[gdal-dev] Meteosat 2nd Gen projection (MSG GEO)

[Frank Warmerdam]

Jose Luis Gomez Dans wrote:
> You are absolutely right (as usual!). The file contains two bands, one with
> Latitude and one with Longitude values for each pixel's location. In a
> sense, it is a set of GCPs stuck in a raster. Since my data are referenced
> with respect to these pixels, I can then work out the Long/Lat values from
> here. i guess that thinking about it, what I would like to do is to shove my
> data into a raster like this (unreference) ENVI raster, and be able to
> project it into something else. So I could still extract the data from my
> reference ENVI raster into GCPs and use that work out the GeoTransform, and
> assign a projection to my rasters.
> 
> Since I have a GCP for each pixel, what would be the best way of dealing
> with this? Extracting gcp lines is not a problem, how can I tell
> gdal_translate to use a 2000x2000 (ish) list of GCPs?

Jose,

There are a few approaches.

One is to extract a subset of GCPs (a 10x10 grib of GCPs would likely be
sufficient).

Another is to create a VRT file that lets GDAL know that the lat and long
bands are geolocation information.  This is fairly involved, and GDAL 1.5
is the first version where gdalwarp properly supports using such
geolocation arrays.  But the approach is documented in:

   http://trac.osgeo.org/gdal/wiki/rfc4_geolocate

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