[GRASS-user] v.net vector orientation when creating a stream network with gauges

[Moritz Lennert]

[Please keep conversations on the list]

On 12/04/16 10:02, Bartolomei.Chris wrote:
> Hi Moritz, Thank you for the feedback but I had already included
> Micha's snapping process in my attempts and while it gets very very
> close, there is still a .00000something meter vector being added (as
> evidenced in the creation of the new layer to re-categorize the
> broken stream vector (due to v.net) and added the length as the
> forward cost. I even tried setting the forward and backward costs to
> positive values for the tiny added vectors but to no avail.

These vectors shouldn't exist. This seems to be a bug in the v.distance 
output coordinates or in the creation of the new points, or at least an 
issue with coordinate precision somewhere in the process.

If you can create a reproducible example, ideally using the North 
Caroline Demo dataset then please post this in a bug report.

>  Part of
> the problem is that v.net.allpairs requires nodes and arcs to be on
> one layer, if you use layer 1 for arcs and layer 2 for nodes then
> create layer 3 for the recategorized arcs you can't tell
> v.net.distance to use layer 3 for arcs ... the only thing that almost
> works is v.neth.path ... but only if there are no stream nodes
> (junctions) between the points.

All these layer issues are resolved in grass7, so once again, I can only 
urge you to convince your client to move to grass7...


> I had passed this information along
> to Markus and also my thanks to Micha, but alas, it does not work if
> there is a stream junction between the added points. Believe me, I
> have tried many many different methods. I attached my data if you
> would like to look at it

No data attached...

>- create a location called "latlon" and a

Have you tried doing all this in a projected location ? Maybe there is 
some issue with working in an unprojected latlong location...

Moritz