s.menu

[Craig Anderson]

> >Does anyone know if there is an upper limit to the amount of sites, s.menu
> >can convert to a rastfile? If I have a very large site file (139203 sites).
> >s.menu seems to read the site file, but skips the SITES to CELL part, returning>to the main menu.....
> >Smaller site files work okay.
> >The alternative, s.to.rast does not work in a lat-lon mapset.
> >
> >
> >Harm Olthof (holthof at ecnsun.ecn.nl)
> 
> Harm;
> 	s.to.rast worked fine for me with a lat.long mapset just be sure
> that when the site data is imported it is long|lat|#site format.  Complete with
> the vertical bars and # sign if you want the sites to be labeled so.  See 
> section 7 of the programmers manual and the s.to.rast man page.
> 
> Craig
> 
> 

>That is strange, I used the correct site format, but all s.to.rast gives me 
>is a raster file with something like 5 isolated cells in it. 
>Meanwhile I solved the original problem by chopping the file im three chunks
>of ca. 50000 sites. It would still be nice to know what the upper limit would be. 
>Thanks,

>Harm Olthof (holthof at ecnsun.ecn.nl) 

Ahhhh you must set your cell resolution finer, if two cells fall in the same
grid cell only the last value will be assigned to that cell and the previous
value will be lost.  For my meterological sites, I finally had to go to 1
minute resolutin to get the s.to.rast to work well.  

What I do is 

- s.menu option 4 to show duplicate sites.
- s.menu option 7 to generate a frequency raster
- r.out.ascii to a file that is accessed by fortran to count up the greater
  than one occurances.  Multiple values in a cell.
- correct any problems 
- s.to.rast at a fine enough resolution
- r.out.ascii to show the total number of sites a a cross check.  This file
  can get large so be careful if you have many sites.  I use r.out.ascii -1lgz.

I realize it is cumbersome and that a program like proprietary-info would do 
it faster with less trouble but then again its not public domain.  Anyway once
your sites are rasterized, they should be done once and for all.

Craig