[GRASSLIST:5703] Re: r.in.gdal problem?

[Markus Neteler]

On Fri, Feb 28, 2003 at 08:39:35PM +0000, Glynn Clements wrote:
> 
> David Orme wrote:
> 
> > I'm using the r.in.gdal module to import ARCINFO GRID format raster  
> > files and I seem to be getting a shift in the extent of the data on  
> > import:
> > 
> > The files contain 1 by 1 degree resolution raster data. In Arc, the NS  
> > extent of the raster data is:
> > 
> > bottom edge of southernmost grid square = 90S
> > topmost edge of northernmost square = 77N
> > 
> > The raster therefore contains 167 rows of data.
> > 
> > When I import this, the file extent is as follows:
> 
> > >  |        Projection: UTM (zone 13)                                             |
> > >  |            N:       77.5    S:      -89.5   Res:     1                       |
> > >  |            E:      330.5    W:     -226.5   Res:     1                       |
> 
> > The number of rows (and columns) are correct but the extent is peculiar  
> > - I'm guessing that r.info returns the centre points of the grid cells  
> > but if this is the case then the centre of the northernmost cell should  
> > be at 76.5N not 77.5N.  This also makes the reported southern extent  
> > odd since the 89.5 S implies that there is still a gridsquare that has  
> > 90S as a bottom edge. Using d.where to inspect the location visually  
> > shows that the top of the northernmost cell in grass is now at 78N and  
> > the bottom of the lowermost cell is at 89S.
> > 
> > Something doesn't add up - I'm expecting to find that it may be me but  
> > can anyone clear this up?
> 
> The GRASS region refers to cell corners, not centres; it seems that
> r.in.gdal is misinterpreting the region information in the grid file.
> 
> You can fix the resulting map manually using r.region, e.g.
> 
> 	r.region map=... n=77 s=-90 ...

Is this a GDAL bug or a GRASS bug? Does this mean that r.in.gdal
is always shifting the data (were a nightmare!)?

Markus