Theissen polygons

[Dave Gerdes]

> From: mccauley at ecn.purdue.edu (Darrell McCauley)
> Reply-To: grassp-list at amber.cecer.army.mil
> To: grassp-list at amber.cecer.army.mil
> In-Reply-To: Darrell McCauley's message of Wed, 1 Apr 92 17:51:02 -0500 <9204012251.AA14387 at sandbar.ecn.purdue.edu>
> Subject: Theissen polygons
> 
> 
> Yesterday I wrote:
> >Is anyone working on something to create Theissen polygons from site files?
> >If not, do you know where I could get src code to get me started?
> 
> I've got some code from netlib and have started work. I assume that there's
> not a problem with incorporating this in GRASS (there's no copyright or
> anything and I will give credit to the original author). Right?
> 
> There seems to be something missing from $GISBASE/man/man5---a
> description of vector files. Is anyone working on this?
> 
A description of ASCII vector files does exist, but managed to miss being
distributed with 4.0.  My understanding is that you can request it from
the Office of GRASS Integration.  I have also placed them for anonymous
ftp from moon.cecer.army.mil in file misc/asc_vect_fmt.tar


> Once I find all bisectors, I'm faced with a question of what to do at
> the edges. Suggestions? I plan to put a rectangle along the borders of
> the DEFAULT region and extend the lines to there.  Good? Bad?

That sounds like a good plan to me.  Just have to make sure that you don't
have any points or polygons that extend outside of the default region.  
Also consider using the CURRENT region instead of the DEFAULT.

> 
> Unless I become more intimately familiar with the library routines, I
> plan write an ascii vector file. Good? Bad? 

I would recommend writing binary vector files, just because it makes one
less step for the user running it.  The upcomming GRASS Programmer's Manual
will give info for writing vector files.  Until then, I recommend looking
at the source code for v.in.ascii (src/mapdev/v.in.ascii) for an example of 
how to create binary vector files.  Or drop me a note if you want me to 
mail you troff files for it.

> 
> Since I'm not real familiar with file formats, I would like to know
> if the following format is okey-doke:
> 
>   A 2
>    2.30000 4.0000
>    5.0000  3.0000

That looks good.  Note that the columns are y,x order.  Don't forget the
header of course.

> After dumping a vector file that I had lying around, I noticed some
> entries with an "A" and some with an "a" (DEAD_AREA?). I guess this
> will become more apparent once I get it working and try to use the
> vector file.

Dead features are those which have been deleted.  They are no longer real.
The codes are changed rather than incurring the overhead of deleting data
from the middle of a file.

> 
> Should I include the original sites as points ("P") or DEAD_DOTs ("p")
> in the vector file? There's already a way to use site_lists to label
> vector files (s.to.dig on moon:incoming).
> 

Points ("P") would be good.  I would probably make that a runtime option whether
they are included or not.

> Darrell
> 
> 


-- 



  Dave Gerdes
  US Army Construction Engineering Research Lab
  Spatial Analysis & Systems Team
  dpgerdes at cerl.cecer.army.mil
  (217) 352-6511 x591