i.rectify2

[Malcolm Williamson]

(Stuff deleted)
> >
> >I have both vector data and rasterdata in my lat-lon mapset. I made the
> >coordinate transformation from ll-->lcc for the vector data with
> >v.proj.  For the raster data I used i.rectify2 as described.  Now, if
> >you make an overlay in the lcc mapset. (the vector map over the raster
> >data) they don't fit. They are shifted up to tens of kilometers. in
> >some of the Greek islands, the 'vector island' and the `raster island`
> >don't even overlap!  In order to check my results I have transformed a
> >vector map (with v.proj) and a raster map which was derived from this
> >vector map (In the lat-lon mapset, using v.to.rast) I would expect that
> >both maps would fit exact in the lcc mapset, as they did in the lat-lon
> >mapset. But no....  So either v.proj is not accurate (but since this is
> >a coordinate transformation for which the transformation formulas are
> >known, I have assumed that my vector data are correct) Or, i.rectify2
> >is not accurate enough. (which I can understand, because the area is
> >large and i.rectify only makes a fit, for the transformation matrix).
> >Or, I am missing something and there is an other solution for this
> >problem.
> 
> I suspect that your second  supposition is the problem---i.rectify2 .
> To check v.proj you are better to compare it to proj of PROJ.4 or
> GCTP but not against i.rectify.
> 
> The clue also seems buried in the "transformation matrix" but since
> I have no experience with i.rectify I can't be of much help other
> than to note that it must be *much* more than a simple scale-rotate-
> shift operation.  That is, i.rectify2 must know all about lcc
> to be able to create a complex non-linear transformation.
> 
> Any i.rectify2 gurus out there?
> 
> Gerald (Jerry) I. Evenden   Internet: gie at charon.er.usgs.gov
> voice: (508)563-6766          Postal: P.O. Box 1027
>   fax: (508)457-2310                  N.Falmouth, MA 02556-1027
> 

I don't claim to be a guru, but I haven't seen the *big question* answered yet:
how many points did you use to rectify? Also, how close was the nearest point
to the area where you are finding your error. This kind of rectification 
process is basically a brute force kind of thing - if you really care about 
where a particular point ends up, you had better tell it where it should be!!
(ie assign points close to it!).

In general, to accurately rectify an entire region, the more points that you
have, spread out fairly evenly over the region, the better. If anyone has had
other experiences, please share them!

-- 
Malcolm D. Williamson - Research Assistant       E-mail: malcolm at cast.uark.edu
Center for Advanced Spatial Technologies      Telephone: (501) 575-6159
Ozark Rm. 12                                        Fax: (501) 575-3846 
University of Arkansas              
Fayetteville, AR 72701