[Gdal-dev] RFC 4: Geolocation Arrays

[Frank Warmerdam]

Ivan Lucena wrote:
> Hi there,
> 
> If I understand correctly RFC 4, you can use the Geolocation bands to 
> warp the image bands when they exists in the same input file (in a 
> HDF-EOS Swath for example).
> 
> This is possible only because the AVHRR, Envisat, HDF and netCDF drivers 
> have add to the Metadata the fields: “GEOLOCATION”, "PIXEL_OFFSET", 
> "LINE_OFFSET", "PIXEL_STEP", "LINE_STEP", "X_BAND", "Y_BAND".

Ivan,

I have some good news.  Per RFC 4 you can also have X_DATASET and
Y_DATASET metadata pointing to another file for the X and Y bands.

   http://www.gdal.org/rfc4_geolocate.html

So you don't really need to reassemble things into an HDF file or
something similar.

> But I have someone here with lots of similar kind of information but in 
> separated Envi file and he want to do the same process... :-)
> 
> For each par of files there is one with a bunch of hyperspectral bands 
> and the other with just two bands with Latitude and Longitude.
> 
> The question is: If one invests time developing a python script to 
> gather those bands in a HDF4 and set the appropriate metadata is 
> gdalwarp going to work? I mean, the metadata will stick to the file and 
> will be interpreted by gdalwarp accordantly?

Unfortunately, I also have some bad news.  The gdalwarp support for
geolocation arrays does not really work.  It is basically partially
implemented but I gave up on it because I could see how to make it work
without quite a bit more work.  So, you cannot use geolocation arrays
with gdalwarp currently.

If I could get this implemented I would finalize RFC 4 and try to get it
passed, but currently I'm stymied.

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | President OSGeo, http://osgeo.org