[GRASS-user] wrong result in v.net.iso for backward direction

[Mira Kattwinkel]

Dear list members,

has nobody any idea what's going on (see below)?

When using forward and backward costs or only backward costs in 
v.net.iso the backward end (i.e. upstream) always gets too short / too 
few segments.

Can anybody please point me into the right direction.

All the best, Mira


> Subject: 	wrong result in v.net.iso for backward direction
> Date: 	Thu, 8 Jun 2017 10:46:42 +0200
> From: 	Mira Kattwinkel <kattwinkel-mira at uni-landau.de>
> To: 	grass-user <grass-user at lists.osgeo.org>
>
> Dear list members
>
> I am using v.net.iso to split a stream network at a certain distance
> from sampling points.
>
> First, I create a network from vector lines (streams) and vector points
> (sampling sites) using v.net. The lines feature have a 'cat' column,
> 'length', 'backward_cost' and 'forward_cost'. I would use -1 for forward
> costs because I am only interested in the upstream part, and length for
> backward costs in v.net.iso:
>
> v.net.iso input=test_edges arc_layer=2 node_layer=3
> output=test_edges_bw_2000  center_cats=55  arc_column=forw_cost
> arc_backward_column=backw_cost costs=2000
>
> However, the backward part of the resulting lines with cat 1 is always
> too short. Likewise, if I give just 1 for the backward costs and set the
> costs to 5, it gets 4 segments with cat 1. Working in both directions at
> the same time gives correct values for the forward end, but too short
> for the backward end. I then realised that the numbers would be correct
> if the first part of the forward end was added to the backward part (see
> attached example). The forward part all the costs (lengths) sum up
> correctly to 2000 (1443.19 + 556.81). For the backward part it would be
> 45.72 + 511.09 = 556.81. However, if the first segment of the forward
> part is added, it gives the correct cost sum (45.72 + 511.09 + 1443.19 =
> 2000).
>
> Do I use the function in the wrong way or is this a bug?
>
> Thanks a lot,
> Mira
URL: 
<http://lists.osgeo.org/pipermail/grass-user/attachments/20170608/97469a58/attachment.png>

URL:<http://lists.osgeo.org/pipermail/grass-user/attachments/20170608/97469a58/attachment-0001.png>

PS

I just realized that it works correctly for length in both direction if
the parameters arc_column and arc_backward_column are not given.
However, for me this is inefficient because I only need the backward end.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/grass-user/attachments/20170609/7a1a7be0/attachment.html>