret);
    }

}