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

20000720: GINI to netCDF (cont.)

>From: David Harper <address@hidden>
>Organization: NCAR/RAP
>Keywords: 200003012325.QAA04248 McIDAS-X GINI netCDF

Dave,

re: Lat,Lon comparisons at UL and LR corners are within 2 parts per 1000

>Actually I made that same comparison.  I think the problem is in the image,
>not the coordinates.  When we look at it in Zebra, White Sands and the river
>are displaced relative to the coordinates.

I looked up the location of White Sands in an atlas and get a location
of 32:50N, 106:20W.  Given that the white sands themselves cover quite
a bit of ground, there is no telling if the numbers I get back from the
center of the white sand feature represent the location listed in the
atlas.  My listings show that the east-west size of the sands is only
about 0.39 degrees across (at 32.9N). A shift of 0.5 degrees would,
therefore, be quite noticable.  Also, when I load the original GINI
image or its AREA copy in McIDAS and put a high resolution map on it, I
get very good agreement with the location of the Rio Grande and the
Texas/Mexico border.

To see how the image looks in McIDAS (with brightnesses exaggerated so
that the river and the sands can be seen easily), please check out:

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

Can you provide me with browser loadable rendition of what you are seeing
in Zebra?

>Have you compared the actual data values at those points?  That might be
>more telling.

The comparison for Lat,Lon and data values is:

point        netCDF brightness     McIDAS AREA brightness
------------+---------------------+-----------------------------
1,1          75                    74
800,800      83                    82

point        netCDF Lat,Lon [deg]  McIDAS AREA Lat,Lon [deg]
------------+---------------------+-----------------------------
1,1          36.32418N  110.7891W  36.32222N  110.79055W
800,800      29.71597N  101.4901W  29.71333N  101.49055W

The discrepency in the brightness values could be the 1 pixel offset
that I mentioned in my first note.

Tom