[Gdal-dev] gdal_translate & srcwin

Charlton Purvis cpurvis at asg.sc.edu
Mon Aug 16 14:14:23 EDT 2004

Hi, folks:

(I've mentioned attachments in this email, but I've put them in this dir instead of on this post.  They are here:  http://nautilus.baruch.sc.edu/~cpurvis/gdal.)

Just starting to dabble w/ GDAL, and I've probably got an obscenely naïve question for the group.

My file's info . . .

gdalinfo HDF4_EOS:EOS_GRID:"dem30ARC_E60N0.hdf":demGRID:LandWater
Driver: HDF4Image/HDF4 Dataset
Size is 14400, 10800
Coordinate System is:
GEOGCS["Unknown datum based upon the Clarke 1866 ellipsoid",
    DATUM["Not specified (based on Clarke 1866 spheroid)",
        SPHEROID["Clarke 1866",6378206.4,294.9786982138982,
Origin = (60000000.000000,0.000000)
Pixel Size = (8333.33333333,-8333.33333333)
Corner Coordinates:
Upper Left  (60000000.000,       0.000) (60000000d 0'214748364800.00"E,  0d 0'0.01"N)
Lower Left  (60000000.000,-90000000.000) (60000000d 0'214748364800.00"E,90000000d 0'322122547200.00"S)
Upper Right (180000000.000,       0.000) (180000000d 0'648540061696.00"E,  0d 0'0.01"N)
Lower Right (180000000.000,-90000000.000) (180000000d 0'648540061696.00"E,90000000d 0'322122547200.00"S)
Center      (120000000.000,-45000000.000) (120000000d 0'433791696896.00"E,45000000d 0'163208757248.00"S)
Band 1 Block=14400x1 Type=Byte, ColorInterp=Gray

If I do a straight gdal_translate of the above HDF in its entirety with a

gdal_translate -of GTiff \
  HDF4_EOS:EOS_GRID:"dem30ARC_E60N0.hdf":demGRID:LandWater \

I'll get something like the attached full_shot.png.  Not very glamorous, but so far so good.

Now what I want to do is to take a small sample of the data and produce an image.  So I was thinking that something like

gdal_translate -of GTiff \
  HDF4_EOS:EOS_GRID:"dem30ARC_E60N0.hdf":demGRI D:LandWater \
  -srcwin 7200 0 1000 1000 purvis2.tif

would give me a 1000x1000 .tif beginning at pixel 7200,0 (didn't see any units for srcwin parameters in the docs).

Well, I do get what appears to be the correct region, but why is my data 'gone', i.e. only a region w/ one value (highlighted in only red, because they all come back value = 0)?  See subsample_shot.png.



Charlton Purvis
(803) 777-8858 : voice
(803) 777-3935 : fax
cpurvis at sc.edu
Baruch Institute
University of South Carolina
Columbia, SC 29208

More information about the Gdal-dev mailing list