[GRASS5] Re: r.in.gdal - precision problem in lib/gis/adj_cellhd.c

[Markus Neteler]

On Thu, Apr 07, 2005 at 04:25:38PM +1200, Hamish wrote:
> > Hacking lib/gis/adj_cellhd.c to permit the import, 
> > I get the following cell header:
> > 
> > grassdata/latlong/PERMANENT/cellhd/srtm90_gtopo30
> > proj:       3
> > zone:       0
> > north:      90:00:00.000001N
> > south:      39:59:59.990613N
> > east:       20:00:00.007509E
> > west:       20:00:00.000001W
> > cols:       4800
> > rows:       6000
> > e-w resol:  0:00:30.000002
> > n-s resol:  0:00:30.000002
> > format:     -1
> > compressed: 2
> > 
> > Probably it's a GRASS/GDAL issue with LatLong data?
> > 
> > Fixing the cell header files manually solved the
> > problem but isn't very elegant.
> > 
> > Any hints are welcome
> 
> 
> It's not trying to add half the cell to the top row of the TIFF, thus
> going beyond 90d 0' 15"?

Probably. But then we would exclude a lot of global data!

> I thought SRTM only made it to 60 deg lat anyway? Space Shuttle is not
> polar orbiting.

Note that these data are the GTOPO30 data enhanced by SRTM, not
the original SRTM data. So north of 60 deg GTOPO30 remained untouched AFAIK
(northern hemisphere).

> 
> I have noticed in the past that v.out.ogr to a shapefile does similar
> strange off-by 1e-9 type errors to double precision columns in the
> data.

Mhh, a systematic problem?

Markus