[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20030904: McIDAS-X UACROSS fails on a local GFS dataset

>From: "Fingerhut, William A" <address@hidden>
>Organization: Lyndon State
>Keywords: 200309041715.h84HF6Ld023992 McIDAS-X v2003 UACROSS

Hi Bill,

>As I continue to test my scripts on Mcidas 2003, I've run into a
>problem creating vertical cross sections from grids.
>
>It appears that XSECTG is being replaced with UACROSS.

Yes, this has been in the works for quite a long time mainly because
XSECTG is not an ADDE application.

>In a pinch I
>could grab the source for XSECTG and compile it,

That is correct.  In fact, the source for XSECTG (xsectg.pgm) is still
included in the Unidata McIDAS distribution.  The makefile action to
build its executable is include in the obsolete keyins section of
the make file.  You could use that information to build the executable
if you really needed it.

>but in the long run we are supposed to use UACROSS.

Yes, absolutely.

>So, I boned up on UACROSS, and composed a command that worked. I came up with
>
>UACROSS '40 120' '45 75' DAT=RTGRIDS/GFS DAY=246 TIME=0 FHOUR=0 PAR=T CINT=5 E
> RASE=YES MAP=YES GRA=1

OK.  I just ran this invocation successfully.

>My script allows me to use both existing grids and to compute new
>grids.  They go into MYDATA/GRIDS.2001 and then I would input them into
>XSECTG. So, I did a GRDCOPY to create a test. The contents of 2001
>are:
>
>GRDLIST MYDATA/GRIDS.2001 NUM=ALL
>Dataset position 2001   Directory Title=
>
>PAR    LEVEL       DAY        TIME    SRC FHOUR     FDAY       FTIME GRID  PRO
>---- --------- ------------ -------- ---- ----- ------------ -------- ----- --
> --
>T    1000 MB   03 SEP 03246 00:00:00  GFS     0 03 SEP 03246 00:00:00 1 MERC
>T     850 MB   03 SEP 03246 00:00:00  GFS     0 03 SEP 03246 00:00:00 2 MERC
>T     700 MB   03 SEP 03246 00:00:00  GFS     0 03 SEP 03246 00:00:00 3 MERC
>T     500 MB   03 SEP 03246 00:00:00  GFS     0 03 SEP 03246 00:00:00 4 MERC
>T     400 MB   03 SEP 03246 00:00:00  GFS     0 03 SEP 03246 00:00:00 5 MERC
>T     300 MB   03 SEP 03246 00:00:00  GFS     0 03 SEP 03246 00:00:00 6 MERC
>T     250 MB   03 SEP 03246 00:00:00  GFS     0 03 SEP 03246 00:00:00 7 MERC
>T     200 MB   03 SEP 03246 00:00:00  GFS     0 03 SEP 03246 00:00:00 8 MERC
>T     150 MB   03 SEP 03246 00:00:00  GFS     0 03 SEP 03246 00:00:00 9 MERC
>Number of grids listed = 9
>GRDLIST - done
>
>I think these grids are a subset of the ones I used in the previous
>UACROSS.

I just created a local GRID file with the GFS T fields in MERC projection:

GRDCOPY RTGRIDS/GFS MYDATA/GRIDS.2001 DAY=246 TIME=0 FHOUR=0 PAR=T LEV=1000 850 
700 500 400 300 250 200 150 GPRO=MERC NUM=9

>Now, I changed the dataset from RTGRIDS/GFS to MYDATA/GRIDS.2001 and
>tried to create a vertical cross section:
>
>UACROSS '40 120' '45 75' DAT=MYDATA/GRIDS.2001 DAY=246 TIME=0 FHOUR=0 PAR=T CI
> NT=5 ERASE=YES MAP=YES GRA=2
>
>This fails and I don't know why.

I have confirmed that UACROSS fails for the local dataset that just contains
T fields.  I don't know why the reason for the failure at the moment.

>UACROSS generates a lot of output (without my asking); I tried changing
>the corridor width, but no success.
>
>  0-hour forecast: T    from GFS  at      0 UTC on 2003246  Proj = 1
>UACROSS: No grid found matching search conditions
>UACROSS: Could not get any TD   grids
>UACROSS: No grid found matching search conditions
>UACROSS: Could not get any SPD  grids
>UACROSS: No grid found matching search conditions
>UACROSS: Could not get any DIR  grids
>UACROSS: No grid found matching search conditions
>UACROSS: Could not get any Z    grids
>UACROSS: No grid found matching search conditions
>UACROSS: Could not get any RH   grids
>UACROSS: No grid found matching search conditions
>UACROSS: Could not get any U    grids
>UACROSS: No grid found matching search conditions
>UACROSS: Could not get any V    grids
>UACROSS: No grid found matching search conditions
>UACROSS: Could not get any W    grids
>UACROSS: No grid found matching search conditions
>UACROSS: Could not get any P    grids

The failure looks like a bug.  UACROSS shouldn't care about any of
these other grids since you are asking for T only.  This bears
investigation.

> Grid Point Locations Collected in SubGrid = 273
>UACROSS:
>UACROSS: Subgrid contains a lot of soundings-- too many to handle quickly.
>UACROSS:
>UACROSS: It would be more efficient to increase the grid sampling
>UACROSS: increment or decrease the corridor width/size unless
>UACROSS: you really need the high resolution.
>UACROSS:
>UACROSS: Only 25 soundings can be selected from the subgrid.
>UACROSS:
> Converting SubGrid Points to Soundings...
>
>   Acquiring records -- Number of levels so far = 200
>   Acquiring records -- Number of levels so far = 400
>   Acquiring records -- Number of levels so far = 600
>   Acquiring records -- Number of levels so far = 800
>   Acquiring records -- Number of levels so far = 1000
>   Acquiring records -- Number of levels so far = 1200
>   Acquiring records -- Number of levels so far = 1400
>   Acquiring records -- Number of levels so far = 1600
>   Acquiring records -- Number of levels so far = 1800
>   Acquiring records -- Number of levels so far = 2000
>   Acquiring records -- Number of levels so far = 2200
>   Acquiring records -- Number of levels so far = 2400
>Acquired sounding records -- Total number of complete levels = 2457
>Sorting the records by sounding number
>Aggregating records into soundings
>Sorting each individual sounding on pressures
>UACROSS: No grid found matching search conditions
>UACROSS: Unable to find a first station in dataset:
>UACROSS:      100 is the closest station at 9999.0 km distance
>UACROSS: Try increasing corridor size beyond  250.00 km to enlarge range for s
> earch
>UACROSS: or reduce at least one of the the grid sampling INCREMENT= keyword va
> lues:
>UACROSS:    Row increment = 5
>UACROSS:    Col increment = 5
>UACROSS: or change the latitude-longitude bounds appropriately.
>
>UACROSS -- done
>
>The e-mail archives have many references to UACROSS, but I don't see
>this one.
>
>Any advice, Bill

I am looking for a workaround at the moment.  I will let you know if/when
I find one.

Tom