[GRASS-dev] Re: [GRASS-user] inaccuracy in v.to.rast results

[Maciej Sieczka]

Dylan Beaudette wrote:
> On Tuesday 19 December 2006 08:24, Yu Zhang wrote:

>> I wonder if anyone can provide a reference to the
>> algorithm used in v.to.rast. I found mismatches
>> between  GRASS raster with Arcview one that were
>> derived from the same polygon layer. v.to.rast in
>> GRASS tends to underestimate the number of cells
>> within certain polygons

> Do you have some images which demonstrate the comparison ?

Hi,

Sorry for jumping in, but it might be related.

I don't know for polygons, but I have similar experience with lines -
not enough cells are rasterised compared to what you would expect -
line's start and end happen to be skipped. Not mentioning that
v.to.rast tries to kindoff approximate the line's shape than to detect
which cells the line actually goes through.

See bug report:

http://intevation.de/rt/webrt?serial_num=4488

The examples mentioned:
http://www.nabble.com/attachment/4566255/0/wrong_cell2.png
http://www.nabble.com/attachment/4566255/2/wrong_cell4.png
http://www.nabble.com/attachment/4566255/1/wrong_cell3.png

http://www.biol.uni.wroc.pl/sieczka/udostepnione/grass/vtorast/wrong_cell.png

r.thin doesn't help:
http://www.biol.uni.wroc.pl/sieczka/udostepnione/grass/vtorast/rthindoesnthelp.png



BTW, two other long standing v.to.rast bugs:

http://intevation.de/rt/webrt?serial_num=5163
(solution by Glynn is there, can someone apply it please?)

http://intevation.de/rt/webrt?serial_num=4480

Maciek