[Gdal-dev] Re: Erdas 7.4 *.gis - wrong values

[Maciek Sieczka]

On śro, 2005-12-14 at 15:07 -0500, Frank Warmerdam wrote:
> On 12/14/05, Maciek Sieczka <werchowyna at epf.pl> wrote:
> > Frank,
> >
> > I got some ERDAS 7.4 *.gis files which where created by LANDIS
> > http://landis.forest.wisc.edu/ and I was asked to import them into
> > Grass.
> >
> > They are supposed to be 16 or 8 bit integer. The value range should be
> > within 0-300. However gdalinfo reports wrong Computed
> > Min/Max=-32256.000,30720.000.

<snip>

> Maciek,
> 
> I have confirmed that GDAL was not making the right decision about
> byte swapping for 16bit LAN/GIS files.  I have committed a fix for
> this, and your file now seems to work OK.
> 
> I appologise for not responding to this sooner.  I was at a client
> site last week, and I lost track of this issue.

No problem! Thanks for fixing this. I'll test soon with GDAL from CVS.
Thank you for you help.

> > P.S.
> > There is also an issue with the georeference handling. The planar
> > coordinates reported by gdalinfo are correct, but a latlong WGS84 CS is
> > forced and latlong cooridnates are printed too. I can override it when
> > importing into Grass, no problem, but looks strange, might be worthy of
> > taking look at too.
> 
> The file has vlaue "0" for the coordinate system field, which is
> apparently Lat/long.  I don't know if my long ago reverse engineering
> of this field is wrong, or if there is some auxilary override I should
> be looking for.  But I don't really have any way to figure out what
> the coordinate system of the file is.

I see. It is weird that for coordinates expressed in millions of units
latlong is assumed. Maybe LANDISII plays lose with CS handling and
assigns "0" while it should be some other value indicating an "unknown
x,y CS".

Cheers,
Maciek


--------------------
W polskim Internecie są setki milionów stron. My przekazujemy Tobie tylko najlepsze z nich!
http://katalog.epf.pl/