[GRASSLIST:5366] Re: r.in.gdal and AVHRR

[Markus Neteler]

On Thu, Jan 23, 2003 at 05:35:17PM -0300, gmpuljak at dinara.gub.uy wrote:
> Hello,
> 
> 	I've downloaded an AVHRR image (level 1B) from NOAA (www.saa.noaa.gov) and I want to import it into GRASS.
> 	According to the list of image formats that r.in.gdal handles it is possible. However, when I try the program doesn't recognize it as a valid image format. How can I tell r.in.gdal which is the image format?
> 
> 	Thanks

You have to be sure to use the latest version of the GDAL library, i.e
GDAL 1.1.8:

NEWS:

GDAL 1.1.8 - Overview of Changes
--------------------------------

 o Implemented HDF 4 read/write support. This includes HDF EOS reading.

 o Implemented Windows BMP read/write support. 

 o Implemented NITF read/write support. 

 o Implemented NOAA Polar Orbiter L1B format driver. 

 o Implemented EOSAT FAST format driver. 

[...]

But note that you will have to find a way to handle the distortions when
geocoding the AVHRR scene.

When importing AVHRR with r.in.gdal, it will generate a POINTS file suitable
for i.points. The problem is that the polygonial approach of i.points will
not rectify the image properly (I did several tests, the resulting map was
shifted up to 25km!).  Unless splines (or a different method) is implemented
into i.points, you will not be able to geocode AVHRR with a reasonable
accuracy using i.points.

We need a volunteer to improve i.points. As splines interpolation is already
implemented in GRASS, things need to be connected.

[BTW: the 'gdalwarp' software also uses polynoms, maybe in future it will
      supports also splines. Then you could rectify AVHRR with 'gdalwarp'
      before importing into GRASS. Or someone improves i.points...]

Sorry for the bad news,

 Markus Neteler