[GRASS-user] Strahler order starting from vector data?

[Jasiewicz Jarosław]

W dniu 25.01.2011 13:03, Hamish pisze:
> Hi,
>
> I see v.strahler and r.stream.order in addons can assign
> Strahler Order to a river network, but how to run that if you
> are starting with an existing vector river network with no
> raster DEM?
Ordering existing river network (without DEM) require topologically 
correct network I can guarantee that r.stream.order can order a network 
created by automatic software like r.stream.extract or r.watershed but 
not about network digitized manually. There are lots of topological and 
directional checks, but such tool is possible in the feature if really 
needed
> could you cross with a matching coastline vector to
> extract end nodes at the river mouths?
With map algebra for example I can cut river network in any place and 
order it
> will even a rough DEM
> like SRTM data be better than nothing in that case?
yes, but it not guarantee that river network will be exactly reproduced
>   how good
> does it have to be? would etopo1 be ok? (1 arc-minute ~ 2 km
> cells)
Depending the resolution end extend you need.
> also, how to get around attaching lake and swamp areas which
> split the river vector line in two?
This is unresolved problem in a general case of network ordering. In 
general lake is treated as a flat area where stream pass trough. (see 
r.watershed of GRASS 7 with beautify flat areas for details). There are 
some early proposition of Mark from early 80s (I do not remember the 
title but I can search if you interested) how to treat such structures 
but there are no common agreement about it. On the other hand lake shall 
be treated as local depression. This is implemented in r.watershed and 
r.stream.extract for GRASS7
> convert their boundaries to
> lines and treat it as a braided river? fill them in then r.thin
Same as above. No common ordering system for braided rivers. This 
problem is commonly called in discussion by hydrologists, but as long as 
there were no uniform ordering system for braided we can do nothing
> them somehow? otherwise topology would be clean.
> how about if the river goes into a cave and out somewhere else?
Another unresolved problem. r.stream was created for terrestrial areas 
not karst system. Modeling of karst system require completely different 
approach
> need to trace that gap by hand? (may be too big for the sloppy=
> option)
Probably yes

Jarek