[gdal-dev] Re: Fwd: Shift of geoTiff

Mallee ben at malleefocus.com.au
Tue Jun 8 23:36:58 EDT 2010


(Ben here - thought I'd have a look at this list)...

Jukka that looks right if the origin is the upper left corner of the upper
left pixel.

According to the GeoTiff spec
(http://www.remotesensing.org/geotiff/spec/geotiff2.5.html#2.5.2), that
would be the 'PixelIsArea' raster grid space - which is the default (see
2.5.2.2).

There is a PixelIsPoint raster grid space (also defined in 2.5.2.2), which
has the origin at the centre of the upper left pixel. In that case (cols -
1) x dx and (rows-1) x dy are right for getting the corner coordinates,
which are the centres and not the extremities of pixels. I have a feeling
the GeoTIFF I sent Paul is one of these types, because the MetaData supplied
with the file implies it. In the MetaData the corner coordinates given are
(cols-1) x dx and (rows-1) x dy away from eachother.

GDALINFO is reporting (AREA_OR_POINT) that the GeoTIFF is a 'Point' type,
which means that it is GDAL that is not rendering the coordinates correctly,
and not a problem with the GeoTIFF.

The GeoTIFFs, by the way, are produced by USGS from Landsat 4, 5 and 7 data
and distributed worldwide, freely, through http://glovis.usgs.gov/. There
will be plenty of people who are half a pixel out! SAGA (and presumably
other GDAL-based GIS) has the same problem with these GeoTIFFs.

Thank you for looking into it

Ben

-- 
View this message in context: http://osgeo-org.1803224.n2.nabble.com/gdal-dev-Shift-of-geoTiff-tp5149184p5156602.html
Sent from the GDAL - Dev mailing list archive at Nabble.com.


More information about the gdal-dev mailing list