20001003: GRDIMAGE issues (cont.)

>From: Bill Fingerhut <address@hidden>
>Organization: Lyndon State
>Keywords: 200009251326.e8PDQ1b04272 McIDAS-X GRDIMAGE

Bill,

Here is what I have done to further check out possible misbehavior by
GRDIMAGE:

o created two GRID files in the dataset MYDATA/GRIDS, GRID0132 and GRID0133
  GRID0132 is the first GRID file in the MYDATA/GRIDS dataset:

GRDLIST MYDATA/GRIDS.ALL FORM=FILE
DATASET NAME: MYDATA/GRIDS
 Dataset Position  Creation Date Max Grids         Directory Title
 ----------------  ------------- --------- -------------------------------
      132             2000270       2000
      133             2000277        200
     8702             2000124       1000
     8802             2000129       1000
     8902             2000123       1000   Eta AWIP grids from NCAR Archive
GRDLIST - done

o copied a sequence of ETA grids to both GRID files using GRDCOPY

GRDLIST MYDATA/GRIDS.132 NUM=ALL
Dataset position 132    Directory Title=
PAR    LEVEL       DAY        TIME    SRC FHOUR     FDAY       FTIME  GRID  PRO
---- --------- ------------ -------- ---- ----- ------------ -------- ----- ----
Z     100 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     1 LAMB
T     100 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     2 LAMB
U     100 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     3 LAMB
V     100 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     4 LAMB
RH    100 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     5 LAMB
U     150 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     6 LAMB
T     150 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     7 LAMB
V     150 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     8 LAMB
Number of grids listed = 8
GRDLIST - done

GRDLIST MYDATA/GRIDS.133 NUM=ALL
Dataset position 133    Directory Title=
PAR    LEVEL       DAY        TIME    SRC FHOUR     FDAY       FTIME  GRID  PRO
---- --------- ------------ -------- ---- ----- ------------ -------- ----- ----
Z     950 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     1 LAMB
U     950 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     2 LAMB
T     950 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     3 LAMB
V     950 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     4 LAMB
RH    950 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     5 LAMB
Z    1000 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     6 LAMB
T    1000 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     7 LAMB
U    1000 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     8 LAMB
V    1000 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00     9 LAMB
RH   1000 MB   03 OCT 00277 00:00:00  ETA     0 03 OCT 00277 00:00:00    10 LAMB
Number of grids listed = 10
GRDLIST - done

As you can see, GRID0133 (MYDATA/GRIDS.133) has fields for levels low in
the atmosphere while GRID0132 has the same set of fields for the upper
atmosphere.  If GRDIMAGE were to always use grids from the first position
in the dataset (132 in this case), then the image that is produced
should be significantly different than the one from the same element in
the second position (133).  The difference should be easily seen, and
the comparison with overlain contour plots of the same field should jump
out at you if they are wrong.

I turned grid 3 of MYDATA/GRIDS.133 (100 mb U) into an image with:

GRDIMAGE MYDATA/GRIDS.132 MYDATA/IMAGES.1232 RANGE=0 255 0 255 GRID=3 MAG=5 5

I turned grid 3 of MYDATA/GRIDS.133 (950 mb T) into an image with:

GRDIMAGE MYDATA/GRIDS.133 MYDATA/IMAGES.1233 RANGE=0 255 0 255 GRID=3 MAG=5 5

I then displayed both images in a 2x2 PANELed frame: MYDATA/IMAGES.1232 in
panels 1 and 2 (top row) and MYDATA/IMAGES.1233 in panels 3 and 4 (bottom
row).  These can be seen at:

http://www.unidata.ucar.edu/staff/tom/grdimage1.gif

You can see right off that the image of the 100 mb U wind component looks
significantly different than the 950 mb Temperature.  Also, the U image
looks like a U image, and the T image looks like a T image (meaning
that the image renditions look correct).

Next, I overlaid what should be the same grid as each image on top of
the image using the following:

GRDDISP MYDATA/GRIDS.132 GRID=3 NAV=C CINT=1 PAN=1

GRDDISP MYDATA/GRIDS.133 GRID=3 NAV=C CINT=1 PAN=3

The result can be seen in:

http://www.unidata.ucar.edu/staff/tom/grdimage2.gif

In both cases, the image and the overlaid grids match.

So, I am perplexed about your results.  It seems that GRDIMAGE on my
7.613 system is working correctly.  

>From address@hidden Tue Oct  3 15:10:40 2000
>Subject: Re: 20000929: GRDIMAGE continued 

>I thought I'd jump in here and add some info.  The server is running
>solaris 8 (x86) and the workstations are running Linux.  The server's
>version of mcidas is 7.612 (compiled w/ gcc & f2c).  The workstations are
>running mcidas 7.600.

So, the server's version of GRDIMAGE should match the one I am using.
This would seem to tell me that the tests Bill was running on the
server should work and the ones on the workstation might not.

Since I am using Linux at home, I will rerun the same tests from there
to see if the OS has anything to do with your problem.

>What's strange in this is that the workstation is
>producing the correct results.

Right.  This is why I will need to get login info so I can test things
out in Bill's environment.  The problem is that I leave for a Policy
Committee meeting tomorrow morning.  I won't be able to attack the
problem until about next Monday.

>Bill, please correct me if I am wrong here regarding the resultsA.  

Me too!

I have the server login, so I will try and jump on and see what I can
see.

Tom