[GRASSLIST:3369] Re: landsat5_bsq

[Glynn Clements]

Maciek Sieczka wrote:

> From: "Glynn Clements" <glynn.clements at virgin.net>
> 
> > > UPPER_LEFT_CORNER=0141500.0042E,0511547.6328N,447671.547,5679365.027;
> > > UPPER_RIGHT_CORNER=0165706.5136E,0505151.0288N,637351.781,5636535.711;
> > > LOWER_RIGHT_CORNER=0162013.7030E,0491632.8889N,597259.322,5458976.331;
> > > LOWER_LEFT_CORNER=0134309.6950E,0493942.1275N,407579.088,5501805.647;
> >
> > This data isn't aligned to any known coordinate system, so you can't
> > just import it directly. You would need to use e.g. i.rectify to
> > transform it to the desired coordinate system.
> 
> I'm not sure what you mean. They are somewhat "projected".
> I mean I know the corners coordinates.

What I mean is that region defined above doesn't form a rectangle in
any likely coordinate system, i.e. the top/bottom edges aren't
horizontal and the left/right edges aren't vertical. At a minimum, the
data is rotated, and may include skew or even a non-affine
transformation.

In GRASS, the X coordinate (easting) of a raster cell is determined
entirely by its column, while the Y coordinate (northing) is
determined entirely by its row.

> UPPER_LEFT_CORNER is 14d15'00.0042"E 51d15'47.6328"N which is 447671.547E
> 5679365.027N meters in the UTM, zone 33, WGS84
> 
> And so forth.
> 
> As you said I rectified the raster myself (using coords as above).
> 
> I get results close to reality but there is a shift of about 1 km beetwen
> the rectified raster and an other TM of the same area - a geotiff which I
> believe to be pretty well projected (verified with a topo map).
> Also the resolution of the rectified raster is different than the "28.5
> meters" stated in the *.HD file.
> 
> Do you have a clue what may be the reason?
> 
> What "order of transformation" should I use for the i.rectify?

I don't have an answer for either of these.

> How do you think: which points in the raster the coords listed above refer
> to? - the black border corners or the "beginnings of the picture" itself?
> (the thumbnail attached)

I don't have an answer for this either; however:

1. GRASS needs the coordinates of the coorners of the raster.

2. You need to know whether the coordinates refer to the corners of
the raster or to those of the "picture" in order to position the data
correctly.

-- 
Glynn Clements <glynn.clements at virgin.net>