[Xastir] Bounding boxes

[Tyler Allison]

> I wouldn't.  I would pick the most westerly of the two left corners as
> the west boundary, the most easterly of the two right corners as the
> east, etc.  That'll guarantee that a rectangle made up of N-S and E-W
> lines will entirely enclose the data.  If you pick the average and the
> map isn't actually aligned with N/S,E/W as this one appears not to be,
> then if the FGD file is used to clip the boundary you'll lose bits of
> the map.  But picking them this way will include bits of the collar, too.

Ahh..I understand..I think.

In the case of the Ohio and Kentucky geotiff files they have no colar (if
I understand the term colar).  The ENTIRE image is the topographic data.
So in theory there is no need to 'remove' any of the pixals.

>
> But I'm concerned about the actual values in Tyler's examples --- USGS
> maps are usually defined by very simple lat/lons --- for example, the
> 7.5-minute quads are aligned so that they have west boundaries of,
> say, 106d-105d52m30s for one map, 105d52m30s-105d45m for the next one
> to the east, 105d45m-105d37m30s for the next, etc.  That the
> re-projected corner points don't map to neat values concerns me ---
> perhaps they were projected back using the wrong datum?  The
> difference seems too great for that, though.

What can I say? ;)  Im using the geotiff fille: f48120a1.tif that was
obtained from one of the URL's in the README.MAPS file. It works on my
system yet the listgeo data that was posted here by me is exactly as you
read.


>
> There's a naming trick that USGS used, and that's exploited by
> mapfgd.pl in the scripts directory --- if you got them from USGS and
> haven't changed the name, maybe you could use that?

Unfortunately the geotiff files from Ohio/kentucky are labeled, for
example, b40.tif.  Not exactly useful for conversion ;)

-Tyler